‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ) ‪( [email protected]‬‬
‫ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ)ﭘﻠﻲﺗﻜﻨﻴﻚ ﺗﻬﺮﺍﻥ(‬
‫ﺩﺍﻧﺸﻜﺪﻩ ﺭﻳﺎﺿﻲ ﻭ ﻋﻠﻮﻡ ﻛﺎﻣﭙﻴﻮﺗﺮ‬

‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪۰‬‬
‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ‬
‫ﻓﺼﻞ ﺻﻔﺮﻡ‬

‫‪۱‬‬ ‫‪ ۱‬ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ‬

‫‪۲‬‬ ‫‪ ۱.۱‬ﻧﻤﺎﻳﺶﻫﺎﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۲‬‬ ‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۱‬‬
‫‪۳‬‬ ‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۱‬‬
‫‪۶‬‬ ‫ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ‪. . . . . . . . . . . . . . .‬‬ ‫‪۳.۱.۱‬‬
‫‪۷‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱‬‬
‫‪۸‬‬ ‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۱‬‬
‫‪۸‬‬ ‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪. . . . . . . . . . . .‬‬ ‫‪۲.۲.۱‬‬
‫‪۱۰‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ‪. . . . . . . .‬‬ ‫‪۳.۲.۱‬‬
‫‪۱۴‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮ ﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ‪. . . . . . . .‬‬ ‫‪۴.۲.۱‬‬
‫‪۱۶‬‬ ‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۱‬‬
‫‪۱۸‬‬ ‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ‪. . . . . . . . . . .‬‬ ‫‪۱.۳.۱‬‬
‫‪۲۱‬‬ ‫ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻣﺎﺷﻴﻨﻲ ﺍﻋﺪﺍﺩ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۱‬‬
‫‪۲۲‬‬ ‫∗ ﺟﺰﺋﻴﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻭ ﻣﻀﺎﻋﻒ ‪. . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۱‬‬
‫‪۲۴‬‬ ‫ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۱‬‬
‫‪۲۵‬‬ ‫ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۱‬‬
‫‪۲۶‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﺭﻳﺎﺿﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۵.۱‬‬
‫‪۳۰‬‬ ‫ﺗﻔﺎﻭﺕﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ ‪. . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۵.۱‬‬

‫‪۳۵‬‬ ‫‪ ۲‬ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﻴﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬

‫‪ ۱.۲‬ﻣﺒﺎﻧﻲ ﺳﻨﺠﺶ ﺧﻄﺎ ‪۳۶ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬

‫ﺁ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﺏ‬
‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬

‫‪۳۶‬‬ ‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۲‬‬

‫‪۳۹‬‬ ‫∗ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ ‪. . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۲‬‬
‫‪۴۱‬‬ ‫ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ )ﺭﻭﻧﺪﻛﺮﺩﻥ( ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲‬‬
‫‪۴۲‬‬ ‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪. . FS (β, m, min, emax‬‬ ‫‪۱.۲.۲‬‬
‫‪۴۲‬‬ ‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪. . . FS (β, m, emin , emax‬‬ ‫‪۲.۲.۲‬‬
‫‪۴۷‬‬ ‫ﺧﻄﺎﻱ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ)ﺧﻄﺎﻱ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ( ‪. . . . . . . .‬‬ ‫‪۳.۲‬‬
‫‪۴۸‬‬ ‫ﺧﻄﺎﻱ ﻧﺎﺷﻲ ﺍﺯ ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۲‬‬
‫‪۴۹‬‬ ‫ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﻭ ﺗﻮﺍﺑﻊ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۲‬‬
‫‪۵۳‬‬ ‫ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ)ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ( ‪. . . . . .‬‬ ‫‪۲.۴.۲‬‬
‫‪۵۴‬‬ ‫ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۴.۲‬‬
‫‪۵۵‬‬ ‫ﺗﺮﻓﻨﺪﻫﺎ ﻭ ﺗﻮﺻﻴﻪﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۲‬‬

‫‪۶۱‬‬ ‫‪ ۳‬ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮ ﺧﻄﻲ‬

‫‪۶۳‬‬ ‫‪ ۴‬ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ‬

‫‪۶۵‬‬ ‫‪ ۱.۴‬ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۶۵‬‬ ‫ﺑﺮﺩﺍﺭ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۴‬‬
‫‪۶۸‬‬ ‫ﻣﺎﺗﺮﻳﺲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۴‬‬
‫‪۷۵‬‬ ‫ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺧﺎﺹ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۱.۴‬‬
‫‪۷۹‬‬ ‫‪ ۲.۴‬ﻣﻌﺮﻓﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۸۲‬‬ ‫ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۴‬‬
‫‪۸۴‬‬ ‫ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۴‬‬
‫‪۸۴‬‬ ‫ﺭﻭﺵﻫﺎﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺮﺑﻌﻲ ‪. . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۲.۴‬‬
‫‪۸۵‬‬ ‫ﺭﻭﺵ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﺭﻭﺵ ﻛﺮﺍﻣﺮ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۲.۴‬‬
‫‪۸۶‬‬ ‫‪ ۳.۴‬ﺭﻭﺵ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۸۶‬‬ ‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻭ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۴‬‬
‫‪۹۰‬‬ ‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪. . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۳.۴‬‬
‫‪۹۷‬‬ ‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۳.۴‬‬
‫‪۱۰۲‬‬ ‫‪ ∗ ۴.۴‬ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۱۰۵‬‬ ‫‪ ۵.۴‬ﺧﻄﺎ ﻭ ﻛﻨﺘﺮﻝ ﺧﻄﺎ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۱۰۷‬‬ ‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۴‬‬
‫‪۱۱۰‬‬ ‫‪ ۶.۴‬ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﻳﮕﺮ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۱۱۰‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . .‬‬ ‫‪۱.۶.۴‬‬
‫‪۱۱۱‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . .‬‬ ‫‪۲.۶.۴‬‬
‫‪۱۱۳‬‬ ‫ﺗﺸﺨﻴﺺ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . .‬‬ ‫‪۳.۶.۴‬‬
‫‪۱۱۵‬‬ ‫‪ ۷.۴‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪۱۱۵‬‬ ‫ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۷.۴‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﺝ‬
‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ‬ ‫ﻓﺼﻞ ﺻﻔﺮﻡ‪:‬‬

‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ‪۱۲۰ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۷.۴‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ‪۱۲۲ . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۷.۴‬‬
‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ‪۱۲۴ . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۷.۴‬‬

‫‪۱۲۵‬‬ ‫‪ ۵‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ‬

‫‪ ۱.۵‬ﺭﻳﺸﻪ ﻭ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪۱۲۶ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﻣﺎﻫﻴﺖ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪۱۳۱ . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۵‬‬
‫‪ ۲.۵‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ‪۱۳۳ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪ ۳.۵‬ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ‪۱۳۶ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪ ۴.۵‬ﺭﻭﺵﻫﺎﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪۱۳۹ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪۱۴۱ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۵‬‬
‫ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪۱۴۴ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۴.۵‬‬
‫ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪۱۴۹ . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۴.۵‬‬
‫‪ ۵.۵‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪۱۵۱ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪۱۵۳ . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۵‬‬
‫‪ ۶.۵‬ﺭﻭﺵ ﻭﺗﺮﻱ ‪۱۵۴ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪۱۵۶ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۷.۵‬‬
‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪۱۵۹ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۷.۵‬‬

‫‪۱۶۱‬‬ ‫‪ ۶‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ ‪۱۶۲ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۶‬‬
‫‪ ۲.۶‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ ‪۱۶۴ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺧﻄﺎﻱ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪۱۶۵ . . . . . . . . . . . . . . . . . . . . . . . . . . . . n‬‬ ‫‪۱.۲.۶‬‬
‫∗ ﻓﺮﻡﻫﺎﻱ ﺩﻳﮕﺮ ﺑﺮﺍﻱ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ‪۱۶۸ . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۶‬‬

‫‪۱۶۹‬‬ ‫‪ ۷‬ﺩﺭﻭﻧﻴﺎﺑﻲ‬
‫‪ ۱.۷‬ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ‪۱۷۰ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﻳﻚ ﺗﺎﺑﻊ ‪۱۷۱ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۷‬‬
‫ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ‪۱۷۳ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۷‬‬
‫‪ ۲.۷‬ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ‪۱۷۵ . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪۱۷۵ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۷‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪۱۷۷ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۷‬‬
‫‪ ۳.۷‬ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪۱۷۹ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪۱۸۰ . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۷‬‬
‫ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪۱۸۰ . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۳.۷‬‬
‫‪ ۴.۷‬ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻤﺪﺟﻤﻠﻪﺍﻱ ‪۱۸۵ . . . . . . . . . . . . . . . . .‬‬
‫‪ ۵.۷‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺁﻥ ‪۱۸۹ . . . . . . . . . . . . . . . . . . . . . .‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬
‫ﺩ‬

‫‪۱۸۹‬‬ ‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﻻﮔﺮﺍﻧﮋ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۷‬‬

‫‪۱۹۱‬‬ ‫ﻧﺘﺎﻳﺠﻲ ﺍﺯ ﻗﻀﻴﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۵.۷‬‬
‫‪۱۹۲‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۶.۷‬‬
‫‪۱۹۲‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . .‬‬ ‫‪۱.۶.۷‬‬
‫‪۱۹۳‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﭘﺎﻳﺪﺍﺭﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . . .‬‬ ‫‪۲.۶.۷‬‬
‫‪۱۹۳‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۷.۷‬‬
‫‪۱۹۳‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻨﺤﻨﻴﻬﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۸.۷‬‬

‫‪۱۹۵‬‬ ‫‪ ۸‬ﺑﺮﺍﺯﺵ‬
‫‪ ۱.۸‬ﺍﻧﮕﻴﺰﻩ ‪۱۹۶ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪ ۲.۸‬ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ‪۱۹۷ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ ‪۲۰۰ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۸‬‬
‫ﺑﺮﺍﺯﺵ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ‪۲۰۱ . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۸‬‬
‫∗ ﺑﺮﺍﺯﺵ ﻳﻚﮔﺎﻧﻪ ﻭ ﭼﻨﺪﮔﺎﻧﻪ ‪۲۰۳ . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۲.۸‬‬
‫‪ ۳.۸‬ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻭ ﻏﻴﺮﺧﻄﻲ ‪۲۰۵ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺧﻲ ﺑﺮﺍﺯﺵﻫﺎﻱ ﻏﻴﺮﺧﻄﻲ ‪۲۰۶ . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۸‬‬
‫‪ ∗ ۴.۸‬ﻧﻜﻮﻳﻲ ﺑﺮﺍﺯﺵ ‪۲۰۹ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬

‫‪۲۱۱‬‬ ‫‪ ۹‬ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‬

‫‪ ۱.۹‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪۲۱۳ . . . . . . . . . . . . . . . . .‬‬
‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ‪۲۱۴ . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۹‬‬
‫‪ ۲.۹‬ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ‪۲۲۲ . . . . . . . . . . . . . . . . .‬‬
‫‪ ۳.۹‬ﻣﺮﺗﺒﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ‪۲۲۳ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪ ∗ ۴.۹‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ‪۲۲۵ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ‪۲۲۶ . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۹‬‬
‫‪ ۵.۹‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ ‪۲۲۷ . . . . . . . . . . . . . . . . . . . . . . . . .‬‬
‫‪ ۶.۹‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺮﻛﺐ ‪۲۲۸ . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۱‬‬
‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬
‫ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨ‬
‫ﻓﺼﻞ ﺍﻭﻝ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۲‬‬ ‫ﻧﻤﺎﻳﺶﻫﺎﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱‬‬
‫‪۲‬‬ ‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۱‬‬
‫‪۳‬‬ ‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۱‬‬
‫‪۶‬‬ ‫ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ‪. . . . . . . . . . . . . . .‬‬ ‫‪۳.۱.۱‬‬
‫‪۷‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱‬‬
‫‪۸‬‬ ‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۱‬‬
‫‪۸‬‬ ‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪. . . . . . . . . . . .‬‬ ‫‪۲.۲.۱‬‬
‫‪۱۰‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ‪. . . . . . . .‬‬ ‫‪۳.۲.۱‬‬
‫‪۱۴‬‬ ‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮ ﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ‪. . . . . . . .‬‬ ‫‪۴.۲.۱‬‬
‫‪۱۶‬‬ ‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۱‬‬
‫‪۱۸‬‬ ‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ‪. . . . . . . . . . .‬‬ ‫‪۱.۳.۱‬‬
‫‪۲۱‬‬ ‫ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻣﺎﺷﻴﻨﻲ ﺍﻋﺪﺍﺩ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۱‬‬
‫‪۲۲‬‬ ‫∗ ﺟﺰﺋﻴﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻭ ﻣﻀﺎﻋﻒ ‪. . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۱‬‬
‫‪۲۴‬‬ ‫ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۱‬‬
‫‪۲۵‬‬ ‫ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۱‬‬
‫‪۲۶‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﺭﻳﺎﺿﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۵.۱‬‬
‫‪۳۰‬‬ ‫ﺗﻔﺎﻭﺕﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ ‪. . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۵.۱‬‬

‫‪۱‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲‬‬

‫ﻳﻚ ﻫﺪﻑ ﺍﺻﻠﻲ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ‪ ،‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺍﺭﺍﻳﻪ ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎ ﻭ ﺭﻭﺵﻫﺎﻳﻲ ﺑﺮﺍﻱ ﺣﻞ ﻣﺴﺎﻳﻠﻲ ﺩﺭ ﻣﻬﻨﺪﺳﻲ ﻭ ﻋﻠﻮﻡ ﺑﺎ‬
‫ﻛﻤﻚ ﻣﺎﺷﻴﻦﻫﺎ‪ .‬ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ‪ ،‬ﻣﻨﻈﻮﺭ ﺍﺯ ﻣﺎﺷﻴﻦ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﻛﺎﻣﭙﻴﻮﺗﺮﻫﺎ ﻭ ﻳﺎ ﻫﺮ ﻧﻮﻉ ﻭﺳﺎﻳﻞ ﺍﻟﻜﺘﺮﻭﻧﻴﻜﻲ‬
‫ﻛﻪ ﻗﺎﺩﺭ ﺑﻪ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻄﺎﻟﻌﻪ ﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ‬
‫ﺍﻳﻦ ﻛﻪ ﺍﻋﺪﺍﺩ ﺭﻛﻦ ﺍﺻﻠﻲ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﻱ ﻣﻲﺑﺎﺷﻨﺪ‪ ،‬ﻟﺬﺍ ﺩﺭ ﺍﺑﺘﺪﺍ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻄﺎﻟﻌﻪ ﻭ‬
‫ﺑﺮﺭﺳﻲ ﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ ﺑﻪ ﺗﺸﺮﻳﺢ ﺍﺻﻮﻝ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻧﺤﻮﻩ‬
‫ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ )ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ( ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻧﻤﺎﻳﺶﻫﺎﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬ ‫‪۱.۱‬‬

‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻭ ﻧﺎﺷﻤﺎﺭﺍ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﻫﻨﺪﺳﻪ‪ ،‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‪ ،‬ﻳﻚ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ‬
‫ﻣﻲﺷﻮﺩ ﻛﻪ ﻫﺮ ﻧﻘﻄﻪ ﺭﻭﻱ ﺍﻳﻦ ﻣﺤﻮﺭ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺣﺴﺎﺑﺎﻥ‪ ،‬ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﺍﻋﺪﺍﺩ ﺍﺯ ﺭﺷﺘﻪﺍﻱ‬
‫ﺷﺎﻣﻞ ﺍﺭﻗﺎﻡ ﻭ ﻋﻼﻳﻢ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﺭﺷﺘﻪ ﺍﺯ ﺍﺭﻗﺎﻡ ﻭ ﻋﻼﻳﻢ ﺑﻪ ﻧﻤﺎﻳﺶ ﺁﻥ ﻋﺪﺩ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ 14 ،‬ﻭ ‪0.25‬‬
‫ﻧﻤﺎﻳﺶﻫﺎﻳﻲ ﺑﺮﺍﻱ ﺭﺑﻊ ﻋﺪﺩ ﻳﻚ ﺍﺳﺖ ﻭ ﻳﺎ ‪ π‬ﻳﻚ ﻧﻤﺎﻳﺶ ﺑﺮﺍﻱ ﻋﺪﺩﻱ ﺍﺳﺖ ﻛﻪ ﺍﺯ ﺗﻘﺴﻴﻢ ﻣﺤﻴﻂ ﺩﺍﻳﺮﻩ ﺑﻪ ﻗﻄﺮ ﺁﻥ ﺣﺎﺻﻞ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻛﺎﻣﻞ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻻﺯﻡ ﺍﺳﺖ ﻳﻚ ﺳﻴﺴﺘﻢﻫﺎﻱ ﻧﻤﺎﻳﺸﻲ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﻳﻚ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ‪ ،‬ﺍﺑﺘﺪﺍ ﻣﺠﻤﻮﻋﻪﺍﻱ‬
‫ﺍﺯ ﺍﺭﻗﺎﻡ ﻭ ﻋﻼﻳﻢ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻗﻮﺍﻋﺪ ﻭ ﻧﺤﻮﻩ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺍﺭﻗﺎﻡ ﻭ ﻋﻼﻳﻢ ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ‬
‫ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﻭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﺮﺳﻮﻡ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻋﺒﺎﺭﺗﻨﺪ ﺍﺯ‪:‬‬

‫• ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬

‫• ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ )ﻭ ﻳﺎ ﻧﻤﺎﻳﺶﻫﺎﻱ ﻣﺸﺎﺑﻪ ﺑﺎ ﺁﻥ ﺍﺯ ﻗﺒﻴﻞ ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ﻭ ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ(‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺳﻴﺴﺘﻢﻫﺎﻱ ﻧﻤﺎﻳﺸﻲ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬ ‫�‬

‫ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ ۱‬ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‪ ،‬ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪.‬‬
‫· · · ‪± dn dn−1 · · · d1 d0 d−1 · · · d−m‬‬ ‫)‪(۱.۱‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪di‬ﻫﺎ ﺭﻗﻢﻫﺎﻱ ﻋﺪﺩ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﻧﺪ ﻭ ﻣﺘﻌﻠﻖ ﺑﻪ ﻣﺠﻤﻮﻋﻪ }‪ {0, 1, · · · , 9‬ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻧﻤﺎﺩ ” ‪ “ .‬ﺑﻪ ﻧﻘﻄﻪ‬
‫ﻣﻤﻴﺰ ﺍﻋﺸﺎﺭ ‪ ۲‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻧﻤﺎﻳﺶ ﻓﻮﻕ ﻧﻘﻄﻪ ﻣﻤﻴﺰ ﺍﻋﺸﺎﺭ ﻧﻘﺶ ﻣﻬﻤﻲ ﺭﺍ ﺩﺍﺭﺩ ﻭ ﺍﺭﺯﺵ ﺍﺭﻗﺎﻡ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ‬
‫ﺻﻮﺭﺕ ﻛﻪ ‪ i‬ﺍﻣﻴﻦ ﺭﻗﻢ ﺑﻌﺪ ﺍﺯ ﻧﻘﻄﻪ ﻣﻤﻴﺰ ﺩﺍﺭﺍﻱ ﺍﺭﺯﺵ ‪ 10−i‬ﻭ ‪ i‬ﺍﻣﻴﻦ ﺭﻗﻢ ﻗﺒﻞ ﺍﺯ ﻧﻘﻄﻪ ﻣﻤﻴﺰ ﺩﺍﺭﺍﻱ ﺍﺭﺯﺵ ‪ 10i−1‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻟﺬﺍ ﻧﻤﺎﻳﺶ ﻓﻮﻕ ﻣﺮﺑﻮﻁ ﺑﻪ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺑﺎ ﻣﻘﺪﺍﺭ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫∞‬
‫!‬
‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬ ‫‪X‬‬
‫‪i=n‬‬
‫‪−i‬‬
‫‪±‬‬ ‫‪i‬‬
‫‪di 10 +‬‬ ‫‪d−i 10‬‬ ‫‪=±‬‬ ‫‪di 10i .‬‬
‫‪i=0‬‬ ‫‪i=1‬‬ ‫∞‪−‬‬

‫ﺩﺭ ﻧﻤﺎﻳﺶ )‪ ،(۱.۱‬ﺍﮔﺮ ﻛﻠﻴﻪ ﺭﻗﻢﻫﺎﻱ ﺑﻌﺪ ﺍﺯ ﻧﻘﻄﻪ ﺍﻋﺸﺎﺭ ﺻﻔﺮ ﺑﺎﺷﻨﺪ ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﺭﺍ ﺻﺤﻴﺢ ‪ ۳‬ﻣﻲﻧﺎﻣﻴﻢ‪ ،‬ﺩﺭ ﻏﻴﺮ ﺍﻳﻦ ﺻﻮﺭﺕ ﻋﺪﺩ‬
‫ﺭﺍ ﻏﻴﺮﺻﺤﻴﺢ ‪ ۴‬ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺍﮔﺮ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﻏﻴﺮ ﺻﻔﺮ ﺑﻌﺪ ﺍﺯ ﻣﻤﻴﺰ ﺍﻋﺸﺎﺭ ﻣﺘﻨﺎﻫﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﺭﺍ ﻣﺨﺘﻮﻡ ‪ ۵‬ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫√‬
‫ﻣﺜﺎﻝ ‪ ،1.234‬ﻋﺪﺩﻱ ﻣﺨﺘﻮﻡ ﻣﻲﺑﺎﺷﺪ ﻭﻟﻲ ﺍﻋﺪﺍﺩ · · · ‪ 2 = 1.4142....‬ﻭ · · · ‪ π = 3.1415‬ﺍﻋﺪﺍﺩﻱ ﻧﺎﻣﺨﺘﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬
‫‪1 Fixed‬‬ ‫‪Point‬‬ ‫‪4 Non-Integer‬‬
‫‪2 Radix‬‬ ‫‪point‬‬ ‫‪5 Terminated‬‬
‫‪3 Integer‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۳‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻻﺯﻡ ﺑﻪ ﺗﻮﺿﻴﺢ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺑﺮﺧﻲ ﻣﺘﻮﻥ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﺍﺭﻗﺎﻡ ﺑﻌﺪ ﺍﺯ ﻣﻤﻴﺰ ﺍﻋﺸﺎﺭ ﺑﻪﻃﻮﺭ ﺗﻨﺎﻭﺑﻲ ﺗﻜﺮﺍﺭ ﻣﻲﺷﻮﻧﺪ ﺭﺍ ﻧﻴﺰ ﺍﻋﺪﺍﺩ ﻣﺨﺘﻮﻡ‬
‫ﻣﺤﺴﻮﺏ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻣﻨﺎﺑﻊ‪ ،‬ﻋﺪﺩ ‪ 1/3 = 0.3‬ﻳﻚ ﻋﺪﺩ ﻣﺨﺘﻮﻡ ﺑﻪ ﺣﺴﺎﺏ ﻣﻲﺁﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻛﺘﺎﺏ ﺗﻌﺮﻳﻒ‬
‫ﺍﻭﻝ ﻣﺪﻧﻈﺮ ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪.‬‬
‫‪P‬‬
‫∞‪،3 + −‬‬ ‫‪i=−1 9‬‬
‫‪−i‬‬
‫ﻧﻤﺎﻳﺶ )‪ (۱.۱‬ﻳﻚ ﻧﻤﺎﻳﺶ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻪﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺑﺮﺍﻱ ﻋﺪﺩ ‪ ،4‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪= 4‬‬
‫ﻧﻤﺎﻳﺶ ﺩﻳﮕﺮ · · · ‪ 3.9 = 3.999‬ﻧﻴﺰ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ ،‬ﻛﻪ ﻧﻤﺎﻳﺶ ﺩﻭﻡ ﻳﻚ ﻧﻤﺎﻳﺶ ﻧﺎﻣﺨﺘﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺍﮔﺮ ﺍﺯ ﺑﻴﻦ ﻧﻤﺎﻳﺶﻫﺎﻱ‬
‫ﻣﺘﻔﺎﻭﺕ ﻳﻚ ﻋﺪﺩ‪ ،‬ﻧﻤﺎﻳﺶ ﻣﺨﺘﻮﻡ ﺭﺍ ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ ﺁﻥﮔﺎﻩ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩﻱ ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﻀﻤﻴﻦ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺳﻴﺴﺘﻢﻫﺎﻱ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﻣﺒﻨﺎﻫﺎﻱ ﺩﻳﮕﺮ‬

‫ﺩﺭ ﻧﻤﺎﻳﺶ )‪ ،(۱.۱‬ﺍﺯ ﭘﺎﻳﻪ ‪ ۶‬ﻳﺎ ﻣﺒﻨﺎﻱ ‪ ۷‬ﺩﻩ ﺍﺳﺘﻔﺎﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﺭﺯﺵ ﺍﺭﻗﺎﻡ ﺑﻪ ﺻﻮﺭﺕ ﺗﻮﺍﻥﻫﺎﻳﻲ ﺍﺯ ‪ 10‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻭﻟﻲ ﻫﺮﻣﺒﻨﺎﻱ ﺩﻳﮕﺮﻱ ﺭﺍ ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺑﻪﻛﺎﺭ ﺑﺮﺩ‪ .‬ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺩﺭ ﻳﻚ ﻣﺒﻨﺎﻱ ﺩﻟﺨﻮﺍﻩ‬
‫‪ ،β‬ﻛﻪ } ‪ β ∈ {2, 3, . . .‬ﻧﻴﺰ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ ،β‬ﻫﺮ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺸﻲ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫‪.‬‬
‫‪± (dn dn−1 · · · d1 d0 d−1 · · · d−m · · · )β‬‬ ‫)‪(۲.۱‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪di‬ﻫﺎ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ β‬ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﻧﺪ ﻭ ﻣﺘﻌﻠﻖ ﺑﻪ ﻣﺠﻤﻮﻋﻪ }‪ {0, 1, · · · , β − 1‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻧﻤﺎﻳﺶ‬
‫ﻓﻮﻕ ﻣﺮﺑﻮﻁ ﺑﻪ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺑﺎ ﻣﻘﺪﺍﺭ‬
‫∞‬
‫!‬
‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬ ‫‪X‬‬
‫‪i=n‬‬
‫‪−i‬‬
‫‪±‬‬ ‫‪i‬‬
‫‪di β +‬‬ ‫‪d−i β‬‬ ‫‪=±‬‬ ‫‪di β i .‬‬
‫‪i=0‬‬ ‫‪i=1‬‬ ‫∞‪−‬‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻟﺨﻮﺍﻩ ‪ β‬ﻧﻴﺰ ﺑﻪﻃﻮﺭ ﻣﺸﺎﺑﻪ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ‪ ،‬ﻣﺨﺘﻮﻡ ﻭ ﻧﺎﻣﺨﺘﻮﻡ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻣﻤﻜﻦ‬
‫‪ 10‬ﺩﺭ‬
‫‪1‬‬
‫ﺍﺳﺖ ﺩﺭ ﻳﻚ ﻣﺒﻨﺎ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ﻣﺨﺘﻮﻡ ﻭ ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻳﮕﺮ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ﻧﺎﻣﺨﺘﻮﻡ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬
‫
‬
‫ﻣﺒﻨﺎﻱ ‪ 10‬ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ﻣﺨﺘﻮﻡ ‪ 0.1‬ﻭﻟﻲ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ 2‬ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ﻧﺎﻣﺨﺘﻮﻡ ‪ 0.0001100110011 2‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬ ‫�‬

‫ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ‪ ۸‬ﺑﺎ ﻣﺒﻨﺎﻱ ‪ ،β‬ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪.‬‬
‫‪± (0 d−1 · · · d−m · · · )β × β e‬‬ ‫)‪(۳.۱‬‬
‫ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ‪ β ∈ {2, 3, . . . } ،‬ﻭ ‪di‬ﻫﺎ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﻣﺠﻤﻮﻋﻪ }‪ {0, 1, · · · , β − 1‬ﻭﺍﻗﻊﺍﻧﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ‬
‫ﻋﺪﺩ ‪ (0.d−1 d−2 · · · d−m · · · )β‬ﺑﻪ ﻣﺎﻧﺘﻴﺲ ‪ ۹‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ ﻭ ‪ e‬ﻋﺪﺩﻱ ﺻﺤﻴﺢ ﻣﻲﺑﺎﺷﺪ ﻭ ﺑﻪ ﻧﻤﺎ ‪ ۱۰‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻋﺪﺩ ‪ 14‬ﺩﺭ ﻣﺒﻨﺎﻱ ‪ β = 10‬ﺑﻪ ﺻﻮﺭﺕ ‪ +0.25 × 100‬ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﻤﺎﻳﺶ‬
‫‪ 40‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﻪ ﺻﻮﺭﺕ ‪ +0.25 × 101‬ﻭ ‪ +0.25 × 10−1‬ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻋﺪﺩ ‪ 14‬ﺩﺭ ﻣﺒﻨﺎﻱ ‪8‬‬ ‫‪1‬‬
‫‪ 10‬ﻭ‬
‫ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺍﻋﺪﺍﺩ ‪4‬‬
‫ﺑﻪ ﺻﻮﺭﺕ ‪ (0.2)8‬ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻧﻤﺎﻳﺶ ﻋﺪﺩ ‪ 4‬ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﺎ ﻣﺒﻨﺎﻱ ‪ 8‬ﺑﻪ ﺻﻮﺭﺕ ‪ + (0.2)8 × 8‬ﺍﺳﺖ‪.‬‬
‫‪0‬‬ ‫‪1‬‬

‫‪ 40‬ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﺎ ﻣﺒﻨﺎﻱ ‪ 8‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶﻫﺎﻱ ‪(0.5)8 × 80 ،(0.24)8 × 81‬‬
‫‪1‬‬
‫‪ 10‬ﻭ‬
‫ﺍﻋﺪﺍﺩ ‪16 ، 4‬‬
‫‪10‬‬

‫ﻭ ‪ (0.146314631 · · · )8 × 8−1‬ﺍﺳﺖ‪.‬‬

‫‪6 Base‬‬ ‫‪9 Mantissa‬‬ ‫)‪(Significant) (Fraction‬‬

‫‪7 Radix‬‬ ‫‪10 Exponent‬‬
‫‪8 Floating‬‬ ‫‪Point‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴‬‬

‫■ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‬

‫ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺳﻴﺴﺘﻢ ﺍﻋﺪﺍﺩ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶﻫﺎﻱ‬
‫ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺍﻋﺪﺍﺩ ﺯﻳﺮ ﻧﻤﺎﻳﺶﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﻋﺪﺩ ‪ 1‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪.‬‬
‫‪0 1 × β1,‬‬ ‫‪.‬‬
‫‪0 01 × β 2 ,‬‬ ‫‪.‬‬
‫‪0 001 × β 3 .‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺍﻋﺪﺍﺩ‬ ‫ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ‪۱۱‬‬‫ﺑﺮﺍﻱ ﺑﺮﻃﺮﻑ ﻛﺮﺩﻥ ﻣﺸﻜﻞ ﻭ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩﻱ‪ ،‬ﺍﺯ‬
‫ﻏﻴﺮ ﺻﻔﺮ‪ ،‬ﻧﻤﺎ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺗﻨﻈﻴﻢ ﺷﻮﺩ ﻛﻪ ﺍﻭﻟﻴﻦ ﺭﻗﻢ ﺑﻌﺪ ﺍﺯ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺑﺎﺷﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺑﺮﺍﻱ‬
‫ﻋﺪﺩ ﺻﻔﺮ ﺑﻲﻣﻌﻨﻲ ﺍﺳﺖ ﻭ ﻛﺎﺭﺑﺮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺁﻥ ﻧﻤﺎﻳﺸﻲ ﺍﺯ ﻋﺪﺩ ﻏﻴﺮﺻﻔﺮ ﻛﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺷﺪﻩ ﺑﺎﺷﺪ )ﺍﻭﻟﻴﻦ ﺭﻗﻢ ﺑﻌﺪ ﺍﺯ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬
‫ﺩﺭ ﻣﺎﻧﺘﻴﺲ ﺁﻥﻫﺎ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺑﺎﺷﺪ( ﺑﻪ ﻧﻤﺎﻳﺶ ﻧﺮﻣﺎﻝ ﻭ ﻳﺎ ﻓﺮﻡ ﻧﺮﻣﺎﻝ ‪ ۱۲‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﻓﺮﻡ ﻏﻴﺮﻧﺮﻣﺎﻝ ﻳﻚ ﻋﺪﺩ‪ ،‬ﻣﺎﻧﺘﻴﺲ ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﺻﻔﺮ ﻭ ﻛﻮﭼﻚﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﻳﻚ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﻓﺮﻡ‬
‫ﻧﺮﻣﺎﻝ ﻳﻚ ﻋﺪﺩ‪ ،‬ﻣﺎﻧﺘﻴﺲ ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ (0.1)β = β −1‬ﻭ ﻛﻮﭼﻚﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ 1‬ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺩﺭ ﻓﺮﻡ ﻧﺮﻣﺎﻝ ﺑﺮﺍﻱ ﺑﺮﺧﻲ ﺍﺯ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻩ ﻭ ﺩﻭ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻋﺪﺩ ﺣﻘﻴﻘﻲ)ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ(‬ ‫ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺑﺎ ﻣﺒﻨﺎﻱ ‪۱۰‬‬ ‫ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺑﺎ ﻣﺒﻨﺎﻱ ‪۲‬‬
‫‪+2‬‬ ‫‪+0.2 × 10‬‬ ‫‪1‬‬
‫‪+(0.1)2 × 22‬‬
‫‪+0.2‬‬ ‫‪+0.2 × 100‬‬ ‫‪+(0.1100110011)2 × 2−2‬‬
‫‪+300‬‬ ‫‪+0.3 × 103‬‬ ‫‪+(0.1001011)2 × 29‬‬
‫‪+4321.768‬‬ ‫‪+0.4321768 × 104‬‬ ‫‪+(0.100001110 · · · 101)2 × 213‬‬
‫‪−53000‬‬ ‫‪−0.53 × 105‬‬ ‫‪−(0.1100111100001)2 × 216‬‬
‫‪+6720000‬‬ ‫‪+0.672 × 10‬‬ ‫‪7‬‬
‫‪+(0.11001101000101)2 × 223‬‬
‫‪+0.000751‬‬ ‫‪+0.751 × 10−3‬‬ ‫‪+(0.0.1100010011)2 × 2−10‬‬

‫ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻣﻨﺎﺳﺐﺗﺮ ﺍﺯ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺛﺎﺑﺖ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺑﺎ ﺍﻧﺪﺍﺯﻩ ﻋﺪﺩ ﺑﺴﻴﺎﺭ‬
‫ﻛﻮﭼﻚ ﺑﺎ ﺻﻔﺮﻫﺎﻱ ﺯﻳﺎﺩ ﻗﺒﻞ ﺍﺯ ﺭﻗﻢ ﻧﺎﺻﻔﺮ ﻭ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺑﺴﻴﺎﺭ ﺑﺰﺭﮒ ﺑﺎ ﺻﻔﺮﻫﺎﻱ ﺯﻳﺎﺩ ﺩﺭ ﺟﻠﻮﻱ ﻋﺪﺩ‪ ،‬ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬
‫ﻣﻨﺎﺳﺐﺗﺮ ﺍﺯ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﻣﻴﺰﺍﻥ ﺑﺰﺭﮔﻲ ﻭ ﻳﺎ ﻛﻮﭼﻜﻲ ﻋﺪﺩ ﺭﺍ ﺑﻬﺘﺮ ﻭ ﺭﺍﺣﺖﺗﺮ‬
‫ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﻣﺘﻮﻥ ﻭ ﮔﺰﺍﺭﺵﻫﺎﻱ ﻋﻠﻤﻲ‪ ،‬ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﻪ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺮﺟﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫♣ ﺑﺮﺍﻱ ﺳﺎﺩﮔﻲ ﺩﺭ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺩﺭ ﻣﺒﻨﺎﻫﺎﻱ ﻏﻴﺮ ﺩﻩ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻧﻮﺷﺘﻦ ﻣﺒﻨﺎﻱ ﻣﺎﻧﺘﻴﺲ ﺍﺟﺘﻨﺎﺏ ﻛﺮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪،‬‬
‫ﻋﺪﺩ ‪ 4‬ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺑﺎ ﻣﺒﻨﺎﻱ ﺩﻭ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ‪ (0.1)2 × 23‬ﺍﺳﺖ‪ .‬ﻣﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺶ ﺳﺎﺩﻩﺗﺮ‬
‫‪ 0.1 × 23‬ﺭﺍ ﻧﻴﺰ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎﻳﺪ ﺗﻮﺟﻪ ﺩﺍﺷﺖ ﻛﻪ ﺩﺭ ﻣﻮﺭﺩ ﻣﺒﻨﺎﻱ ﻣﺎﻧﺘﻴﺲ ﺍﺑﻬﺎﻣﻲ ﺍﻳﺠﺎﺩ ﻧﻤﻲﺷﻮﺩ‪ ،‬ﭼﺮﺍ ﻛﻪ ﻣﺒﻨﺎﻱ ﺩﻭ ﺑﻮﺩﻥ‬
‫ﻣﺎﻧﺘﻴﺲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻭﺟﻮﺩ ﭘﺎﻳﻪ ‪ 2‬ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﺗﺸﺨﻴﺺ ﺩﺍﺩ‪.‬‬
‫♣ ﺑﻪ ﻣﻨﻈﻮﺭ ﺳﺎﺩﮔﻲ ﺩﺭ ﻧﻮﺷﺘﻦ ﺍﻋﺪﺍﺩ ﺩﺭ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﺍﺯ ﻧﻤﺎﺩ ‪ E‬ﻭ ﻳﺎ ‪ e‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﻧﻤﺎﻳﺶ‬
‫‪ a × 10n‬ﺑﺎ ﻧﻤﺎﻳﺶ ‪ a E n‬ﻭ ﻳﺎ ‪ a en‬ﺟﺎﻳﮕﺰﻳﻦ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻧﻤﺎﺩﻫﺎ ﺑﺮﺍﻱ ﺻﻔﺤﻪ ﻧﻤﺎﻳﺶ ﻣﺎﺷﻴﻦ ﺣﺴﺎﺏﻫﺎ ﻭ ﺧﺮﻭﺟﻲﻫﺎﻱ‬
‫ﻣﺘﻨﻲ ﻣﻨﺎﺳﺐﺗﺮ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﻋﺪﺩ ‪ 0.12309 × 10−5‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‬
‫‪.‬‬ ‫‪.‬‬
‫‪ 0 12309 e−5.‬ﻳﺎ ‪0 12309 E −5‬‬
‫ﻧﻤﺎﻳﺶﻫﺎﻱ ﺩﻳﮕﺮﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ﻭ ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ ﻧﻴﺰ ﺭﺍﻳﺞ ﻣﻲﺑﺎﺷﺪ ﺍﻳﻦ ﻧﻤﺎﻳﺶﻫﺎ ﺷﺒﻴﻪ ﺑﻪ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﻣﻌﺮﻓﻲ ﺁﻥﻫﺎ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪11 Normalization‬‬ ‫‪12 Normalized‬‬ ‫‪form‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫■ ◦ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻋﻠﻤﻲ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬

‫ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ﻫﻤﺎﻧﻨﺪ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺍﺳﺖ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﻣﺎﻧﺘﻴﺲ ﺑﻪ ﺻﻮﺭﺕ ﻣﺘﻔﺎﻭﺗﻲ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﺩ‪ .‬ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ‬
‫ﺩﺭ ﻣﺒﻨﺎﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﻗﺎﺑﻞ ﺗﻌﺮﻳﻒ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺑﺮﺍﻱ ﺳﺎﺩﮔﻲ‪ ،‬ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﺑﺎ ﻣﺒﻨﺎﻱ ﺩﻩ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ‪۱۳‬‬

‫ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‬

‫‪.‬‬
‫‪± d0 d−1 d−2 . . . · · · × 10e‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪di‬ﻫﺎ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺑﻪ ﻣﻨﻈﻮﺭ ﺣﻔﻆ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﻏﻴﺮﺻﻔﺮ‪ ،‬ﺷﺮﻁ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ‬
‫‪ d0 ̸= 0‬ﺍﻋﻤﺎﻝ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﻣﺎﻧﺘﻴﺲ ﻋﺪﺩﻱ ﺑﻴﻦ ‪ 1‬ﻭ ‪ 10‬ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ‪s := d0.d−1 d−2 . . .‬‬
‫ﺁﻥﮔﺎﻩ ‪.1 ≤ s ≤ 10‬‬
‫ﺑﺮﺍﻱ ﺳﻬﻮﻟﺖ ﺩﺭ ﻧﻮﺷﺘﻦ ﺍﻋﺪﺍﺩ ﻋﻠﻤﻲ ﻧﻴﺰ ﺍﺯ ﻧﻤﺎﺩ ‪ E‬ﻭ ﻳﺎ ‪ e‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ﺑﺮﺧﻲ ﺍﺯ ﺍﻋﺪﺍﺩ‬
‫ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬ ‫ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ‬ ‫ﻧﻤﺎﻳﺶ ﻋﻠﻤﻲ ﺑﺎ ‪e‬‬

‫‪+2‬‬ ‫‪+2.0 × 10‬‬ ‫‪0‬‬
‫‪+2.0e+0‬‬
‫‪+300‬‬ ‫‪+3.0 × 10‬‬ ‫‪2‬‬
‫‪+3.0e+2‬‬
‫‪+4321.768‬‬ ‫‪+4.321768 × 103‬‬ ‫‪+4.321768e+3‬‬
‫‪−53000‬‬ ‫‪−5.3 × 10‬‬ ‫‪4‬‬
‫‪−5.3e+4‬‬
‫‪+6720000000‬‬ ‫‪+6.72 × 10‬‬ ‫‪9‬‬
‫‪+6.72 e+9‬‬
‫‪−1‬‬
‫‪+0.2‬‬ ‫‪+2.0 × 10‬‬ ‫‪+2.0e − 1‬‬
‫‪−9‬‬
‫‪+0.00000000751‬‬ ‫‪+7.51 × 10‬‬ ‫‪+7.51 e−9‬‬

‫■ ◦ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻬﻨﺪﺳﻲ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬

‫ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ ‪ ۱۴‬ﺍﺯ ﻣﺒﻨﺎﻱ ‪ 10‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﻣﺸﺎﺑﻪ ﻧﻤﺎﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺍﺳﺖ‪ ،‬ﺑﺎ ﺍﻳﻦ‬
‫ﺗﻔﺎﻭﺕ ﻛﻪ ﻣﺎﻧﺘﻴﺲ ﺍﻋﺪﺍﺩ ﻏﻴﺮﺻﻔﺮ ﻋﺪﺩﻱ ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ 1‬ﻭ ﻛﻮﭼﻚﺗﺮ ﺍﺯ ‪ 1000‬ﺍﺳﺖ ﻭ ﻧﻤﺎ ﻋﺪﺩﻱ ﺻﺤﻴﺢ ﻭ ﻣﻀﺮﺏ ‪3‬‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻧﻤﺎﻳﺶ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺑﺎ ﻧﻤﺎﺩ ﻣﻬﻨﺪﺳﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪.‬‬
‫‪± d2 d1 d0 d−1 d−2 . . . · · · × 103e‬‬
‫ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﻧﻴﺰ ﺍﺯ ﻧﻤﺎﺩ ‪ E‬ﻭ ﻳﺎ ‪ e‬ﺑﺮﺍﻱ ﺳﺎﺩﮔﻲ ﺩﺭ ﻧﻮﺷﺘﻦ ﻋﺪﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ ﺑﺮﺧﻲ ﺍﺯ‬
‫ﺍﻋﺪﺍﺩ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬ ‫ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ‬ ‫ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ ﺑﺎ ‪E‬‬

‫‪+2‬‬ ‫‪+2.0 × 10‬‬ ‫‪0‬‬
‫‪+2.0 E +0‬‬
‫‪+300‬‬ ‫‪+300.0 × 100‬‬ ‫‪+300.0 E +0‬‬
‫‪+4321.768‬‬ ‫‪+4.321768 × 103‬‬ ‫‪+4.321768 E +3‬‬
‫‪−53000‬‬ ‫‪−53.0 × 10‬‬ ‫‪3‬‬
‫‪−53.0 E +3‬‬
‫‪+6720000000‬‬ ‫‪+6.72 × 10‬‬ ‫‪9‬‬
‫‪+6.72 E +9‬‬
‫‪−3‬‬
‫‪+0.2‬‬ ‫‪+200.0 × 10‬‬ ‫‪+200.0 E − 3‬‬
‫‪−9‬‬
‫‪+0.00000000751‬‬ ‫‪+7.51 × 10‬‬ ‫‪+7.51 E −9‬‬
‫‪13 scientific‬‬ ‫)‪notation(scientific form‬‬ ‫‪14 Engineering‬‬ ‫)‪notation(engineering form‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶‬‬

‫ﻓﻠﺴﻔﻪ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﺁﻥ ﺍﺳﺖ ﻛﻪ‪ ،‬ﺩﺭ ﻧﻤﺎﻳﺶ ﻋﺪﺩ ﺑﺎ ﻧﻤﺎﺩ ﻣﻬﻨﺪﺳﻲ‪ ،‬ﺑﻪ ﺳﺎﺩﮔﻲ ﻣﻲﺗﻮﺍﻥ ﺍﻧﺪﺍﺯﻩ ﻋﺪﺩ ﺭﺍ ﺑﺮ ﺣﺴﺐ ﭘﻴﺸﻮﻧﺪﻫﺎﻱ‬
‫ﻣﺘﺮﻳﻚ ‪ ۱۵‬ﺍﺯ ﻗﺒﻴﻞ ﻛﻴﻠﻮ‪ ،‬ﻣﮕﺎ‪ ،‬ﻣﻴﻠﻲ‪ ،‬ﻧﺎﻧﻮ ﻭ ﻏﻴﺮﻩ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺍﻳﻦ ﭘﻴﺸﻮﻧﺪﻫﺎ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫ﻗﺎﺑﻞ ﺍﺳﺘﻔﺎﺩﻩ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺑﺎ ﺍﻧﺪﺍﺯﻩ‬ ‫ﻗﺎﺑﻞ ﺍﺳﺘﻔﺎﺩﻩ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺑﺎ ﺍﻧﺪﺍﺯﻩ‬
‫ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﻫﺰﺍﺭ‬ ‫ﻛﻮﭼﮓﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﻳﻚ ﻫﺰﺍﺭﻡ‬
‫ﻣﻘﺪﺍﺭ ﻧﻤﺎﺩ ﻧﺎﻡ ﭘﻴﺸﻮﻧﺪ‬ ‫ﻣﻘﺪﺍﺭ ﻧﻤﺎﺩ ﻧﺎﻡ ﭘﻴﺸﻮﻧﺪ‬
‫‪Kilo‬‬ ‫‪K‬‬ ‫‪10+3‬‬ ‫‪milli‬‬ ‫‪m‬‬ ‫‪10−3‬‬
‫‪Mega‬‬ ‫‪M‬‬ ‫‪10+6‬‬ ‫‪micro‬‬ ‫‪µ‬‬ ‫‪10−6‬‬
‫‪Giga‬‬ ‫‪G‬‬ ‫‪10+9‬‬ ‫‪nano‬‬ ‫‪n‬‬ ‫‪10−9‬‬
‫‪Tera‬‬ ‫‪T‬‬ ‫‪10‬‬ ‫‪+12‬‬
‫‪pico‬‬ ‫‪p‬‬ ‫‪10−12‬‬
‫‪Peta‬‬ ‫‪P‬‬ ‫‪10+15‬‬ ‫‪femto‬‬ ‫‪f‬‬ ‫‪10−15‬‬
‫‪Exa‬‬ ‫‪E‬‬ ‫‪10+18‬‬ ‫‪atto‬‬ ‫‪a‬‬ ‫‪10−18‬‬
‫‪Zetta‬‬ ‫‪Z‬‬ ‫‪10+21‬‬ ‫‪zepto‬‬ ‫‪z‬‬ ‫‪10−21‬‬
‫‪Yotta‬‬ ‫‪Y‬‬ ‫‪10+24‬‬ ‫‪yocto‬‬ ‫‪y‬‬ ‫‪10−24‬‬

‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻋﺪﺩ ‪ 32560000000‬ﻛﻪ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ﻣﻬﻨﺪﺳﻲ ‪ 32.56 E 9‬ﺍﺳﺖ ﻭ ﻣﻲﺗﻮﺍﻥ ﺁﻥ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪ 32.56 Giga‬ﻧﻴﺰ ﻧﻮﺷﺖ‪.‬‬

‫ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺩﺭ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬ ‫�‬

‫ﻣﺎ ﺍﻧﺠﺎﻡ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺍﻋﺪﺍﺩ ﺭﺍ ﺩﺭ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺁﻣﻮﺧﺘﻪﺍﻳﻢ‪ .‬ﺑﻪ ﺭﺍﺣﺘﻲ ﻣﻲﺗﻮﺍﻥ ﺭﻭﺵ ﺍﻧﺠﺎﻡ ﭼﻬﺎﺭ ﻋﻤﻞ‬
‫ﺍﺻﻠﻲ ﺭﻭﻱ ﺍﻋﺪﺍﺩ ﺑﺎ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺭﺍ ﺍﺳﺘﻨﺘﺎﺝ ﻧﻤﻮﺩ‪ ،‬ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺁﻥ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺟﻤﻊ)ﺗﻔﺮﻳﻖ( ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬

‫ﺍﮔﺮ ﻧﻤﺎﻱ ﺩﻭ ﻋﺪﺩ ﻳﻜﺴﺎﻥ ﺑﺎﺷﻨﺪ‪ ،‬ﺑﺮﺍﻱ ﺟﻤﻊ)ﺗﻔﺮﻳﻖ( ﺩﻭ ﻋﺪﺩ ﻛﺎﻓﻲ ﺍﺳﺖ ﻣﺎﻧﺘﻴﺲﻫﺎﻱ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ)ﺗﻔﺮﻳﻖ( ﻛﻨﻴﻢ ﻭ ﻧﻤﺎ‬
‫ﻫﻢ ﻛﻪ ﻫﻤﺎﻥ ﻧﻤﺎﻱ ﻣﺸﺘﺮﻙ ﺍﺳﺖ‪ .‬ﺍﻟﺒﺘﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﺑﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻧﻴﺎﺯ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺩﺭ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ ﺩﻭ ﻋﺪﺩ‪ ،‬ﻧﻤﺎﻫﺎ ﻳﻜﺴﺎﻥ ﻧﺒﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﺎﻳﺶ ﻋﺪﺩ ﺑﺎ ﻧﻤﺎﻱ ﻛﻮﭼﻚﺗﺮ ﺭﺍ ﻃﻮﺭﻱ ﺗﻐﻴﻴﺮ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﻧﻤﺎﻱ ﺁﻥ ﺑﺎ ﻧﻤﺎﻱ ﻋﺪﺩ‬
‫ﺩﻳﮕﺮ ﻳﻜﺴﺎﻥ ﮔﺮﺩﺩ‪ .‬ﺳﭙﺲ ﻋﻤﻞ ﺟﻤﻊ)ﺗﻔﺮﻳﻖ( ﺭﺍ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺍﻟﺒﺘﻪ ﻣﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺶ ﻋﺪﺩ ﺑﺎ ﻧﻤﺎﻱ ﺑﺰﺭﮒﺗﺮ ﺭﺍ ﻧﻴﺰ ﺗﻐﻴﻴﺮ ﺩﺍﺩ‬
‫ﻛﻪ ﻧﻤﺎﻱ ﺁﻥ ﺑﺎ ﻧﻤﺎﻱ ﻋﺪﺩ ﺩﻳﮕﺮ ﻳﻜﺴﺎﻥ ﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﺑﻌﺪ ﺍﺯ ﺟﻤﻊ)ﺗﻔﺮﻳﻖ(‪ ،‬ﺑﻪ ﺍﺣﺘﻤﺎﻝ ﺯﻳﺎﺩ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻭﺟﻮﺩ‬
‫ﺩﺍﺭﺩ‪.‬‬

‫■ ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬

‫ﺑﺮﺍﻱ ﺿﺮﺏ ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﺑﺎﻳﺪ ﻣﺎﻧﺘﻴﺲﻫﺎ ﺭﺍ ﺩﺭ ﻫﻢ ﺿﺮﺏ ﻛﺮﺩ ﻭ ﻧﻤﺎﻫﺎ ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺗﻘﺴﻴﻢ ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ‪ ،‬ﻣﺎﻧﺘﻴﺲﻫﺎ ﺑﺮ ﻫﻢ ﺗﻘﺴﻴﻢ ﻭ ﻧﻤﺎﻫﺎ ﺍﺯ ﻳﻚﺩﻳﮕﺮ ﺗﻔﺮﻳﻖ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫■ ﻣﻘﺎﻳﺴﻪ ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‬

‫ﺑﺮﺍﻱ ﻣﻘﺎﻳﺴﻪ ﺩﻭ ﻋﺪﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﺩﺭ ﺍﺑﺘﺪﺍ ﻋﻼﻣﺖ ﺩﻭ ﻋﺪﺩ ﻣﻘﺎﻳﺴﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻋﺪﺩ ﺑﺎ ﻋﻼﻣﺖ ﻣﺜﺒﺖ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﻋﺪﺩ ﺑﺎ ﻋﻼﻣﺖ‬
‫ﻣﻨﻔﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﻣﻘﺎﻳﺴﻪ ﺩﻭ ﻋﺪﺩ ﻣﺜﺒﺖ‪ ،‬ﺍﻭﻝ ﻧﻤﺎﻱ ﺩﻭ ﻋﺪﺩ ﻣﻘﺎﻳﺴﻪ ﻣﻲﺷﻮﺩ)ﺍﻟﺒﺘﻪ ﻧﻤﺎﻳﺶ ﺩﻭ ﻋﺪﺩ ﺑﺎﻳﺪ ﻧﺮﻣﺎﻝ ﺑﺎﺷﺪ ﻭ ﻧﻤﺎﻱ‬
‫‪15 metric‬‬ ‫‪prefix‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻋﺪﺩ ‪ 0‬ﻧﻴﺰ ﺑﺮﺍﻳﺮ ﺻﻔﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﻮﺩ(‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺁﻥ ﻋﺪﺩﻱ ﻛﻪ ﻧﻤﺎﻱ ﺁﻥ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ ﺑﻪ ﻋﻨﻮﺍﻥ ﻋﺪﺩ ﺑﺰﺭﮒﺗﺮ‬
‫ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﻧﻤﺎﻱ ﺩﻭ ﻋﺪﺩ ﻣﺜﺒﺖ ﻳﻜﺴﺎﻥ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺁﻥ ﻋﺪﺩﻱ ﺑﺰﺭﮒﺗﺮ ﺍﺳﺖ ﻛﻪ ﻣﺎﻧﺘﻴﺲ ﺁﻥ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ .‬ﺩﺭ‬
‫ﻣﻮﺭﺩ ﺍﻋﺪﺍﺩ ﻣﻨﻔﻲ ﻧﻴﺰ ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ ﻣﻲﺗﻮﺍﻥ ﻣﻘﺎﻳﺴﻪ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱−۱‬ﺟﻤ ‪ ،‬ﺗﻔﺮﯾﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﯿﻢ ﺩﻭ ﻋﺪﺩ ﻣﻤﯿﺰ ﺷﻨﺎﻭﺭ(‬

‫▷ ﺩﻭ ﻋﺪﺩ ‪ x‬ﻭ ‪ y‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺁﻥﻫﺎ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ β = 10‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪x = 0.9341 × 103 ,‬‬ ‫‪y = 0.321 × 10−2 .‬‬
‫ﺑﺮﺍﻱ ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﺩﻭ ﻋﺪﺩ‪ ،‬ﺍﺑﺘﺪﺍ ﺑﺎﻳﺪ ﻧﻤﺎﻫﺎ ﺩﺭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺭﺍ ﻳﻜﺴﺎﻥ ﻛﻨﻴﻢ‪ .‬ﻧﻤﺎ ﺩﺭ ﻧﻤﺎﻳﺶ ‪ y‬ﻛﻮﭼﻚﺗﺮ ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﻧﻤﺎﻳﺶ ‪y‬‬
‫ﺭﺍ ﻃﻮﺭﻱ ﺗﻐﻴﻴﺮ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﻧﻤﺎ ﺑﺮﺍﺑﺮ ‪ 3‬ﺷﻮﺩ‬
‫‪y = 0.321 × 10−2 = 0.00000321 × 103 .‬‬
‫ﺣﺎﻝ ﻣﻲﺗﻮﺍﻥ ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‬
‫‬ ‫‬ ‫‬ ‫‬
‫‪x + y = 0.9341 × 10 + 0.00000321 × 103 = 0.93410321 × 103‬‬
‫‪3‬‬

‫‬ ‫ ‬ ‫‬
‫‪x − y = 0.9341 × 103 − 0.00000321 × 103 = 0.93409679 × 103 .‬‬
‫ﺑﺮﺍﻱ ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ‪ ،‬ﻟﺰﻭﻣﻲ ﺑﻪ ﻳﻜﺴﺎﻥﺳﺎﺯﻱ ﻧﻤﺎﻫﺎ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪.‬‬
‫‬ ‫ ‬ ‫‬
‫‪x ∗ y = 0.9341 × 103 ∗ 0.321 × 10−2‬‬

‫‪=(0.9341 ∗ 0.321) × 103+(−2) = 0.2998461 × 101 ,‬‬

‫‬ ‫ ‬ ‫‬
‫‪x/y = 0.9341 × 103 / 0.321 × 10−2‬‬

‫‪=(0.9341/0.321) × 103−(−2) = 2.9099 · · · × 105 = 0.29099 · · · × 106 .‬‬

‫ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻓﻮﻕ‪ ،‬ﺑﻪ ﻏﻴﺮ ﺍﺯ ﻋﻤﻞ ﺗﻘﺴﻴﻢ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﻭﺟﻮﺩ ﻧﺪﺍﺷﺖ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﻛﻠﻴﻪ ﻋﻤﻠﻴﺎﺕ‬
‫ﺣﺴﺎﺑﻲ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻌﺪ ﺍﺯ ﻣﺤﺎﺳﺒﻪ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻧﺘﻴﺠﻪ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪.‬‬

‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫‪۲.۱‬‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﻛﻠﻴﺎﺕ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪ .‬ﻗﺒﻞ ﺍﺯ ﺷﺮﻭﻉ ﺑﺤﺚ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻧﻜﺎﺕ ﺯﻳﺮ ﻣﺪﻧﻈﺮ‬
‫ﻗﺮﺍﺭ ﮔﻴﺮﻧﺪ‪.‬‬
‫• ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺍﺑﺘﺪﺍ ﻻﺯﻡ ﺍﺳﺖ ﻳﻚ ﻣﺒﻨﺎﻱ ‪ β‬ﺍﻧﺘﺨﺎﺏ ﺷﻮﺩ ﻭ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺁﻥ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ .‬ﺑﻪ ﻟﺤﺎﻅ‬
‫ﺗﺌﻮﺭﻱ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻲﺗﻮﺍﻧﻨﺪ ﺍﺯ ﻫﺮ ﻣﺒﻨﺎﻳﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻨﺪ‪ .‬ﺍﻣﺎ ﺑﻪ ﺧﺎﻃﺮ ﻣﻼﺣﻈﺎﺕ ﺳﺨﺖﺍﻓﺰﺍﺭﻱ ﺩﺭ ﺍﻛﺜﺮ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺯ ﻣﺒﻨﺎﻱ‬
‫ﺩﻭ )ﻭ ﻳﺎ ﻣﺒﻨﺎﻫﺎﻱ ﻭﺍﺑﺴﺘﻪ ﺑﻪ ﺁﻥ ﻳﻌﻨﻲ ‪ ۸‬ﻭ ‪ (۱۶‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫• ﻣﺎﺷﻴﻦﻫﺎ ﺍﻋﺪﺍﺩ ﺭﺍ ﺩﺭ ﺣﺎﻓﻈﻪ ﺧﻮﺩ ﺫﺧﻴﺮﻩ ﻣﻲﻛﻨﻨﺪ ﻭ ﺣﺎﻓﻈﻪ ﻣﺎﺷﻴﻦﻫﺎ ﻣﺘﻨﺎﻫﻲ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻣﺎﺷﻴﻦﻫﺎ ﻳﻚ ﺣﺎﻓﻈﻪ ﻣﺘﻨﺎﻫﻲ‬
‫)ﺑﺎ ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﺭﻗﻢ( ﺑﺮﺍﻱ ﻫﺮ ﻋﺪﺩ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻧﺪ‪.‬‬
‫• ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ﻳﻚ ﻋﺪﺩ‪ ،‬ﻋﻼﻭﻩ ﺑﺮ ﻣﻘﺪﺍﺭ ﻋﺪﺩ ﺑﺎﻳﺪ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ‪ ۱۶‬ﺁﻥ ﻋﺪﺩ ﻧﻴﺰ ﺗﻌﻴﻴﻦ ﺷﻮﺩ‪ .‬ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ‬
‫ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﻋﺪﺩ ﺑﻪ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ﺁﻥ ﻋﺪﺩ ﻧﻴﺰ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﺩﺭ ﺯﺑﺎﻥ ‪ ،C‬ﺳﻪ‬
‫ﻣﺘﻐﻴﺮ ﺑﺎ ﺳﻪ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ﻣﺘﻔﺎﻭﺕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻛﻨﻴﻢ‪:‬‬
‫‪16 Data‬‬ ‫‪type‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸‬‬

‫;‪int n‬‬ ‫;‪float x; double y‬‬

‫ﺣﺎﻝ‪ ،‬ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﻣﺘﻐﻴﺮ ‪ n‬ﻣﺘﻔﺎﻭﺕ ﺍﺯ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪ x‬ﻭ ‪ y‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺍﻧﺪﺍﺯﻩ ﺣﺎﻓﻈﻪ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ‬
‫ﺷﺪﻩ ﺑﺮﺍﻱ ﻣﺘﻔﻴﺮﻫﺎﻱ ‪ x‬ﻭ ‪ y‬ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺑﺨﺶ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﻭ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﻳﻚ ﻋﺪﺩ‬
‫ﺑﺎ ﻗﺎﻟﺐ ﺻﺤﻴﺢ ﻛﺎﻓﻲ ﺍﺳﺖ ﺍﺭﻗﺎﻡ ﺁﻥ ﺭﺍ ﺩﺭ ﻛﺎﻣﭙﻴﻮﺗﺮ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ .‬ﺍﻣﺎ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﻛﻤﻲ ﭼﺎﻟﺸﻲﺗﺮ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺍﺑﺘﺪﺍ‪ ،‬ﺑﻪ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺍﺷﺎﺭﻩ ﻣﻲﺷﻮﺩ ﺳﭙﺲ ﺑﺎ ﺍﻟﮕﻮﺑﺮﺩﺍﺭﻱ ﺍﺯ ﻧﻤﺎﻳﺶﻫﺎﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻭ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ‪ ،‬ﺩﻭ ﻧﻮﻉ‬
‫ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺍﺭﺍﻳﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫�‬

‫ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ ﻛﻪ ﺍﺯ ﻣﺒﻨﺎﻱ ‪ β‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪ .‬ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻳﻚ ﻋﺪﺩ ﺑﺎ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ﺻﺤﻴﺢ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‪ ،‬ﻗﺴﻤﺘﻲ‬
‫ﺍﺯ ﺣﺎﻓﻈﻪ ﺑﺎ ﺍﻧﺪﺍﺯﻩ ﻣﺸﺨﺺ ﺑﺮﺍﻱ ﺁﻥ ﻋﺪﺩ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﭼﻮﻥ‪ ،‬ﺣﺎﻓﻈﻪ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺩﺍﺭﺍﻱ ﻃﻮﻝ ﻣﺘﻨﺎﻫﻲ ﺍﺳﺖ‪،‬‬
‫ﻟﺬﺍ ﻫﻤﻪ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺭﺍ ﻧﻤﻲﺗﻮﺍﻥ ﺫﺧﻴﺮﻩ ﻧﻤﻮﺩ‪ .‬ﺑﻠﻜﻪ ﺗﻌﺪﺍﺩﻱ ﺍﺯ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﻛﻪ ﻣﺘﻮﺍﻟﻲ ﻧﻴﺰ ﻣﻲﺑﺎﺷﻨﺪ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺩﺭ ﺣﺎﻓﻈﻪ‬
‫ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺫﺧﻴﺮﻩ ﻧﻤﻮﺩ‪ .‬ﻳﻚ ﺭﺍﻩ ﺳﺮﺭﺍﺳﺖ ﻭ ﺳﺎﺩﻩ ﺩﺭ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﻭ ﻫﻤﭽﻨﻴﻦ‬
‫ﻋﻼﻣﺖ ﻋﺪﺩ ﺭﺍ ﺩﺭ ﺣﺎﻓﻈﻪ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺫﺧﻴﺮﻩ ﻧﻤﻮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﻛﺎﻣﭙﻴﻮﺗﺮﻫﺎﻱ ﺍﻣﺮﻭﺯﻱ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻣﻨﺎﺳﺐﺗﺮﻱ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺩﺭﺱ‪ ،‬ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﻭ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺁﻥﻫﺎ ﭼﻨﺪﺍﻥ ﺑﻪ ﻛﺎﺭ ﻧﻤﻲﺁﻳﺪ ﻭ ﻟﺬﺍ ﺑﻪ ﺁﻥ ﻭ ﺟﺰﺋﻴﺎﺕ ﻣﺮﺑﻮﻁ ﺑﻪ ﺁﻥ ﭘﺮﺩﺍﺧﺘﻪ‬
‫ﻧﻤﻲﺷﻮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎ ﻣﻄﺎﻟﻌﻪ ﻧﺤﻮﻩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ )ﺩﺭ ﺩﻭ ﺑﺨﺶ ﺑﻌﺪ(‪ ،‬ﺧﻮﺍﻧﻨﺪﻩ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺮﺧﻲ ﺟﺰﺋﻴﺎﺕ ﺩﺭ ﻣﻮﺭﺩ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ‬
‫ﺻﺤﻴﺢ ﺭﺍ ﻧﻴﺰ ﺍﺳﺘﻨﺘﺎﺝ ﻧﻤﺎﻳﺪ‪.‬‬

‫◦ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬ ‫�‬

‫ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺑﺎ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ )ﺑﺎ ﻣﺒﻨﺎﻱ ‪ (β‬ﺍﻟﮕﻮﺑﺮﺩﺍﺭﻱ ﻧﻤﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪ ،‬ﻣﻘﺪﺍﺭﻱ ﻣﺘﻨﺎﻫﻲ ﺍﺯ ﺣﺎﻓﻈﻪ ﺑﺎ ﻗﺎﻟﺐ ﺑﻨﺪﻱ ﺯﻳﺮ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‬

‫• ﻳﻚ ﺑﻴﺖ ﺑﺮﺍﻱ ﻋﻼﻣﺖ‪.‬‬

‫• ‪ n‬ﺭﻗﻢ ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﺭﻗﺎﻡ ﻗﺒﻞ ﺍﺯ ﻣﻤﻴﺰ ﺍﻋﺸﺎﺭ)ﻗﺴﻤﺖ ﺻﺤﻴﺢ(‪،‬‬
‫• ‪ m‬ﺭﻗﻢ ﺑﺮﺍﻱ ﺍﺭﻗﺎﻡ ﻣﻤﻴﺰ ﺑﻌﺪ ﺍﺯ ﺍﻋﺸﺎﺭ )ﻗﺴﻤﺖ ﺍﻋﺸﺎﺭﻱ(‪،‬‬

‫ﺷﻤﺎﻱ ﺍﻳﻦ ﻗﺎﻟﺐﺑﻨﺪﻱ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻗﺴﻤﺖ ﺻﺤﻴﺢ‬ ‫ﻗﺴﻤﺖ ﺍﻋﺸﺎﺭ‬
‫ﺑﻴﺖ ﻋﻼﻣﺖ‬

‫) ‪ n‬ﺭﻗﻢ(‬ ‫) ‪ m‬ﺭﻗﻢ(‬
‫‪. .‬‬ ‫‪.‬‬ ‫‪...‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪....‬‬ ‫‪.‬‬ ‫‪.‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﻧﻤﺎﻳﺶ ﻓﻮﻕ‪ ،‬ﻫﺮ ﺭﻗﻢ ﺍﺯ ﺗﻌﺪﺍﺩﻱ ﺑﻴﺖ ﺗﺸﻜﻴﻞ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺍﮔﺮ ‪ β = 2‬ﺁﻥﮔﺎﻩ ﻫﺮ ﺭﻗﻢ ﻣﺘﺸﻜﻞ ﺍﺯ ﻳﻚ‬
‫ﺑﻴﺖ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺍﮔﺮ ﻣﺎﺷﻴﻦ ﺍﺯ ﻣﺒﻨﺎﻱ ﺩﻭ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺮﺍﻱ ﻫﺮ ﺭﻗﻢ ﻳﻚ ﺑﻴﺖ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﺒﻨﺎﻫﺎﻱ‬
‫‪ β = 8‬ﻭ ‪ β = 16‬ﻫﺮ ﺭﻗﻢ ﺑﻪ ﺗﺮﺗﻴﺐ ﻣﻌﺎﺩﻝ ﺳﻪ ﻭ ﭼﻬﺎﺭ ﺑﻴﺖ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﻌﺮﻓﻲ ﺷﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﻓﻮﻕ‪ ،‬ﻓﻘﻂ ﺍﻋﺪﺍﺩﻱ ﺑﺎ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺭﻗﻢ ﺻﺤﻴﺢ ﻭ ﺣﺪﺍﻛﺜﺮ ‪ m‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻓﺮﻡ ﻛﻠﻲ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪.‬‬
‫‪±(dn−1 dn−2 · · · d1 d0 d−1 · · · d−m )β ,‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪di‬ﻫﺎ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ β‬ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﺩﺭ ﻣﺠﻤﻮﻋﻪ }‪ {0, 1, . . . , β − 1‬ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪ .‬ﻣﺠﻤﻮﻋﻪﻱ ﺍﻋﺪﺍﺩ ﻓﻮﻕ‬
‫ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ )‪ F0 (β, m, n‬ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪:‬‬
‫‪n‬‬ ‫‪o‬‬
‫=‪F0 (β, m, n) :‬‬ ‫‪.‬‬
‫}‪± (dn−1 dn−2 · · · d1 d0 d−1 · · · d−m )β dn−1 , . . . , d−m ∈ {0, 1, . . . , β − 1‬‬

‫ﻣﺜﺎﻝ ‪)۲−۱‬ﻣﺠﻤﻮﻋﻪ )‪(F0 (10, 1, 1‬‬

‫ﻣﺠﻤﻮﻋﻪ )‪ F0 (10, 1, 1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻋﻀﺎﻱ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻛﻪ ﺣﺪﺍﻛﺜﺮ ﻳﻚ ﺭﻗﻢ ﺻﺤﻴﺢ‬
‫ﻭ ﺣﺪﺍﻛﺜﺮ ﻳﻚ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺍﺭﻧﺪ‪ .‬ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‬
‫‪−9.9, −9.8, . . . , −0.1, 0.0, +0.1, . . . , +9.8, +9.9‬‬
‫ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﺗﻮﺯﻳﻊ ﺍﻋﺪﺍﺩ ﻓﻮﻕ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺍﻋﺪﺍﺩ ﻓﻮﻕ ﺭﻭﻱ ﻣﺤﻮﺭ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺷﻜﻞ ﺍﻋﺪﺍﺩﻱ ﻛﻪ‬
‫ﻗﺴﻤﺖ ﺍﻋﺸﺎﺭ ﺁﻥﻫﺎ ﺻﻔﺮ ﺍﺳﺖ )ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ( ﺑﺎ ﺧﻂﻫﺎﻱ ﺑﻠﻨﺪﺗﺮ ﻭ ﺿﺨﻴﻢﺗﺮ ﻣﺸﺨﺺ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫‪−xmax‬‬ ‫‪+xmax‬‬
‫)‪F0 (10, 1, 1‬‬
‫‪−9.9‬‬
‫‪−9‬‬
‫‪−8‬‬
‫‪−7‬‬
‫‪−6‬‬
‫‪−5‬‬
‫‪−4‬‬
‫‪−3‬‬
‫‪−2‬‬
‫‪−1‬‬

‫‪+1‬‬
‫‪+2‬‬
‫‪+3‬‬
‫‪+4‬‬
‫‪+5‬‬
‫‪+6‬‬
‫‪+7‬‬
‫‪+8‬‬
‫‪+9‬‬
‫‪+9.9‬‬
‫‪0‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺍﻋﺪﺍﺩ ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﺩﺭ ﺑﺎﺯﻩ ]‪ [−9.9, +9.9‬ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺑﻌﺪ‪ ،‬ﺗﺎﺛﻴﺮ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ m‬ﻭ ‪ n‬ﺩﺭ ﺗﻮﺯﻳﻊ ﺍﻋﺪﺍﺩ ﻣﺠﻤﻮﻋﻪ )‪ F0 (β, m, n‬ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۳−۱‬ﻣﺠﻤﻮﻋەﻫﺎﯼ )‪ F0 (2, 2, 3) ،F0 (2, 2, 2‬ﻭ )‪(F0 (2, 3, 3‬‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﺍﺯ ﻣﺒﻨﺎﻱ ‪ β = 2‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ‪،‬ﻣﺠﻤﻮﻋﻪﻫﺎﻱ )‪ F0 (2, 2, 3) ،F0 (2, 2, 2‬ﻭ )‪F0 (2, 3, 3‬‬
‫ﻣﺸﺨﺺ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪−xmax‬‬ ‫‪+xmax‬‬
‫)‪F0 (2, 2, 2‬‬
‫‪−3.75‬‬

‫‪+3.75‬‬
‫‪−3‬‬

‫‪−2‬‬

‫‪−1‬‬

‫‪+1‬‬

‫‪+2‬‬

‫‪+3‬‬

‫‪0‬‬

‫‪−xmax‬‬ ‫‪+xmax‬‬
‫)‪F0 (2, 2, 3‬‬
‫‪−7.75‬‬

‫‪+7.75‬‬
‫‪−7‬‬

‫‪−6‬‬

‫‪−5‬‬

‫‪−4‬‬

‫‪−3‬‬

‫‪−2‬‬

‫‪−1‬‬

‫‪+1‬‬

‫‪+2‬‬

‫‪+3‬‬

‫‪+4‬‬

‫‪+5‬‬

‫‪+6‬‬

‫‪+7‬‬

‫‪0‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰‬‬

‫‪−xmax‬‬ ‫‪+xmax‬‬
‫)‪F0 (2, 3, 3‬‬
‫‪−7.875‬‬

‫‪+7.875‬‬
‫‪−7‬‬

‫‪−6‬‬

‫‪−5‬‬

‫‪−4‬‬

‫‪−3‬‬

‫‪−2‬‬

‫‪−1‬‬

‫‪+1‬‬

‫‪+2‬‬

‫‪+3‬‬

‫‪+4‬‬

‫‪+5‬‬

‫‪+6‬‬

‫‪+7‬‬
‫‪0‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺍﻓﺰﺍﻳﺶ ‪ m‬ﺑﺎﻋﺚ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﻓﺎﺻﻠﻪ ﺍﻋﺪﺍﺩ ﻛﻤﺘﺮ ﺷﻮﺩ )ﺭﺯﻭﻟﻮﺷﻦ ﺑﺎﻻ ﻣﻲﺭﻭﺩ ﻭ ﺩﻗﺖ ﺑﻴﺸﺘﺮ‬
‫ﻣﻲﺷﻮﺩ(‪ ،‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺍﻓﺰﺍﻳﺶ ‪ n‬ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﺤﺪﻭﺩﻩ ﺍﻋﺪﺍﺩ ﮔﺴﺘﺮﺩﻩﺗﺮ ﺷﻮﺩ‪.‬‬

‫■ ﺧﻮﺍﺹ ﻣﺠﻤﻮﻋﻪ )‪F0 (β, m, n‬‬

‫‪ F0 (β, m, n) .۱‬ﻳﻚ ﺯﻳﺮﻣﺠﻤﻮﻋﻪ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ R‬ﻣﻲﺑﺎﺷﺪ ﻭ ﺗﻌﺪﺍﺩ ﺍﻋﻀﺎﻱ ﺁﻥ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪2 × βn × βm − 1‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻭﺟﻮﺩ ‪ −1‬ﺩﺭ ﻋﺒﺎﺭﺕ ﻓﻮﻕ‪ ،‬ﺑﻪ ﺍﻳﻦ ﺩﻟﻴﻞ ﺍﺳﺖ ﻛﻪ ﻋﺪﺩ ﺻﻔﺮ ﺩﺍﺭﺍﻱ ﺩﻭ ﻧﻤﺎﻳﺶ ‪ + 0 · · · 0.0 · · · 0‬ﻭ‬
‫‪ − 0 · · · 0.0 · · · 0‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪ .۲‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ )‪ F0 (β, m, n‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪ n‬ﺭﻗﻢ‬ ‫‪ m‬ﺭﻗﻢ‬
‫‪z‬‬ ‫|}‬ ‫‪{ z‬‬ ‫|}‬ ‫{‬
‫‪xmax‬‬ ‫‪.‬‬
‫)‪= (β − 1) · · · (β − 1) (β − 1) · · · (β − 1‬‬
‫‪X‬‬
‫‪n−1‬‬
‫=‬ ‫‪(β − 1) × β i‬‬
‫‪i=−m‬‬

‫‪= β n − β −m‬‬ ‫)ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻓﺮﻣﻮﻝ ﺗﺼﺎﻋﺪ ﻫﻨﺪﺳﻲ(‬

‫‪ .۳‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﺍﺯ ﻟﺤﺎﻅ ﻗﺪﺭﻣﻄﻠﻖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ )‪ F0 (β, m, n‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﻋﺒﺎﺭﺕ‬
‫ﺍﺳﺖ ﺍﺯ‬
‫‪ n‬ﺭﻗﻢ‬ ‫‪ m−1‬ﺭﻗﻢ‬
‫{ |} ‪z }| { z‬‬
‫‪xmin‬‬ ‫‪.‬‬
‫‪= 0 · · · 0 0 · · · 0 1 = β −m‬‬
‫‪ .۴‬ﺗﻮﺯﻳﻊ ﺍﻋﺪﺍﺩ ﻣﻮﺟﻮﺩ ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻳﻜﻨﻮﺍﺧﺖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺩﺭ ﺑﺎﺯﻩ ] ‪[−xmax , xmax‬‬
‫ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪ .‬ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﺩﻭ ﻋﺪﺩ ﻣﺘﻮﺍﻟﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪.β −m‬‬

‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‬ ‫�‬

‫ﻏﺎﻟﺐ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺑﺎ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ﺍﻋﺸﺎﺭﻱ ﺍﺯ ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺍﻟﮕﻮﺑﺮﺩﺍﺭﻱ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺑﻪ‬
‫ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﻳﻦ ﻣﺎﺷﻴﻦﻫﺎ‪ ،‬ﺣﺎﻓﻈﻪ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺮﺍﻱ ﻳﻚ ﻋﺪﺩ ﺍﻋﺸﺎﺭﻱ ﺭﺍ ﺑﺮﺍﻱ ﻧﮕﻪﺩﺍﺭﻱ ﻋﻼﻣﺖ‪ ،‬ﻣﺎﻧﺘﻴﺲ ﻭ ﻧﻤﺎﻱ‬
‫ﻋﺪﺩ ﻣﻮﺭﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻨﺪ‪ .‬ﺩﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‪ ،‬ﻣﺎﻧﺘﻴﺲ ﻣﻲﺗﻮﺍﻧﺪ ﺑﻲﻧﻬﺎﻳﺖ ﺭﻗﻢ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ ﻭ ﻧﻤﺎ ﻣﻲﺗﻮﺍﻧﺪ ﻫﺮ ﻋﺪﺩ ﺻﺤﻴﺤﻲ‬
‫ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﻪ ﻋﻠﺖ ﻣﺤﺪﻭﺩﻳﺖ ﺣﺎﻓﻈﻪ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺭﻭﻱ ﺗﻌﺪﺍﺩ ﺭﻗﻢﻫﺎﻱ ﻣﺎﻧﺘﻴﺲ ﻭ ﻫﻤﭽﻨﻴﻦ ﻛﺮﺍﻥ ﺑﺎﻻ ﻭ ﻛﺮﺍﻥ‬
‫ﭘﺎﻳﻴﻦ ﻧﻤﺎ ﻣﺤﺪﻭﺩﻳﺖ ﻗﺎﻳﻞ ﺷﻮﻳﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻳﻲ ﻛﻪ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺭﺍ ﺍﻟﮕﻮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻨﺪ‪ ،‬ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﺭﻗﻢ‬
‫)ﻣﺜﻼ ‪ m‬ﺭﻗﻢ( ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﻫﻤﭽﻨﻴﻦ ﺑﺮﺍﻱ ﻧﻤﺎ ﻛﺮﺍﻥ ﭘﺎﻳﻴﻦ )ﻣﺜﻼ ‪ (emin‬ﻭ ﻛﺮﺍﻥ ﺑﺎﻻﻱ )ﻣﺜﻼ ‪(emax‬‬
‫ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ‪ ،‬ﺷﻤﺎﻱ ﺣﺎﻓﻈﻪ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ‬
‫ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻣﺎﻧﺘﻴﺲ‬ ‫ﻧﻤﺎ‬

‫ﺑﻴﺖ ﻋﻼﻣﺖ‬
‫)‪ m‬ﺭﻗﻢ(‬ ‫)ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﺑﻴﻦ ‪ emin‬ﻭ ‪(emax‬‬
‫‪. .‬‬ ‫‪.‬‬ ‫‪...‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬

‫ﺩﺭ ﺍﻳﻦ ﺭﻭﻳﻜﺮﺩ‪ ،‬ﻗﺴﻤﺖ ﻧﻤﺎ ﺍﺯ ﺗﻌﺪﺍﺩﻱ ﺑﻴﺖ ﺗﺸﻜﻴﻞ ﺷﺪﻩ ﺍﺳﺖ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺑﺎ ﺍﻳﻦ ﺑﻴﺖﻫﺎ ﺑﺘﻮﺍﻥ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ﻣﺘﻮﺍﻟﻲ‬
‫‪emin , emin + 1, . . . , emax − 1, emax‬‬
‫ﺭﺍ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺳﻴﺴﺘﻢ ﻓﺮﻡ ﻛﻠﻲ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪.‬‬
‫‪± 0 d−1 · · · d−m × β e ,‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪di‬ﻫﺎ ﺍﺭﻗﺎﻡ ﻋﺪﺩ ﺩﺭ ﻣﺒﻨﺎﻱ ‪ β‬ﻣﻲﺑﺎﺷﻨﺪ ﻭ ‪ e‬ﻧﻤﺎﻱ ﻋﺪﺩ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﻋﺪﺩ ﺻﺤﻴﺢ ﺑﻴﻦ ‪ emin‬ﻭ ‪ emax‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ‬
‫ﺷﺮﻁ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ‪ d−1 ̸= 0‬ﺑﺮﺍﻱ ﺗﻀﻤﻴﻦ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﻟﺤﺎﻅ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ‬
‫ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ) ‪ FN (β, m, emin , emax‬ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ ﻭ ﺩﺍﺭﻳﻢ‪:‬‬
‫=‪FN (β, m, emin , emax ) :‬‬
‫‪n‬‬ ‫‪o‬‬
‫‪.‬‬
‫‪± 0 d−1 · · · d−m × β e di ∈ {0, 1, . . . , β−1}, d−1 ̸= 0, e ∈ {emin , emin + 1, . . . , emax } .‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦﻫﺎ ﻫﻤﻮﺍﺭﻩ ﺷﺮﻁ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺍﻋﻤﺎﻝ ﻣﻲﮔﺮﺩﺩ ﻭ ﻟﺬﺍ ﻋﺪﺩ ﺻﻔﺮ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦﻫﺎ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۴−۱‬ﻣﺠﻤﻮﻋﻪ )‪(FN (2, 3, −1, 2‬‬

‫ﻣﺠﻤﻮﻋﻪ )‪ FN (2, 3, −1, 2‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺩﺍﺭﺍﻱ ‪ ۳۲‬ﻋﻀﻮ ﺍﺳﺖ‪ ،‬ﻛﻪ ﺍﻋﻀﺎﻱ ﻣﺜﺒﺖ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻋﺒﺎﺭﺕ‬
‫ﻫﺴﺘﻨﺪ ﺍﺯ‪:‬‬
‫‪.‬‬
‫‪0 100×2−1 = 0 25 0 101×2−1 = 0 3125‬‬‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪0 110×2−1 = 0 375 0 111×2−1 = 0 4375‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪0.100×2‬‬ ‫‪0‬‬
‫‪= 0.5‬‬ ‫‪0.101×2‬‬ ‫‪0‬‬
‫‪= 0.625‬‬ ‫‪0.110×2‬‬ ‫‪0‬‬
‫‪= 0.75‬‬ ‫‪0.111×2‬‬ ‫‪0‬‬
‫‪= 0.875‬‬
‫‪0.100×2‬‬ ‫‪+1‬‬
‫‪= 1.0‬‬ ‫‪0.101×2‬‬ ‫‪+1‬‬
‫‪= 1.25‬‬ ‫‪0.110×2‬‬ ‫‪+1‬‬
‫‪= 1.5‬‬ ‫‪0.111×2‬‬ ‫‪+1‬‬
‫‪= 1.75‬‬
‫‪0.100×2‬‬ ‫‪+2‬‬
‫‪= 2.0‬‬ ‫‪0.101×2‬‬ ‫‪+2‬‬
‫‪= 2.5‬‬ ‫‪0.110×2‬‬ ‫‪+2‬‬
‫‪= 3.0‬‬ ‫‪0.111×2‬‬ ‫‪+2‬‬
‫‪= 3.5‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺎﻧﺘﻴﺲﻫﺎ ﺍﻋﺪﺍﺩﻱ ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻭ ﻫﺴﺘﻨﺪ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺍﻋﺪﺍﺩ ﻓﻮﻕ )ﻛﻪ ﺍﻋﺪﺍﺩ ﻣﺜﺒﺖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ )‪FN (2, 3, −1, 2‬‬
‫ﻣﻲﺑﺎﺷﻨﺪ( ﺭﻭﻱ ﻣﺤﻮﺭ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪xmin‬‬ ‫‪xmax‬‬
‫‪0.100 × 2−1 = 0.25‬‬
‫‪0.101 × 2−1 = 0.3125‬‬
‫‪0.110 × 2−1 = 0.375‬‬
‫‪0.111 × 2−1 = 0.4375‬‬
‫‪0.100 × 20 = 0.5‬‬
‫‪0.101 × 20 = 0.625‬‬
‫‪0.110 × 20 = 0.75‬‬
‫‪0.111 × 20 = 0.875‬‬
‫‪0.100 × 2+1 = 1.0‬‬

‫‪0.101 × 2+1 = 1.25‬‬

‫‪0.110 × 2+1 = 1.5‬‬

‫‪0.111 × 2+1 = 1.75‬‬

‫‪0.100 × 2+2 = 2.0‬‬

‫‪0.101 × 2+2 = 2.5‬‬

‫‪0.110 × 2+2 = 3.0‬‬

‫‪0.111 × 2+2 = 3.5‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺑﻪ ﭼﻬﺎﺭ ﺩﺳﺘﻪ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‬
‫ﻛﻪ ﺍﻋﻀﺎﻱ ﻫﺮ ﺩﺳﺘﻪ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪﺍﻧﺪ‪ .‬ﺗﻮﺟﻪ ﮔﺮﺩﺩ ﻛﻪ ﭼﻮﻥ ﭼﻬﺎﺭ ﺣﺎﻟﺖ ﻣﺘﻔﺎﻭﺕ ﺑﺮﺍﻱ ﻧﻤﺎ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻟﺬﺍ ﻣﺠﻤﻮﻋﻪ ﺑﻪ ﭼﻬﺎﺭ‬
‫ﺯﻳﺮ ﻣﺠﻤﻮﻋﻪ ﺍﺯ ﺍﻋﺪﺍﺩ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲‬‬

‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﻗﺮﻳﻨﻪ ﺍﻋﺪﺍﺩ ﻓﻮﻕ ﻧﻴﺰ ﺩﺭ ﻣﺠﻤﻮﻋﻪ )‪ FN (2, 3, −1, 2‬ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺗﻤﺎﻡ ﺍﻋﻀﺎﻱ‬
‫)‪ FN (2, 3, −1, 2‬ﺭﻭﻱ ﻣﺤﻮﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪−xmax‬‬ ‫‪−xmin xmin‬‬ ‫‪xmax‬‬
‫‪− 3.5‬‬

‫‪+ 3.5‬‬
‫‪− 2.0‬‬

‫‪− 1.0‬‬

‫‪− 0.5‬‬
‫‪− 0.25‬‬

‫‪+ 0.25‬‬
‫‪+ 0.5‬‬

‫‪+ 1.0‬‬

‫‪+ 2.0‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﻀﻮ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ‪ xmax = 3.5‬ﻭ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﻨﺼﺮ ﻣﺜﺒﺖ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ‬
‫‪ xmin = 0.25‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻋﺪﺩ ﺻﻔﺮ ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻧﻴﺴﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ﻳﻚ ﻓﺎﺻﻠﻪ ﺧﺎﻟﻲ ﻧﺴﺒﺘ ًﺎ ﺑﺰﺭﮒ ﺩﺭ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﭘﻮﺷﺶ‬
‫ﺩﺍﺩﻩ ﻧﻤﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﻋﻠﺖ ﺁﻥ ﺍﻧﺠﺎﻡ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫■ ﺧﻮﺍﺹ ﻣﺠﻤﻮﻋﻪ ) ‪FN (β, m, emin , emax‬‬

‫‪ FN (β, m, emin , emax ) .۱‬ﻳﻚ ﺯﻳﺮﻣﺠﻤﻮﻋﻪ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ R‬ﻣﻲﺑﺎﺷﺪ ﻭ ﺗﻌﺪﺍﺩ ﺍﻋﻀﺎﻱ ﺁﻥ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫)‪2 × (β − 1) × β m−1 × (emax − emin + 1‬‬
‫‪ .۲‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ) ‪ FN (β, m, emin , emax‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪ m‬ﺭﻗﻢ‬
‫‪z‬‬ ‫|}‬ ‫{‬
‫‪xmax‬‬ ‫‪.‬‬
‫‪= 0 (β − 1) · · · (β − 1) ×β emax‬‬
‫‪X‬‬
‫‪m‬‬
‫‪= (β − 1)β emax‬‬ ‫‪β −i‬‬
‫‪i=1‬‬
‫‪−m‬‬
‫‪=β‬‬ ‫‪emax‬‬
‫‪(1 − β‬‬ ‫)‬
‫‪ .۳‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ) ‪ FN (β, m, emin , emax‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪m−1‬‬
‫{ |} ‪z‬‬
‫‪.‬‬
‫‪xmin = 0 1 0 · · · 0 ×β emin = β emin −1‬‬
‫‪ .۴‬ﺗﻮﺯﻳﻊ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﻳﻜﻨﻮﺍﺧﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﺍﻋﺪﺍﺩ ﻣﺘﻔﺎﻭﺕ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺍﮔﺮ ) ‪ x ∈ FN (β, m, emin , emax‬ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ‪ 0.d−1 · · · d−m × β e‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪ ﺍﺯ ﺁﻥ ﺩﺭ‬
‫ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‬ ‫ { |} ‪z‬‬
‫‪m−1‬‬

‫‪.‬‬
‫‪x+ = 0 d−1 · · · d−m + 0 0 · · · 0 1 × β e .‬‬‫‪.‬‬
‫ﻟﺬﺍ ﻓﺎﺻﻠﻪ ﺑﻴﻦ ‪ x‬ﻭ ‪ x+‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪m−1‬‬
‫{ |} ‪z‬‬
‫‪.‬‬
‫‪0 0 · · · 0 1 × β e = β e−m‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻓﺎﺻﻠﻪ ﻋﺪﺩ ‪ x‬ﺑﺎ ﻋﺪﺩ ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪﻱ ﻋﻼﻭﻩ ﺑﺮ ﺍﻧﺪﺍﺯﻩ ‪ m‬ﺑﻪ ﻣﻘﺪﺍﺭ ‪ e‬ﻳﻌﻨﻲ ﻧﻤﺎﻱ ﻋﺪﺩ ‪ x‬ﻧﻴﺰ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪.‬‬
‫ﻫﺮﭼﻪ ﺍﻧﺪﺍﺯﻩ ﻋﺪﺩ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ‪ e‬ﺑﺰﺭﮒﺗﺮ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻓﺎﺻﻠﻪ ﺗﺎ ﻋﺪﺩ ﺑﻌﺪﻱ ﺑﻴﺸﺘﺮ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﺳﻪ ﻣﺜﺎﻝ ﺑﻌﺪ‪ ،‬ﺗﺎﺛﻴﺮ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ emin ،m‬ﻭ ‪ emax‬ﺑﺮ ﺗﻮﺯﻳﻊ ﺍﻋﺪﺍﺩ ﻣﺠﻤﻮﻋﻪ ) ‪ FN (β, m, emin , emax‬ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻣﺜﺎﻝ ‪)۵−۱‬ﺗﺎﺛﯿﺮ ‪ m‬ﺑﺮ ﺗﻮﺯﯾﻊ ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ ﻣﻤﯿﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ(‬

‫ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ‪ ،‬ﺍﻋﻀﺎﻱ ﻣﺜﺒﺖ ﻣﺠﻤﻮﻋﻪ )‪ FN (2, m, −1, 2‬ﺑﻪﺍﺯﺍﻱ ‪ m = 2, 3, 4, 5‬ﺭﻭﻱ ﻣﺤﻮﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪xmin‬‬
‫)‪FN (2, 2, −1, 2‬‬ ‫‪xmax‬‬

‫‪0.25‬‬ ‫‪0.5‬‬ ‫‪1.0‬‬ ‫‪2.0‬‬ ‫‪3.0‬‬

‫‪xmin‬‬
‫)‪FN (2, 3, −1, 2‬‬ ‫‪xmax‬‬

‫‪0.25‬‬ ‫‪0.5‬‬ ‫‪1.0‬‬ ‫‪2.0‬‬ ‫‪3.5‬‬

‫‪xmin‬‬
‫)‪FN (2, 4, −1, 2‬‬ ‫‪xmax‬‬

‫‪0.25‬‬ ‫‪0.5‬‬ ‫‪1.0‬‬ ‫‪2.0‬‬ ‫‪3.75‬‬

‫‪xmin‬‬
‫)‪FN (2, 5, −1, 2‬‬ ‫‪xmax‬‬

‫‪0.25‬‬ ‫‪0.5‬‬ ‫‪1.0‬‬ ‫‪2.0‬‬ ‫‪3.875‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺗﻐﻴﻴﺮ ﭘﺎﺭﺍﻣﺘﺮ ‪ m‬ﺩﺭ ﻣﻘﺪﺍﺭ ‪ xmin‬ﺗﺎﺛﻴﺮﻱ ﻧﺪﺍﺭﺩ ﻭ ﺗﺎﺛﻴﺮ ﺍﻧﺪﻛﻲ ﺭﻭﻱ ‪ xmax‬ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪،‬‬
‫ﺍﻓﺰﺍﻳﺶ ﻣﻘﺪﺍﺭ ‪ m‬ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﺮﺍﻛﻢ ﺍﻋﺪﺍﺩ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [xmin , xmax‬ﺍﻓﺰﺍﻳﺶ ﻳﺎﺑﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۶−۱‬ﺗﺎﺛﯿﺮ ‪ emin‬ﺑﺮ ﺗﻮﺯﯾﻊ ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ ﻣﻤﯿﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ(‬

‫ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ‪ ،‬ﺍﻋﻀﺎﻱ ﻣﺜﺒﺖ ﻣﺠﻤﻮﻋﻪ )‪ FN (2, 3, emin , 1‬ﺑﻪﺍﺯﺍﻱ ‪ emin = −1, −2, −3, −4‬ﺭﻭﻱ ﻣﺤﻮﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬
‫ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪xmin‬‬ ‫)‪FN (2, 3, −1, 1‬‬ ‫‪xmax‬‬
‫‪1.75‬‬
‫‪0.25‬‬

‫‪0.5‬‬

‫‪1.0‬‬

‫‪xmin‬‬ ‫)‪FN (2, 3, −2, 1‬‬ ‫‪xmax‬‬

‫‪1.75‬‬
‫‪0.125‬‬

‫‪0.25‬‬

‫‪0.5‬‬

‫‪1.0‬‬

‫‪xmin‬‬ ‫)‪FN (2, 3, −3, 1‬‬ ‫‪xmax‬‬

‫‪1.75‬‬
‫‪0.0625‬‬
‫‪0.125‬‬

‫‪0.25‬‬

‫‪0.5‬‬

‫‪1.0‬‬

‫‪xmin‬‬ ‫)‪FN (2, 3, −4, 1‬‬ ‫‪xmax‬‬

‫‪1.75‬‬
‫‪0.03125‬‬
‫‪0.0625‬‬
‫‪0.125‬‬

‫‪0.25‬‬

‫‪0.5‬‬

‫‪1.0‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﺸﺎﻫﺪﻩ ﻣﻲﺷﻮﺩ ﺑﺎ ﻛﺎﻫﺶ ﻣﻘﺪﺍﺭ ‪ emin‬ﺩﺳﺘﻪ ﺍﻋﺪﺍﺩﻱ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺑﻪ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻗﺒﻠﻲ ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺍﻳﻦ ﺩﺳﺘﻪ ﺍﻋﺪﺍﺩ ﺩﺭ ﻧﺰﺩﻳﻜﻲ ﺻﻔﺮ ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ ﻭ ﺑﺎﻋﺚ ﻛﻮﭼﻚﺗﺮ ﺷﺪﻥ ﻓﻀﺎﻱ ﺧﺎﻟﻲ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۷−۱‬ﺗﺎﺛﯿﺮ ‪ emax‬ﺑﺮ ﺗﻮﺯﯾﻊ ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ ﻣﻤﯿﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ(‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﺠﻤﻮﻋﻪ ) ‪ FN (2, 3, 1, emax‬ﺑﻪﺍﺯﺍﻱ ‪ emax = 1, 2, 3, 4‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ ﺍﻋﻀﺎﻱ ﻣﺜﺒﺖ‬
‫ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪﻫﺎ ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪xmin‬‬ ‫‪xmax‬‬ ‫)‪FN (2, 3, −1, 1‬‬

‫‪1.75‬‬
‫‪0.25‬‬
‫‪0.5‬‬

‫‪1.0‬‬

‫‪xmin‬‬ ‫‪xmax‬‬ ‫)‪FN (2, 3, −1, 2‬‬

‫‪3.5‬‬
‫‪0.25‬‬
‫‪0.5‬‬

‫‪1.0‬‬

‫‪2.0‬‬

‫‪xmin‬‬ ‫‪xmax‬‬ ‫)‪FN (2, 3, −1, 3‬‬

‫‪7.0‬‬
‫‪0.25‬‬
‫‪0.5‬‬

‫‪1.0‬‬

‫‪2.0‬‬

‫‪4.0‬‬

‫‪xmin‬‬ ‫)‪FN (2, 3, −1, 4‬‬ ‫‪xmax‬‬

‫‪14.0‬‬
‫‪0.25‬‬
‫‪0.5‬‬

‫‪1.0‬‬

‫‪2.0‬‬

‫‪4.0‬‬

‫‪8.0‬‬

‫ﻣﺸﺎﻫﺪﻩ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎ ﺍﻓﺰﺍﻳﺶ ﻣﻘﺪﺍﺭ ‪ ،emax‬ﺩﺳﺘﻪ ﺍﻋﺪﺍﺩﻱ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺑﻪ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻗﺒﻠﻲ ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻃﻮﺭﻱ‬
‫ﻛﻪ ﺍﻧﺪﺍﺯﻩ ﺍﻋﺪﺍﺩ ﺍﺿﺎﻓﻪ ﺷﺪﻩ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﺍﻧﺪﺍﺯﻩ ﺍﻋﺪﺍﺩ ﻗﺒﻠﻲ ﺍﺳﺖ ﻭ ﺩﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﺍﻋﺪﺍﺩ ﻣﺜﺒﺖ )ﻭ ﺩﺭ ﺳﻤﺖ ﭼﭗ ﺍﻋﺪﺍﺩ ﻣﻨﻔﻲ(‬
‫ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪.‬‬

‫ﻳﻚ ﻣﺰﻳﺖ ﺳﻴﺴﺘﻢ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﻧﺴﺒﺖ ﺑﻪ ﺳﻴﺴﺘﻢ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻫﻢﺯﻣﺎﻥ ﺍﻋﺪﺍﺩ ﺑﺰﺭﮒ‬
‫ﻭ ﻛﻮﭼﻚ ﺭﺍ ﺩﺍﺧﻞ ﺧﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺍﺯ ﻣﻌﺎﻳﺐ ﺁﻥ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﭘﻮﺷﺶ ﻧﺪﺍﺩﻥ ﻋﺪﺩ ﺻﻔﺮ ﻭ ﻫﻤﻴﻦﻃﻮﺭ ﭘﻮﺷﺶ ﻧﺪﺍﺩﻥ ﻳﻚ ﻧﺎﺣﻴﻪ ﺍﻃﺮﺍﻑ‬
‫ﺻﻔﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺑﺨﺶ ﺑﻌﺪ‪ ،‬ﺍﻳﻦ ﺍﺷﻜﺎﻻﺕ ﺑﺎ ﺍﺿﺎﻓﻪ ﻛﺮﺩﻥ ﺍﻋﺪﺍﺩ ﻏﻴﺮﻧﺮﻣﺎﻟﻲ ﻣﻮﺳﻮﻡ ﺑﻪ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﺑﺮﻃﺮﻑ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮ ﻣﺒﻨﺎﻱ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﺷﺪ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪ ،‬ﻟﺤﺎﻅ ﻛﺮﺩﻥ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻣﻨﺠﺮ ﺑﻪ ﻣﻨﺤﺼﺮﺑﻪ‬
‫ﻓﺮﺩﻱ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﮔﺮﺩﻳﺪ‪ .‬ﺍﻣﺎ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻋﺪﺩ ﺻﻔﺮ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﻧﺒﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﻳﻚ ﻧﺎﺣﻴﻪ ﻧﺴﺒﺘﺎ‬
‫ﺑﺰﺭﮒ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﭘﻮﺷﺶ ﺩﺍﺩﻩ ﻧﺸﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺭﻓﻊ ﺍﻳﻦ ﻣﺸﻜﻞ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺗﻐﻴﻴﺮ ﻛﻮﭼﻜﻲ ﺩﺭ ﻗﺎﻋﺪﻩ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪،‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺗﻌﺪﺍﺩﻱ ﻋﺪﺩ ﻏﻴﺮﻧﺮﻣﺎﻝ‪ ،‬ﻣﻮﺳﻮﻡ ﺑﻪ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ‪ ،۱۷‬ﺑﺮﺍﻱ ﭘﻮﺷﺶ ﺩﺍﺩﻥ ﺻﻔﺮ ﻭ ﻓﻀﺎﻱ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﺑﻪ‬
‫) ‪ FN (β, m, emin , emax‬ﺍﺿﺎﻓﻪ ﺷﻮﺩ ﻭ ﺩﺭ ﺿﻤﻦ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﻧﻤﺎﻳﺶ ﻧﻴﺰ ﺣﻔﻆ ﺷﻮﺩ‪ .‬ﻣﺎﺷﻴﻦﻫﺎﻳﻲ ﻛﻪ ﺍﺯ ﺍﻳﻦ ﺭﻭﻳﻜﺮﺩ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻨﺪ ﺑﻪ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬
‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‪ ،‬ﻫﻤﺎﻧﻨﺪ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪،‬‬
‫ﺑﻪ ﺗﻌﺪﺍﺩ ‪ m‬ﺭﻗﻢ ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﻧﻤﺎﻱ ﻫﺮ ﻋﺪﺩ ﺩﺍﺭﺍﻱ ﻛﺮﺍﻥ ﺑﺎﻻ ﻭ ﭘﺎﺋﻴﻦ ‪ emax‬ﻭ ‪ emin‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ‬
‫‪17 Subnormal‬‬ ‫)‪numbers(Denormal numbers‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻣﺎﻧﺘﻴﺲ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﻧﻤﺎﻱ ﺁﻥﻫﺎ ﺑﺮﺍﺑﺮ ‪ emin‬ﺍﺳﺖ ﻣﻲﺗﻮﺍﻧﺪ ﻏﻴﺮﻧﺮﻣﺎﻝ ﺑﺎﺷﺪ ﻭﻟﻲ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﻧﻤﺎﻱ ﺁﻥﻫﺎ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ‪ emin‬ﺍﺳﺖ‬
‫ﺑﺎﻳﺪ ﻧﺮﻣﺎﻝ ﺑﺎﺷﻨﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻋﺪﻡ ﺍﻋﻤﺎﻝ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﺑﺎ ﻣﺎﻧﺘﻴﺲ ‪ emin‬ﺑﺎﻋﺚ ﺑﻪ ﻫﻢ ﺧﻮﺭﺩﻥ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ‬
‫ﻧﻤﺎﻳﺶ ﻧﻤﻲﮔﺮﺩﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺗﺸﺮﻳﺢ ﻧﻜﺘﻪ ﻓﻮﻕ‪ ،‬ﻋﺪﺩ ‪ x = 0.0d−2 · · · d−m × β‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﻋﺪﺩ ‪ x‬ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﺩﺍﺭﺍﻱ‬
‫‪emin‬‬

‫ﻣﺎﻧﺘﻴﺲ ﺣﺪﺍﻛﺜﺮ ‪ m‬ﺭﻗﻤﻲ ﺍﺳﺖ ﻭ ﻧﺮﻣﺎﻝ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﻤﺎﻱ ‪ x‬ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ ﺑﺮﺍﺑﺮ ‪ emin‬ﺍﺳﺖ‪ .‬ﻋﺪﺩ ‪ x‬ﺩﺭ ﻣﺠﻤﻮﻋﻪ‬
‫) ‪ FN (β, m, emin , emax‬ﻗﺮﺍﺭ ﻧﺪﺍﺭﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﻛﻠﻴﻪ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺸﻲ ﺑﺎ ﻣﺎﻧﺘﻴﺲ ﺣﺪﺍﻛﺜﺮ ‪ m‬ﺭﻗﻤﻲ ﻭ ﻧﻤﺎﻱ ‪ emin‬ﺑﺎﺷﻨﺪ‬
‫ﻭ ﻫﻤﭽﻨﻴﻦ ﻏﻴﺮﻧﺮﻣﺎﻝ ﺑﺎﺷﻨﺪ ﺩﺭ ) ‪ FN (β, m, emin , emax‬ﻗﺮﺍﺭ ﻧﺪﺍﺭﺩ‪ .‬ﻟﺬﺍ ﺍﺿﺎﻓﻪ ﻛﺮﺩﻥ ﭼﻨﻴﻦ ﺍﻋﺪﺍﺩﻱ ﺑﻪ ) ‪FN (β, m, emin , emax‬‬
‫ﺑﺎﻋﺚ ﺑﺮﻫﻢ ﺧﻮﺭﺩﻥ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﻧﻤﺎﻳﺶ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻋﺪﻡ ﺍﻧﺠﺎﻡ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﻧﻤﺎﻳﺸﻲ ﺑﻪ‬
‫ﻓﺮﻡ ‪ 0.0d−2 · · · d−m × β emin‬ﺩﺍﺭﻧﺪ‪ ،‬ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻭﻻ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﻧﻤﺎﻳﺶ ﺑﻪ ﻫﻢ ﻧﺨﻮﺭﺩ‪ ،‬ﺛﺎﻧﻴﺎ ﺍﻋﺪﺍﺩﻱ ﺟﺪﻳﺪ‬
‫ﺑﻪ ﻣﺠﻤﻮﻋﻪ ) ‪ FN (β, m, emin , emax‬ﺍﺿﺎﻓﻪ ﺷﻮﺩ‪ ،‬ﻛﻪ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺟﺪﻳﺪ ﻧﺎﺣﻴﻪ ﺧﺎﻟﻲ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﺭﺍ ﭘﻮﺷﺶ ﻣﻲﺩﻫﻨﺪ‪.‬‬
‫ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‪ ،‬ﻫﻤﻪ ﭼﻴﺰ ﺷﺒﻴﻪ ﺑﻪ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ ﺍﺳﺖ ﻭ ﺗﻨﻬﺎ ﻗﺎﻋﺪﻩ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ ﻋﻮﺽ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ‪ ،‬ﻭﻗﺘﻲ ﻧﻤﺎ ﺑﺮﺍﺑﺮ ‪ emin‬ﺑﺎﺷﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﺮﻣﺎﻝﺳﺎﺯﻱ‬
‫ﺭﺍ ﺍﻧﺠﺎﻡ ﻧﺪﺍﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦﻫﺎ ﻋﻼﻭﻩ ﺑﺮ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ) ‪ FN (β, m, emin , emax‬ﺍﻋﺪﺍﺩ ﻏﻴﺮﻧﺮﻣﺎﻟﻲ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻧﻴﺰ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﺧﻮﺍﻫﻨﺪ ﺑﻮﺩ‪:‬‬
‫‪.‬‬
‫‪± 0 0d−2 · · · d−m × β emin ,‬‬ ‫}‪di ∈ {0, 1, . . . , β − 1‬‬
‫ﻛﻪ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﻏﻴﺮﻧﺮﻣﺎﻝ ﻫﻤﺎﻥ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ) ‪ FS (β, m, emin , emax‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻭ ﺩﺍﺭﻳﻢ‪:‬‬
‫‬
‫‪.‬‬
‫}‪FS (β, m, emin , emax ) := FN (β, m, emin , emax ) ∪ ± 0 0d−2 · · · d−m × β emin di ∈ {0, 1, . . . , β − 1‬‬
‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﺍﺯ ﻧﻈﺮ ﺍﻧﺪﺍﺯﻩ ﺍﺯ ﺗﻤﺎﻣﻲ ﺍﻋﺪﺍﺩ ﻧﺮﻣﺎﻝ )ﻭ ﻟﺬﺍ ﺍﺯ ‪ (x min‬ﻛﻮﭼﻚﺗﺮ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﺑﻌﺪ‪،‬‬
‫ﺍﻳﻦ ﻣﻮﺿﻮﻉ ﻭ ﻣﻮﻗﻌﻴﺖ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۸−۱‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﻤﯿﺰ ﺷﻨﺎﻭﺭ ﻧﯿﻤەﻧﺮﻣﺎﻝ )‪(FS (2, 3, −1, 1‬‬
‫ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ‪ ،‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ )‪ FN (2, 3, −1, 1‬ﻭ )‪ FS (2, 3, −1, 1‬ﺭﻭﻱ ﻣﺤﻮﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﺸﺨﺺ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫‪−xmax‬‬ ‫‪−xmin‬‬ ‫‪xmin‬‬ ‫‪xmax‬‬

‫)‪FN (2, 3, −1, 1‬‬
‫‪− 1.75‬‬

‫‪+ 1.75‬‬
‫‪− 1.0‬‬

‫‪− 0.5‬‬

‫‪− 0.25‬‬

‫‪+ 0.25‬‬

‫‪+ 0.5‬‬

‫‪+ 1.0‬‬

‫‪−xmax‬‬ ‫‪−xmin‬‬ ‫ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‬ ‫‪xmin‬‬ ‫‪xmax‬‬

‫)‪FS (2, 3, −1, 1‬‬
‫‪− 0.011 × 2−1‬‬
‫‪− 0.010 × 2−1‬‬
‫‪− 0.001 × 2−1‬‬

‫‪+ 0.001 × 2−1‬‬

‫‪+ 0.010 × 2−1‬‬
‫‪+ 0.011 × 2−1‬‬
‫‪− 1.75‬‬

‫‪+ 1.75‬‬
‫‪− 1.0‬‬

‫‪− 0.5‬‬

‫‪− 0.25‬‬

‫‪+ 0.25‬‬

‫‪+ 0.5‬‬

‫‪+ 1.0‬‬
‫‪0.000 × 2 e‬‬
‫‪= −0.1875‬‬
‫‪= −0.125‬‬
‫‪= −0.0625‬‬

‫‪= +0.0625‬‬
‫‪= +0.125‬‬
‫‪= +0.1875‬‬
‫‪= 0.0‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺗﻨﻬﺎ ﺗﻔﺎﻭﺕ ﺍﻳﻦ ﺩﻭ ﻣﺠﻤﻮﻋﻪ ﺩﺭ ﺍﻋﺪﺍﺩ ﻏﻴﺮﻧﺮﻣﺎﻟﻲ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺑﺎﺯﻩ ) ‪ (−xmin , xmin‬ﺑﻪ‬
‫ﺻﻮﺭﺕ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﭘﺨﺶ ﺷﺪﻩﺍﻧﺪ ﻭ ﻧﺎﺣﻴﻪ ﺧﺎﻟﻲ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﺭﺍ ﭘﺮ ﻛﺮﺩﻩﺍﻧﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻋﺪﺩ ﺻﻔﺮ ﻧﻴﺰ ﺗﻮﺳﻂ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻋﺪﺩﻱ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﭘﻮﺷﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺧﻮﺍﺹ ﻣﺠﻤﻮﻋﻪ ) ‪FS (β, m, emin , emax‬‬

‫‪ FS (β, m, emin , emax ) .۱‬ﻳﻚ ﺯﻳﺮﻣﺠﻤﻮﻋﻪ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ R‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﻋﻀﺎﻱ ﻧﺮﻣﺎﻝ ﺁﻥ ﺑﺮﺍﺑﺮ‬
‫)‪2 × (β − 1) × β m−1 × (emax − emin + 1‬‬
‫ﻭ ﺗﻌﺪﺍﺩ ﺍﻋﻀﺎﻱ ﻏﻴﺮﻧﺮﻣﺎﻝ ﺁﻥ ﺑﺮﺍﺑﺮ‬
‫‪2 × β m−1 − 1‬‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺗﻌﺪﺍﺩ ﻛﻞ ﺍﻋﻀﺎﻱ ) ‪ FS (β, m, emin , emax‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪2 × (β − 1) × β m−1 × (emax − emin + 1) + 2 × β m−1 − 1‬‬

‫‪ .۲‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ) ‪ FS (β, m, emin , emax‬ﻫﻤﺎﻥ ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﺩﺭ ) ‪ FN (β, m, emin , emax‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫‪ .۳‬ﺻﻔﺮ ﻋﻀﻮﻱ ﺍﺯ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺍﺳﺖ ﻭ ﺩﻭ ﻧﻤﺎﻳﺶ )ﻳﻜﻲ ﺑﺎ ﻋﻼﻣﺖ ﻣﺜﺒﺖ ﻭ ﻳﻜﻲ ﺑﺎ ﻋﻼﻣﺖ ﻣﻨﻔﻲ( ﺑﺮﺍﻱ ﺻﻔﺮ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪:‬‬
‫‪m‬‬ ‫‪m‬‬
‫{ |} ‪z‬‬ ‫{ |} ‪z‬‬
‫‪.‬‬
‫‪+ 0 0 · · · 0 ×β emin ,‬‬ ‫‪.‬‬
‫‪− 0 0 · · · 0 ×β emin .‬‬

‫‪ .۴‬ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ‪ x min = β emin −1‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺑﻠﻜﻪ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ‬
‫ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪m−1‬‬
‫{ |} ‪z‬‬
‫‪.‬‬
‫‪xsmin = 0 0 · · · 0 1 × β emin = β emin −m‬‬
‫ﺩﺭ ﻭﺍﻗﻊ ‪ xsmin‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﻭ ‪ x min‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﻧﺮﻣﺎﻝ ﺩﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺍﺳﺖ‪.‬‬

‫‪ .۵‬ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﺩﺭ ﺑﺎﺯﻩ ) ‪ (−xmin , xmin‬ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ ﻭ ﺩﺍﺭﺍﻱ ﺗﻮﺯﻳﻊ ﻳﻜﻨﻮﺍﺧﺖ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬

‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ‬ ‫‪۳.۱‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺯ ﻳﻚ ﺣﺎﻓﻈﻪ ﻣﺘﻨﺎﻫﻲ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻨﺪ‪ ،‬ﻟﺬﺍ ﻗﺎﺩﺭ ﻧﻴﺴﺘﻨﺪ ﻛﻪ ﺗﻤﺎﻣﻲ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‬
‫ﺭﺍ ﺩﺭ ﺧﻮﺩ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺩﻫﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﺎﺷﻴﻦﻫﺎ ﺗﻨﻬﺎ ﻣﻲﺗﻮﺍﻧﻨﺪ ﺯﻳﺮﻣﺠﻤﻮﻋﻪ ﻛﻮﭼﻜﻲ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺭﺍ ﺫﺧﻴﺮﻩ ﻭ‬
‫ﻧﻤﺎﻳﺶ ﺩﻫﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺭﺍ ﺑﻪ ﺩﻭ ﺩﺳﺘﻪ ﺯﻳﺮ ﺗﻘﺴﻴﻢ ﻣﻲﻛﻨﻴﻢ‪:‬‬

‫• ﺍﻋﺪﺍﺩ ﻗﺎﺑﻞ ﻧﻤﺎﻳﺶ ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ)ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ‪(۱۸‬‬

‫• ﺍﻋﺪﺍﺩ ﻏﻴﺮﻗﺎﺑﻞ ﻧﻤﺎﻳﺶ ﺩﺭ ﻣﺎﺷﻴﻦ)ﺍﻋﺪﺍﺩ ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ‪(۱۹‬‬

‫ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪18 Machine‬‬ ‫‪Numbers‬‬ ‫‪19 Non-Machine‬‬ ‫‪Numbers‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ ﻭ ﻏﯿﺮﻣﺎﺷﯿﻨ‬ ‫ﺗﻌﺮﯾﻒ ‪۱−۱‬‬

‫ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﻛﻪ ﺍﺯ ﺣﺎﻓﻈﻪ ﻣﺘﻨﺎﻫﻲ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪ ،‬ﻓﻘﻂ ﻣﻲﺗﻮﺍﻥ ﺯﻳﺮﻣﺠﻤﻮﻋﻪﺍﻱ ﺍﺯ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺭﺍ‬
‫ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺩﺍﺩ‪ .‬ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﺩﺭ ﻣﺎﺷﻴﻦ ﺭﺍ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻣﻲﮔﻮﻳﻴﻢ ﻭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺩﻳﮕﺮ ﺭﺍ ﺍﻋﺪﺍﺩ‬
‫ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻪ ﻧﻮﻉ ﻣﺎﺷﻴﻦ ﻭ ﺳﻴﺴﺘﻤﻲ ﻛﻪ ﺩﺭ ﺁﻥ ﻣﺎﺷﻴﻦ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﺩ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪ .‬ﺑﻪ‬
‫ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻤﺎﻳﺶ ﻧﻘﻄﻪ ﺛﺎﺑﺖ )ﺑﺎ ﻣﺒﻨﺎﻱ ‪ n ،β‬ﺭﻗﻢ ﺻﺤﻴﺢ ﻭ ‪ m‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ( ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻋﺒﺎﺭﺗﻨﺪ‬
‫ﺍﺯ )‪ .F0 (β, n, m‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‪ ،‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫) ‪.FS (β, m, emin , emax‬‬
‫ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ‪ x ∈ R‬ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ‪ x‬ﺭﺍ ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ .‬ﺳﻪ ﺣﺎﻟﺖ ﺭﺥ ﻣﻲ ﺩﻫﺪ‪:‬‬

‫‪ x .۱‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ‪،‬‬

‫‪ x .۲‬ﻳﻚ ﻋﺪﺩ ﺧﺎﺭﺝ ﺍﺯ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﺍﺳﺖ‪،‬‬
‫‪ x .۳‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﻳﻚ ﻋﺪﺩ ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﺣﺎﻟﺖ ﺍﻭﻝ‪ ،‬ﻛﻪ ‪ x‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ‪ ،‬ﻋﺪﺩ ‪ x‬ﺑﺪﻭﻥ ﻫﻴﭻ ﺗﻐﻴﻴﺮﻱ ﺩﺭ ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭﺣﺎﻟﺖ ﺩﻭﻡ‪ ،‬ﻛﻪ ‪x‬‬
‫ﺧﺎﺭﺝ ﺍﺯ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ ،‬ﺫﺧﻴﺮﻩ ‪ x‬ﺍﻣﻜﺎﻥﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ‪ xmax‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻗﺎﺑﻞ ﻧﻤﺎﻳﺶ ﺩﺭ‬
‫ﺳﻴﺴﺘﻢ ﻣﻲﺑﺎﺷﺪ ﻭ ﺗﺒﻌﺎ ﺍﻋﺪﺍﺩﻱ ﻛﻪ ﺩﺭ ﺧﺎﺭﺝ ﺍﺯ ﺑﺎﺯﻩ ] ‪ [−xmax , xmax‬ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ ﺭﺍ ﻧﻤﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺶ ﺩﺍﺩ‪ .‬ﺩﺭ ﭼﻨﻴﻦ ﻭﺿﻌﻴﺘﻲ‬
‫ﮔﻮﻳﻴﻢ ﻛﻪ ﺳﺮﺭﻳﺰﻱ ‪ ۲۰‬ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫ﺳﺮﺭﯾﺰﯼ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۱‬‬
‫ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﻛﻪ ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺁﻥ ﺑﺮﺍﺑﺮ ‪ x max‬ﺍﺳﺖ‪ ،‬ﺍﮔﺮ ﺑﺨﻮﺍﻫﻴﻢ ﻋﺪﺩ ‪ x‬ﻛﻪ ‪ |x| > xmax‬ﺭﺍ ﺫﺧﻴﺮﻩ‬
‫ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﻋﻤﻞ ﻣﻤﻜﻦ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺍﺻﻄﻼﺡ ﻣﻲﮔﻮﺋﻴﻢ ﺳﺮﺭﻳﺰﻱ ﻳﺎ ﺧﻄﺎﻱ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪.‬‬

‫ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ ﻣﺸﺨﺺ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ‬

‫[‬ ‫]‬
‫‪−xmax‬‬ ‫‪xmax‬‬

‫ﺍﻣﺎ ﺩﺭ ﺣﺎﻟﺖ ﺳﻮﻡ‪ ،‬ﻛﻪ ‪ x‬ﻣﺎﺷﻴﻨﻲ ﻧﻤﻲﺑﺎﺷﺪ ﻭﻟﻲ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ ،‬ﺑﺎﺯ ﻫﻢ ﻧﻤﻲﺗﻮﺍﻥ ﻋﺪﺩ ﺭﺍ ﺩﺭ ﻣﺎﺷﻴﻦ‬
‫ﺫﺧﻴﺮﻩ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺍﻋﻤﺎﻝ ﺗﻐﻴﻴﺮﻱ ﺩﺭ ‪ ،x‬ﺁﻥ ﺭﺍ ﺑﻪ ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺗﺒﺪﻳﻞ ﻛﺮﺩ ﻭ ﺩﺭ ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻛﺮﺩ‪ .‬ﺍﻳﻦ ﻛﺎﺭ ﺑﻪ‬
‫ﺭﻭﻧﺪ ﻛﺮﺩﻥ ‪ ۲۱‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﻧﮕﺎﺷﺘﻲ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻣﻮﺟﻮﺩ ﺩﺭ ] ‪ [−xmax , +xmax‬ﺭﺍ ﺑﻪ‬
‫ﻃﻮﺭ ﻣﻨﺎﺳﺒﻲ ﺑﻪ ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﺑﺮﺭﺳﻲ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫‪20 Overflow‬‬ ‫‪21 Rounding‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸‬‬

‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‬ ‫�‬

‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻳﻲ ﺑﺎ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ) ‪ ،FS (β, m, emin , emax‬ﺍﮔﺮ ﻧﮕﺎﺷﺖ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ rd‬ﻧﺸﺎﻥ ﺩﻫﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫ﺍﻳﻦ ﻧﮕﺎﺷﺖ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫) ‪rd : [−xmax , +xmax ] −→ FS (β, m, emin , emax‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺍﮔﺮ ‪ x‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ .rd(x) = x‬ﺍﻣﺎ ﺍﮔﺮ ‪x‬‬
‫ﻣﺎﺷﻴﻨﻲ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻮﺳﻂ ﺩﻭﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺩﻳﮕﺮ ﺍﺣﺎﻃﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻓﺮﺽ ﻛﻨﻴﻢ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x−‬ﻭ ‪ x+‬ﺑﺎﺷﻨﺪ ﻭ‬
‫ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ .x− < x < x+ :‬ﻃﺒﻴﻌﻲ ﺍﺳﺖ ﻛﻪ )‪ rd(x‬ﺑﺎﻳﺪ ﺑﺮﺍﺑﺮ ‪ x−‬ﻭ ﻳﺎ ‪ x+‬ﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﺑﺘﺪﺍ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ‬
‫ﻫﺮ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ ،[−xmax , +xmax‬ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ‪ x−‬ﻭ ‪ x+‬ﺭﺍ ﺗﻌﻴﻴﻦ ﻛﻨﻴﻢ‪.‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ x‬ﻋﺪﺩﻱ ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﺑﺎﺷﺪ‪ .‬ﻣﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺮﺍﻱ ‪ x‬ﺍﺭﺍﻳﻪ‬
‫ﻧﻤﻮﺩ‪:‬‬
‫‪.‬‬
‫‪x = ±0 d−1 d−2 · · · d−m d−m−1 · · · × β e‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ }‪ di ∈ {0, 1, . . . , β − 1‬ﻭ ‪ .emin ≤ e ≤ emax‬ﺑﺴﺘﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ x‬ﻣﺜﺒﺖ ﺑﺎﺷﺪ ﻳﺎ ﻣﻨﻔﻲ‪ ،‬ﺍﻋﺪﺍﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ‪ x−‬ﻭ ‪ x+‬ﺑﻪ ﺷﻜﻞ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ ﺩﻭ ﺣﺎﻟﺖ ﺑﺮﺍﻱ ‪ x‬ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﮔﺮ ‪ x‬ﻋﺪﺩﻱ ﻣﺜﺒﺖ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪ x = +0.d−1 d−2 · · · d−m d−m−1 · · · × β e‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫
‬
‫‪.‬‬
‫‪x− := +0 d−1 d−2 · · · d−m × β e ,‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪x+ := + 0 d−1 d−2 · · · d−m + 0 0| ·{z‬‬
‫‪· · 0} 1 × β e‬‬
‫‪m−1‬‬
‫ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﺑﻬﺘﺮ‪ ،‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪.‬‬
‫‪x = +0 d−1 d−2 · · · d−m d−m−1 · · · × β e‬‬
‫‪−xmax‬‬ ‫‪0‬‬ ‫‪xmax‬‬
‫×‬

‫‪.‬‬
‫‪x− = 0 d−1 d−2 · · · d−m × β e‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪x+ = (0 d−1 d−2 · · · d−m + 0 0 · · · 01) × β e‬‬

‫ﺍﻣﺎ ﺍﮔﺮ ‪ x‬ﻋﺪﺩﻱ ﻣﻨﻔﻲ ﺑﻪ ﺻﻮﺭﺕ ‪ x = −0.d−1 d−2 · · · d−m d−m−1 · · · × β e‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫
‬
‫‪.‬‬
‫‪x− := − 0 d−1 d−2 · · · d−m + 0 |0 ·{z‬‬ ‫‪.‬‬
‫‪· · 0} 1 × β e ,‬‬ ‫‪.‬‬
‫‪x+ := −0 d0−1d−2 · · · d−m × β e‬‬
‫‪m−1‬‬
‫ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﺑﻬﺘﺮ‪ ،‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪.‬‬
‫‪x = −0 d−1 d−2 · · · d−m d−m−1 · · · × β e‬‬
‫‪−xmax‬‬ ‫‪xmax‬‬
‫‪0‬‬
‫×‬

‫‪.‬‬ ‫‪.‬‬
‫‪x− = −(0 d−1 d−2 · · · d−m + 0 0 · · · 01) × β e‬‬ ‫‪.‬‬
‫‪x+ = −0 d−1 d−2 · · · d−m × β e‬‬

‫ﺣﺎﻝ ﺑﻪ ﻧﺤﻮﻩ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺑﺮﻣﻲﮔﺮﺩﻳﻢ‪ .‬ﺭﻭﺵﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﺗﻤﺎﻣﻲ ﺍﻳﻦ ﺭﻭﺵﻫﺎ‪ ،‬ﺍﮔﺮ ‪x‬‬
‫ﻋﺪﺩﻱ ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻳﺎ ‪ x−‬ﻳﺎ ‪ x+‬ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻭﻧﺪ ﺷﺪﻩ ‪ x‬ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺍﻳﻦ‬
‫ﻛﻪ ﻛﺪﺍﻡﻳﻚ ﺍﻧﺘﺨﺎﺏ ﺷﻮﺩ ﺑﻪ ﻧﻮﻉ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪ .‬ﺍﺯ ﺟﻤﻠﻪ ﺍﻧﻮﺍﻉ ﺭﻭﻧﺪ ﻛﺮﺩﻥ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ‪،۲۲‬‬
‫‪22 Rounding‬‬ ‫‪toward zero‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۹‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ ‪ ،۲۳‬ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﭘﺎﻳﻴﻦ ‪ ۲۴‬ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ ۲۵‬ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪.‬‬

‫■ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ‬

‫ﺍﮔﺮ ] ‪ x ∈ [−xmax , +xmax‬ﻋﺪﺩﻱ ﻏﻴﺮ ﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﻧﺪ ﺷﺪﻩ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ‪ x‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺍﻭﻟﻴﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬
‫ﻛﻪ ﺩﺭ ﺳﺮ ﺭﺍﻩ ﺣﺮﻛﺖ ﺍﺯ ‪ x‬ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺭﻭﻧﺪ ﺷﺪﻩ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ‪ x‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬
‫ﺍﺳﺖ ﻛﻪ ﻛﻮﭼﻚﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ x‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺭﻭﻧﺪ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ rd0‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺩﺭ ﺷﻜﻞ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ‬
‫ﺻﻔﺮ ﺑﻪ ﺗﺼﻮﻳﺮ ﻛﺸﻴﺪﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪−xmax‬‬ ‫‪xmax‬‬
‫‪x‬‬ ‫‪0‬‬ ‫‪y‬‬
‫×‬ ‫×‬
‫‪x−‬‬ ‫‪x+‬‬ ‫‪y−‬‬ ‫‪y+‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺍﮔﺮ ‪ x‬ﻋﺪﺩﻱ ﻣﻨﻔﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ )‪ rd0 (x‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ x+‬ﻭ ﺍﮔﺮ ‪ x‬ﻋﺪﺩﻱ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ )‪ rd0 (x‬ﺑﺮﺍﺑﺮ ‪x−‬‬
‫ﺍﺳﺖ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺍﮔﺮ ﻋﺪﺩ ﺣﻴﻘﻴﻖ ﻭ ﻣﻨﻔﻲ ‪ x‬ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ‪ x = −0.d−1 · · · d−m d−(m+1) · · · × β‬ﺑﺎﺷﺪ‪،‬‬
‫‪e‬‬

‫ﺁﻥﮔﺎﻩ ‪ x+ = −0.d−1 · · · d−m × β e‬ﻭ ﺍﮔﺮ ‪ x‬ﻣﺜﺒﺖ ﻭ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ‪x = +0.d−1 · · · d−m d−(m+1) · · · × β e‬‬
‫ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ .x− = +0.d−1 · · · d−m × β e‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦ ) ‪ ،FS (β, m, emin , emax‬ﺿﺎﺑﻄﻪ ﺭﻭﻧﺪ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ‬
‫ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫
‬
‫‪.‬‬ ‫‪.‬‬
‫‪rd0 ± 0 d−1 · · · d−m d−(m+1) · · · × β e = ± 0 d−1 · · · d−m × β e‬‬
‫ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﺑﺮﺍﻱ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﻳﻚ ﻋﺪﺩ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ‪ ،‬ﻛﺎﻓﻲ ﺍﺳﺖ ﻛﻪ ‪ m‬ﺭﻗﻢ ﺍﻭﻝ ﺍﻋﺸﺎﺭ ﺩﺭ ﻧﻤﺎﻳﺶ )؟؟( ﻧﮕﻪ ﺩﺍﺷﺘﻪ ﺷﻮﺩ‬
‫ﺑﻘﻴﻪ ﺍﺭﺍﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺁﻥ ﻋﺪﺩ ﺣﺬﻑ ﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺧﺎﻃﺮ‪ ،‬ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﺑﻪ ﺑﺮﺵﺩﺍﺩﻥ ‪ ۲۶‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫■ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ ﻭ ﭘﺎﻳﻴﻦ‬

‫ﺭﻭﻧﺪ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ ﻋﺪﺩ ] ‪ x ∈ [−xmax , +xmax‬ﺭﺍ ﺑﺎ )‪ rdu (x‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻭ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺍﻭﻟﻴﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺰﺭﮒﺗﺮ‬
‫ﻳﺎ ﻣﺴﺎﻭﻱ ‪ .x‬ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ ﺑﻪ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﻲﻧﻬﺎﻳﺖ ‪ ۲۷‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪−xmax‬‬ ‫‪xmax‬‬
‫‪x‬‬ ‫‪0‬‬ ‫‪y‬‬
‫×‬ ‫×‬
‫‪x−‬‬ ‫‪x+‬‬ ‫‪y−‬‬ ‫‪y+‬‬

‫ﺩﺭ ﻭﺍﻗﻊ ﭼﻪ ‪ x‬ﻣﻨﻔﻲ ﺑﺎﺷﺪ ﻳﺎ ﻣﺜﺒﺖ‪ ،‬ﺁﻥﮔﺎﻩ ‪ .rdu (x) = x+‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ x+‬ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﻣﻨﻔﻲ ﻭ ﻣﺜﺒﺖ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﻣﺘﻔﺎﻭﺗﻲ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‪:‬‬
‫‪23 Rounding‬‬ ‫‪up‬‬ ‫‪26‬‬ ‫‪Cutting‬‬
‫‪24 Rounding‬‬ ‫‪down‬‬ ‫‪27 rounding‬‬ ‫‪towards positive infinity‬‬
‫‪25 rounding to the nearest machine number‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰‬‬

‫
‬
‫‪.‬‬ ‫‪.‬‬
‫‪rdu −0 d−1 · · · d−m d−(m+1) · · · × β e = −0 d−1 · · · d−m × β e ,‬‬
‫‬ ‫‬
‫
‬
‫‪.‬‬ ‫‪.‬‬
‫‪rdu +0 d−1 · · · d−m d−(m+1) · · · × β e = + 0 d−1 · · · d−m + 0 0| ·{z‬‬ ‫‪.‬‬
‫‪· · 0} 1 × β e .‬‬
‫‪m−1‬‬

‫ﺭﻭﻧﺪ ﺑﻪ ﺳﻤﺖ ﭘﺎﻳﻴﻦ ) ﻭ ﻳﺎ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻣﻨﻔﻲ ﺑﻲﻧﻬﺎﻳﺖ ‪ (۲۸‬ﺭﺍ ﺑﺎ ‪ rdd‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻭ ﻣﺸﺎﺑﻪ ﺑﺎ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ‬
‫ﺳﻤﺖ ﺑﺎﻻ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ‪:‬‬
‫
‬ ‫‬ ‫‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪· · 0} 1 × β e ,‬‬
‫‪rdl −0 d−1 · · · d−m d−(m+1) · · · × β e = − 0 d−1 · · · d−m + 0 0| ·{z‬‬
‫‪m−1‬‬
‫
‬
‫‪.‬‬
‫‪rdl +0 d−1 · · · d−m d−(m+1) · · · × β‬‬ ‫‪e‬‬
‫‪.‬‬
‫‪= +0 d−1 · · · d−m × β .‬‬
‫‪e‬‬

‫■ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬

‫ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ x‬ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﻛﻪ ﮔﺮﺩﻛﺮﺩﻥ ﻧﻴﺰ ﻧﺎﻣﻴﺪﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻳﺎﻓﺘﻦ ﻳﻚ ﻋﺪﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﻛﻪ ﻛﻤﺘﺮﻳﻦ ﻓﺎﺻﻠﻪ ﺭﺍ ﺍﺯ ‪ x‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﻧﮕﺎﺷﺖ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺑﺎ ‪ rdr‬ﻧﺸﺎﻥ‬
‫ﺩﻫﻴﻢ‪.‬‬
‫ﻣﺎﺷﻴﻦ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪ FS (β, m, emin , emax‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ ﻭ ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x‬ﻋﺪﺩﻱ ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺑﺎﺯﻩ ] ‪[−xmax , +xmax‬‬
‫ﺑﺎﺷﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﮔﻔﺘﻪ ﺷﺪ‪ x ،‬ﺗﻮﺳﻂ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x−‬ﻭ ‪ x+‬ﺍﺣﺎﻃﻪ ﺷﺪﻩ ﺍﺳﺖ ﻭ ﻫﺮﻛﺪﺍﻡ ﻛﻪ ﺑﻪ ‪ x‬ﻧﺰﺩﻳﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺑﻪ‬
‫ﻋﻨﻮﺍﻥ ﺭﻭﻧﺪﺷﺪﻩ ‪ x‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻋﺪﺩ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [x− , x+‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﻧﻘﻄﻪ ﻭﺳﻂ ﺍﻳﻦ ﺑﺎﺯﻩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‬ ‫‬
‫‪.‬‬ ‫‪1‬‬
‫‪x∗ = 0 d−1 · · · d−m + 0 |0 ·{z‬‬
‫‪2‬‬
‫‪.‬‬
‫‪· · 0} 1 × β e‬‬
‫‪m−1‬‬

‫ﺣﺎﻝ ﺍﮔﺮ ∗‪ x > x‬ﺁﻥﮔﺎﻩ ‪ x+‬ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻪ ‪ x‬ﻣﻲﺑﺎﺷﺪ ﻭ ﺍﮔﺮ ‪ x < x‬ﺁﻥﮔﺎﻩ ‪ x‬ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬
‫‪−‬‬ ‫∗‬

‫ﺑﻪ ‪ x‬ﻣﻲﺑﺎﺷﺪ ﻭ ﺍﮔﺮ ∗‪ x = x‬ﺁﻥﮔﺎﻩ ﻓﺎﺻﻠﻪ ‪ x+‬ﻭ ‪ x−‬ﺗﺎ ‪ x‬ﻳﻜﺴﺎﻥ ﺍﺳﺖ ﻭ ﻫﺮﺩﻭ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﺑﻪ ‪ x‬ﻣﻌﺮﻓﻲ ﻧﻤﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺿﺎﺑﻄﻪ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x− ,‬‬ ‫∗‪x < x‬‬
‫‪‬‬
‫‪‬‬
‫‪rdr (x) = x+ ,‬‬ ‫∗‪x > x‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪ x+ ,‬ﻳﺎ ‪x−‬‬ ‫∗‪x = x‬‬

‫ﺩﺭ ﺣﺎﻻﺗﻲ ﻛﻪ ∗‪ x = x‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺩﻟﺨﻮﺍﻩ ﻣﻲﺗﻮﺍﻧﻴﻢ ﮔﺮﺩﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﭘﺎﺋﻴﻦ ﻭ ﻳﺎ ﺑﺎﻻ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪ .‬ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﻛﻪ ﺩﺭ‬
‫ﻧﻴﻤﻲ ﺍﺯ ﻣﻮﺍﺭﺩ ﮔﺮﺩﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ ﻭ ﺩﺭ ﺑﺎﻗﻲ ﮔﺮﺩﻛﺮﺩﻥ ﺭﺍ ﺑﻪ ﺳﻤﺖ ﭘﺎﺋﻴﻦ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻳﻚ ﺭﻗﻢ ﻣﺎﻧﺘﻴﺲ )ﻣﺜﻼ‬
‫ﺭﻗﻢ ‪m − 1‬ﺍﻡ( ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺍﮔﺮ ﺍﻳﻦ ﺭﻗﻢ ﻓﺮﺩ ﺑﻮﺩ‪ ،‬ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x+‬ﻭ ﺍﮔﺮ ﺯﻭﺝ ﺑﻮﺩ‪ x− ،‬ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻭﻧﺪ ﺷﺪﻩ ﻋﺪﺩ‬
‫ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻧﺴﺒﺖ ﺑﻪ ﺩﻳﮕﺮ ﺭﻭﻧﺪﻛﺮﺩﻥﻫﺎ ﻣﻌﻘﻮﻝﺗﺮ ﻭ ﻣﻨﺎﺳﺐﺗﺮ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ‪ .‬ﺍﻣﺎ ﺍﻧﺠﺎﻡ ﺁﻥ‬
‫ﭘﻴﭽﻴﺪﻩﺗﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻩ ﺑﺎ ﻣﻘﺎﻳﺴﻪ ﺭﻗﻢ )‪(m + 1‬ﺍﻡ ﻣﺎﻧﺘﻴﺲ ﺑﺎ ﻋﺪﺩ ‪ 5‬ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﻭ ﻳﺎ ﻫﻤﺎﻥ ﮔﺮﺩﻛﺮﺩﻥ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪28 Rounding‬‬ ‫‪towards positive infinity‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۱‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻣﺎﺷﻴﻨﻲ ﺍﻋﺪﺍﺩ‬ ‫‪۴.۱‬‬

‫ﺩﺭ ﺩﻫﻪﻫﺎﻱ ‪ ۱۹۶۰‬ﻭ ‪ ۱۹۷۰‬ﺳﺎﺯﻧﺪﮔﺎﻥ ﻣﺨﺘﻠﻒ ﭘﺮﺩﺍﺯﻧﺪﻩﻫﺎﻱ ﻛﺎﻣﭙﻴﻮﺗﺮﻱ ﺍﺯ ﺭﻭﻳﻜﺮﺩﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﺮﺩﻧﺪ‪ .‬ﺍﮔﺮﭼﻪ ﺍﻛﺜﺮ ﺁﻥﻫﺎ ﺍﺯ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﺑﺎ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﺮﺩﻧﺪ‪ ،‬ﻟﻴﻜﻦ ﺗﻔﺎﻭﺕﻫﺎﺋﻲ ﺩﺭ ﺍﻧﺘﺨﺎﺏ ﻣﺒﻨﺎ‬
‫ﻭ ﺍﻧﺪﺍﺯﻩﻫﺎﻱ ﻣﺎﻧﺘﻴﺲ ﻭ ﻧﻤﺎ‪ ،‬ﻧﺤﻮﻩ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻭ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﺗﻔﺎﻭﺕﻫﺎﻳﻲ ﺩﺍﺷﺘﻨﺪ‪ .‬ﺍﻳﻦ ﺗﻔﺎﻭﺕﻫﺎ ﻣﻨﺠﺮ ﺑﻪ‬
‫ﻧﺎﺳﺎﺯﮔﺎﺭﻱ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﻳﻜﺪﻳﮕﺮ ﻣﻲﺷﺪ‪ .‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻳﻚ ﺑﺮﻧﺎﻣﻪ ﻛﺎﻣﭙﻴﻮﺗﺮﻱ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺨﺘﻠﻒ ﻣﻨﺠﺮ ﺑﻪ ﺟﻮﺍﺏﻫﺎﻱ‬
‫ﻣﺘﻔﺎﻭﺕ ﻣﻲﮔﺮﺩﻳﺪ ﻭ ﺣﺘﻲ ﺑﺮﺧﻲ ﻧﺮﻡﺍﻓﺰﺍﺭﻫﺎ ﺭﻭﻱ ﻳﻚ ﻣﺎﺷﻴﻦ ﺑﻪ ﺩﺭﺳﺘﻲ ﻋﻤﻞ ﻣﻲﻛﺮﺩﻧﺪ ﺍﻣﺎ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺩﻳﮕﺮ ﻋﻤﻠﻜﺮﺩ‬
‫ﺩﺭﺳﺘﻲ ﻧﺪﺍﺷﺘﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﺩﺭ ﺳﺎﻝ ‪ ۱۹۸۵‬ﺍﺯ ﺳﻮﻱ ﺍﻧﺴﺘﻴﺘﻮﻱ ﻣﻬﻨﺪﺳﻲ ﺑﺮﻕ ﻭ ﺍﻟﻜﺘﺮﻭﻧﻴﻚ ‪ (IEEE) ۲۹‬ﺍﺳﺘﺎﻧﺪﺍﺭﺩﻱ ﺑﺮﺍﻱ‬
‫ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﻭﺿﻊ ﺷﺪ‪ .‬ﺳﭙﺲ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﺩﺭ ﺳﺎﻝ ‪ ۱۹۸۹‬ﺍﺯ ﺳﻮﻱ ﻛﻤﻴﺴﻴﻮﻥ ﺑﻴﻦﺍﻟﻤﻠﻠﻲ ﺍﻟﻜﺘﺮﻭﻧﻴﻚ ‪ (IEC) ۳۰‬ﻣﻮﺭﺩ ﺗﺎﻳﻴﺪ‬
‫ﻗﺮﺍﺭ ﮔﺮﻓﺖ ﻭ ﺍﺯ ﺁﻥ ﭘﺲ ﺳﺎﺯﻧﺪﮔﺎﻥ ﭘﺮﺩﺍﺯﻧﺪﻩﻫﺎ‪ ،‬ﻣﺎﻧﻨﺪ ‪ Intel‬ﻭ ‪ ،Motorola‬ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﺭﺍ ﺩﺭ ﺳﺎﺧﺖ ﻣﺤﺼﻮﻻﺕ ﺧﻮﺩ ﺑﻪﻛﺎﺭ‬
‫ﺑﺮﺩﻧﺪ‪ .‬ﺍﻭﻟﻴﻦ ﻧﮕﺎﺭﺵ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ‪ ،‬ﻛﻪ ﺩﺭ ﺳﺎﻝ ‪ ۱۹۸۵‬ﺍﺭﺍﻳﻪ ﺷﺪ‪ ،‬ﺑﻪ ‪ IEEE 754-1985‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻧﮕﺎﺭﺵﻫﺎﻱ ﺩﻳﮕﺮ‬
‫ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﺑﺎ ﻧﺎﻡﻫﺎﻱ ‪ IEEE 754-2008‬ﻭ ‪ IEEE 754-2019‬ﺩﺭ ﺭﺍﺳﺘﺎﻱ ﻫﺮﭼﻪ ﺑﻴﺸﺘﺮ ﺑﺎﻻﺑﺮﺩﻥ ﻛﺎﺭﺍﻳﻲ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ‬
‫ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺩﺭ ﺧﺼﻮﺹ ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺧﻼﺻﻪ ﺍﺭﺍﺋﻪ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ‬
‫ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﻓﻘﻂ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻭ ﻧﻤﺎﻳﺶ ﺍﻋﺪﺍﺩ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻠﻜﻪ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﻗﻮﺍﻋﺪﻱ ﻧﻴﺰ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ‪ ،‬ﻧﺤﻮﻩ‬
‫ﺭﻭﻧﺪﻛﺮﺩﻥ ﻭ ﻏﻴﺮﻩ ﻧﻴﺰ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﻣﺒﺎﺣﺚ ﺑﻌﺪ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﺁﻥﻫﺎ ﺍﺷﺎﺭﻩ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﺮﺍﻱ ﺟﺰﺋﻴﺎﺕ ﺑﻴﺸﺘﺮ ﺩﺭ ﺧﺼﻮﺹ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ‬
‫‪ IEEE‬ﺑﻪ ]؟ [ ﺍﺭﺟﺎﻉ ﺷﻮﺩ‪.‬‬
‫ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺳﻴﺴﺘﻢ ﻧﻤﺎﻳﺸﻲ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺭﺍ ﺍﻟﮕﻮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﺪ‪ .‬ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺩﺭ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ‪ ،‬ﺫﺧﻴﺮﻩ ﺍﻋﺪﺍﺩ ﻣﺒﺘﻨﻲ‬
‫ﺑﺮ ﻧﻤﺎﻳﺶ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻧﺮﻣﺎﻝ‪-‬ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﻭ ﻣﺠﻤﻮﻋﻪ ) ‪ FS (β, m, emin , emax‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺍﻟﺰﺍﻡ ﻣﻲﻛﻨﺪ ﻛﻪ‬
‫ﻋﻼﻭﻩ ﺑﺮ ﺍﻋﺪﺍﺩ ) ‪ ،FS (β, m, emin , emax‬ﻧﻤﺎﺩﻫﺎﻱ ‪ −Inf ،+Inf‬ﻭ ‪ ۳۱ NaN‬ﻧﻴﺰ ﺑﻪ ﻋﻨﻮﺍﻥ ﺍﻋﺪﺍﺩ ﺧﺎﺹ ﺗﻮﺳﻂ ﻣﺎﺷﻴﻦﻫﺎ ﺩﺭ‬
‫ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻳﻦ ﻧﻤﺎﺩﻫﺎ ﺩﺭ ﻛﻨﺎﺭ ) ‪ ،FS (β, m, emin , emax‬ﺑﻪ ﻋﻨﻮﺍﻥ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﻧﺪ ﻭ ﺑﻪ‬
‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺧﺎﺹ ﻳﺎ ﺍﺳﺘﺜﻨﺎﻳﻲ ‪ ۳۲‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﺯ ‪ ±Inf‬ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ∞ ‪ ±‬ﻭ ﺍﺷﺎﺭﻩ ﺑﻪ ﺳﺮﺭﻳﺰﻱ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺑﺨﻮﺍﻫﻴﻢ ‪ x > x max‬ﺭﺍ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ ،‬ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺧﺎﺹ ‪ +Inf‬ﺫﺧﻴﺮﻩ ﺧﻮﺍﻫﺪ ﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ‬
‫‪ −1‬ﺭﺍ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺧﺎﺹ ‪ −Inf‬ﺫﺧﻴﺮﻩ ﺧﻮﺍﻫﺪ ﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺍﺯ ‪NaN‬‬ ‫ﻭﻗﺘﻲ ﺑﺨﻮﺍﻫﻴﻢ ﺣﺎﺻﻞ ﻋﺒﺎﺭﺕ ‪0‬‬
‫ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﺣﺎﻟﺖﻫﺎﻱ ﻣﺒﻬﻢ ﻭ ﺗﻌﺮﻳﻒ ﻧﺸﺪﻩ ﻣﺎﻧﻨﺪ ﻣﻮﺍﺭﺩ ﺯﻳﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‬
‫√‬ ‫‪0‬‬ ‫∞‬
‫‪−1,‬‬ ‫‪,‬‬ ‫‪0 × ∞,‬‬ ‫‪∞ − ∞,‬‬ ‫···‪,‬‬
‫‪0‬‬ ‫∞‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‬
‫∪ ) ‪FS+ (β, m, emin , emax ) := FS (β, m, emin , emax‬‬ ‫‪− Inf, +Inf, NaN .‬‬ ‫)‪(۴.۱‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ‪:‬‬

‫» ﻣﻘﺪﺍﺭ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ emin ،m ،β‬ﻭ ‪ emax‬ﺩﺭ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﭼﻪ ﻣﻲﺑﺎﺷﺪ؟ «‬

‫ﺩﺭ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE 745-2008‬ﻫﻔﺖ ﺍﻧﺘﺨﺎﺏ ﻣﺘﻔﺎﻭﺕ ﺑﺮﺍﻱ ﺍﻳﻦ ﭘﺎﺭﺍﻣﺘﺮﻫﺎ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﻫﺮ ﺍﻧﺘﺨﺎﺏ ﺑﻪ ﻳﻚ ﻓﺮﻣﺖ ﻣﻮﺳﻮﻡ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻣﻘﺎﺩﻳﺮ ﭘﺎﺭﺍﻣﺘﺮﻫﺎ ﺩﺭ ﺍﻳﻦ ﻫﻔﺖ ﻓﺮﻣﺖ ﺑﻪ ﺷﺮﺡ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪29‬‬ ‫‪Institute of Electrical and Electronic Engineering‬‬ ‫‪31 Not‬‬ ‫) ‪a Number(NaN‬‬
‫‪30‬‬ ‫‪International Electronical Commission‬‬ ‫‪32 Exception‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲‬‬

‫ﻧﺎﻡ ﻓﺮﻣﺖ ﺩﺭ‬ ‫ﺗﻌﺪﺍﺩ ﺑﺎﻳﺖﻫﺎﻱ‬ ‫ﻣﺒﻨﺎ‬ ‫ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ‬ ‫ﻛﺮﺍﻥ ﭘﺎﻳﻴﻦ‬ ‫ﻛﺮﺍﻥ ﺑﺎﻻﻱ‬
‫‪IEEE 745-2008‬‬ ‫ﻻﺯﻡ ﺩﺭ ﻓﺮﻣﺖ‬ ‫)‪(β‬‬ ‫ﻣﺎﻧﺘﻴﺲ )‪(m‬‬ ‫ﻧﻤﺎ ) ‪(emin‬‬ ‫ﻧﻤﺎ ) ‪(emax‬‬
‫‪Binary16‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪11‬‬ ‫‪−13‬‬ ‫‪16‬‬
‫‪Binary32‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪24‬‬ ‫‪−125‬‬ ‫‪128‬‬
‫‪Binary64‬‬ ‫‪8‬‬ ‫‪2‬‬ ‫‪53‬‬ ‫‪−1021‬‬ ‫‪1024‬‬
‫‪Binary128‬‬ ‫‪16‬‬ ‫‪2‬‬ ‫‪113‬‬ ‫‪−16381‬‬ ‫‪16384‬‬
‫‪Decimal32‬‬ ‫‪4‬‬ ‫‪10‬‬ ‫‪7‬‬ ‫‪−95‬‬ ‫‪96‬‬
‫‪Decimal64‬‬ ‫‪8‬‬ ‫‪10‬‬ ‫‪16‬‬ ‫‪−382‬‬ ‫‪385‬‬
‫‪Decimal128‬‬ ‫‪16‬‬ ‫‪10‬‬ ‫‪34‬‬ ‫‪−6142‬‬ ‫‪6145‬‬

‫ﺍﺭ ﺑﻴﻦ ﻓﺮﻣﺖﻫﺎﻱ ﻓﻮﻕ‪ ،‬ﺩﻭ ﻓﺮﻣﺖ ‪ Binary32‬ﻭ ‪ Binary64‬ﺭﺍﻳﺞﺗﺮ ﻭ ﭘﺮﻛﺎﺭﺑﺮﺩﺗﺮ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﻪ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ‪۳۳‬‬

‫ﻭ ﺩﻗﺖ ﻣﻀﺎﻋﻒ ‪ ۳۴‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻳﻦ ﺩﻭ ﻓﺮﻣﺖ ﺑﻪ ﺗﺮﺗﻴﺐ ﻣﻌﺎﺩﻝ ﻧﻮﻉ ﺩﺍﺩﻩﺍﻱ ‪ float‬ﻭ ‪ double‬ﺩﺭ ﺯﺑﺎﻥ ‪ C‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻣﺠﻤﻮﻋﻪ‬
‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﺖﻫﺎ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‬
‫‪ :‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ‬ ‫∪ )‪FS (2, 24, −125, 128‬‬ ‫‪− Inf, +Inf, NaN ,‬‬
‫‬
‫‪ :‬ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻀﺎﻋﻒ‬ ‫‪FS (2, 53, −1021, 1024) ∪ − Inf, +Inf, NaN .‬‬
‫ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻭ ﺧﺼﻮﺻﺎ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻀﺎﻋﻒ ﺷﺎﻣﻞ ﺍﻋﺪﺍﺩ ﺑﺴﻴﺎﺭ ﺑﺰﺭﮒ ﻭ ﺍﻋﺪﺍﺩ ﺑﺴﻴﺎﺭ‬
‫ﻛﻮﭼﻚ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﻛﻬﻜﺸﺎﻥﻫﺎ ﻭ ﻗﻄﺮ ﻫﺴﺘﻪ ﻳﻚ ﺍﻟﻜﺘﺮﻭﻥ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺍﻋﺪﺍﺩ ﺍﻳﻦ ﻓﺮﻣﺖﻫﺎ ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪.‬‬
‫ﺟﺰﺋﻴﺎﺕ ﺗﻮﺯﻳﻊ ﺍﻋﺪﺍﺩ ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﺖﻫﺎ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬

‫ﻧﺎﻡ‬ ‫ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ‬ ‫ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ‬ ‫ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ‬

‫ﻓﺮﻣﺖ‬ ‫ﻣﺜﺒﺖ )ﻧﻴﻤﻪﻧﺮﻣﺎﻝ(‬ ‫ﻣﺜﺒﺖ ﻧﺮﻣﺎﻝ‬ ‫ﻧﺮﻣﺎﻝ‬
‫‪21−m‬‬ ‫‪2emin −1‬‬ ‫) ‪2emax (1 − 2−m‬‬
‫ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ‬ ‫≃ ‪2−149‬‬ ‫≃ ‪2−126‬‬ ‫≃ ) ‪2128 (1 − 2−24‬‬
‫‪1.401×10−45‬‬ ‫‪1.176×10−38‬‬ ‫‪3.403×10+38‬‬

‫ﺩﻗﺖ ﻣﻀﺎﻋﻒ‬ ‫≃ ‪2−1074‬‬ ‫≃ ‪2−1022‬‬ ‫≃ ) ‪21024 (1 − 2−53‬‬

‫‪4.941×10−324‬‬ ‫‪2.225×10−308‬‬ ‫‪1.798×10+308‬‬

‫ﺩﺭ ﺑﺨﺶ ﺑﻌﺪ ﺟﺰﺋﻴﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﻗﺎﻟﺐﺑﻨﺪﻱ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻭ ﻣﻀﺎﻋﻒ ﺩﺭ ﺣﺎﻓﻈﻪ ﺑﺮﺍﻱ ﺧﻮﺍﻧﻨﺪﮔﺎﻥ ﻋﻼﻗﻪﻣﻨﺪ ﺍﺭﺍﻳﻪ‬
‫ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫∗ ﺟﺰﺋﻴﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻭ ﻣﻀﺎﻋﻒ‬ ‫�‬

‫ﺩﺭ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ ،IEEE‬ﺑﺮﺍﻱ ﻫﺮ ﻋﺪﺩ ‪ ۳۲‬ﺑﻴﺖ )ﭼﻬﺎﺭ ﺑﺎﻳﺖ( ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺍﺯ ﺍﻳﻦ ‪ ۳۲‬ﺑﻴﺖ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ‬
‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﺖ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺑﺎ ﻫﺮ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻳﻚ ﺭﺷﺘﻪ ‪۳۲‬ﺗﺎﻳﻲ ﺍﺯ ﺍﺭﻗﺎﻡ ‪ 0‬ﻭ ‪ 1‬ﻣﺘﻨﺎﻇﺮ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻣﻮﺟﻮﺩ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‪ FS (2, 24, −125, 128) :‬ﺑﻌﻼﻩ‬
‫‪33 Single‬‬ ‫‪precision‬‬ ‫‪34 Double‬‬ ‫‪precision‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۳‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺍﻋﺪﺍﺩ ﺧﺎﺹ ‪ ±Inf‬ﻭ ‪ .NaN‬ﻧﻤﺎﻱ ﺍﻋﺪﺍﺩ ﻣﻮﺟﻮﺩ ﺩﺭ )‪ FS (2, 24, −125, 128‬ﺑﻴﻦ ‪ −125‬ﻭ ‪ 128‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻳﻌﻨﻲ ﺑﻪ ﺗﻌﺪﺍﺩ‬
‫‪ 128 − (−125) + 1 = 254‬ﻣﻘﺪﺍﺭ ﻣﺘﻔﺎﻭﺕ ﺑﺮﺍﻱ ﻧﻤﺎﻱ ﺍﻋﺪﺍﺩ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﺍﻳﻦ ‪ 254‬ﻧﻤﺎ ﺑﻪ‬
‫‪ ۸‬ﺑﻴﺖ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎ ‪ ۸‬ﺑﻴﺖ‪ ۲۵۶ ،‬ﻋﺪﺩ ﺻﺤﻴﺢ ﻣﺘﻔﺎﻭﺕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ‪ .‬ﻣﺎﻧﺘﻴﺲ ﻫﻢ ﻛﻪ ﻋﺪﺩﻱ ‪ ۲۴‬ﺭﻗﻤﻲ )ﺩﺭ‬
‫ﻣﺒﻨﺎﻱ ﺩﻭ( ﻣﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺑﺎ ﻟﺤﺎﻅ ﻛﺮﺩﻥ ﺑﻴﺖ ﻋﻼﻣﺖ‪ ،‬ﺩﺭ ﻣﺠﻤﻮﻉ ﻣﺎ ﺑﻪ ﻳﻚ ﺭﺷﺘﻪ ‪ ۳۳‬ﺑﻴﺘﻲ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻛﻠﻴﻪ ﺍﻋﺪﺍﺩ ﻣﻮﺟﻮﺩ ﺩﺭ‬
‫)‪ FS (2, 24, −125, 128‬ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻣﺎ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﮔﻔﺘﻪ ﺷﺪ‪ ،‬ﺩﺭ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ‪ ۳۲‬ﺑﻴﺖ ﻭ ﻧﻪ ‪ ۳۳‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻋﺪﺩ ﺩﺭ ﻧﻈﺮ‬
‫ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻣﺸﻜﻞ ﻳﻚ ﺑﻴﺖ ﭼﮕﻮﻧﻪ ﺣﻞ ﻣﻲﺷﻮﺩ؟ ﺩﺭ ﭘﺎﺳﺦ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻝ‪،‬‬
‫ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺩﻗﻴﻘﺎ ﺭﻭﻳﻜﺮﺩ ﻓﻮﻕ ﺑﺮﺍﻱ ﻣﺘﻨﺎﻇﺮ ﻛﺮﺩﻥ ﺭﺷﺘﻪ‪-‬ﺑﻴﺘﻲ ﺑﺎ ﻫﺮ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺍﻧﺠﺎﻡ ﻧﻤﻲﺩﻫﺪ‪.‬‬
‫ﺑﻠﻜﻪ ﺑﻪ ﺻﻮﺭﺕ ﻫﻮﺷﻤﻨﺪﺍﻧﻪﺗﺮﻱ ﺍﻳﻦ ﺗﺨﺼﻴﺺ ﺭﺍ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﺪ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ‪ ۳۲ ،‬ﺑﻴﺖ ﺩﺭﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺑﻪ ‪ ۱‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻋﻼﻣﺖ‪ ۲۳ ،‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ ﻭ ‪ ۸‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻧﻤﺎ‬
‫ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﺗﻘﺴﻴﻢﺑﻨﺪﻱ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺑﻴﺖ ﻋﻼﻣﺖ‬

‫‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ‬ ‫‪ ۲۳‬ﺑﻴﺖ ﻣﺎﻧﺘﻴﺲ‬

‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺑﻌﺪ ﺍﺯ ﺑﻴﺖ ﻋﻼﻣﺖ‪ ،‬ﺑﻴﺖﻫﺎﻱ ﻧﻤﺎ )ﻭ ﻧﻪ ﻣﺎﻧﺘﻴﺲ( ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪ .‬ﺍﻳﻦ ﺁﺭﺍﻳﺶ ﺑﺎﻋﺚ ﺳﺎﺩﮔﻲ ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﺩﻭ‬
‫ﻋﺪﺩ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻧﺤﻮﻩ ﻣﻘﺪﺍﺭﺩﻫﻲ ﺑﻴﺖ ﻋﻼﻣﺖ‪ ،‬ﺑﻴﺖﻫﺎﻱ ﻧﻤﺎ ﻭ ﺑﻴﺖﻫﺎﻱ ﻣﺎﻧﺘﻴﺲ ﺩﺭ ﺯﻳﺮ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫• ﺑﺎ ﻳﻚ ﺑﻴﺖ ﻋﻼﻣﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻋﻼﻣﺖ ﻋﺪﺩ ﺭﺍ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺍﮔﺮ ﺩﺭ ﺑﻴﺖ ﻋﻼﻣﺖ ‪ 0‬ﺫﺧﻴﺮﻩ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﻣﺜﺒﺖ ﻭ‬
‫ﺍﮔﺮ ‪ 1‬ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﻣﻨﻔﻲ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‬
‫• ﺑﺎ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﻣﻲﺗﻮﺍﻥ ﺍﻋﺪﺍﺩ ﺻﺤﻴﺢ ‪ 0‬ﺍﻟﻲ ‪ 225‬ﺭﺍ ﻣﺸﺨﺺ ﻛﺮﺩ ﻛﻪ ﻣﻌﺎﺩﻝ ﺍﻋﺪﺍﺩ ‪ −126‬ﺍﻟﻲ ‪ 129‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺍﮔﺮ ﻋﺪﺩ ‪ 0‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﺎ ﻣﻌﺎﺩﻝ ‪ −126‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺩﺭ ﻭﺍﻗﻊ ﺍﮔﺮ ﻋﺪﺩ ̂‪ e‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﺎ ﺑﺮﺍﺑﺮ ‪ e = ê − 126‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻧﻤﺎﻱ‬
‫ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﺑﻴﻦ ‪ −125‬ﻭ ‪ 128‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﺮﺍﻱ ﺍﻳﻦ ﻧﻤﺎﻫﺎ ﺍﻋﺪﺍﺩ ‪ 1‬ﺍﻟﻲ ‪ 254‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ‬
‫ﻧﻤﺎ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﻋﺪﺍﺩ ‪ 0‬ﻭ ‪ 255‬ﺭﺍ ﻧﻴﺰ ﺫﺧﻴﺮﻩ ﻛﺮﺩ ﻛﻪ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺗﻮﺿﻴﺢ‬
‫ﺩﺍﺩﻩ ﺧﻮﺍﻫﺪ ﺷﺪ‪ ،‬ﻣﻘﺎﺩﻳﺮ ‪ 0‬ﻭ ‪ 1‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺑﺮﺍﻱ ﻣﻨﻈﻮﺭﻫﺎﻱ ﺧﺎﺻﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫• ﺣﺎﻝ ﺑﻪ ﺳﺮﺍﻍ ﺫﺧﻴﺮﻩ ﻣﺎﻧﺘﻴﺲ ﻣﻲﺭﻭﻳﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﻋﺪﺍﺩ ﺑﺎ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﺩﺍﺭﺍﻱ ﻣﺎﻧﻴﺲ ‪ ۲۴‬ﺑﻴﺘﻲ ﻣﻲﺑﺎﺷﻨﺪ‪،‬‬
‫ﺍﻣﺎ ﺍﺯ ‪ ۳۲‬ﺑﻴﺖ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺮﺍﻱ ﺍﻳﻦ ﺍﻋﺪﺍﺩ‪ ۲۳ ،‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻭﻟﻴﻦ‬
‫ﺭﻗﻢ ﻣﺎﻧﺘﻴﺲ ﺫﺧﻴﺮﻩ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺍﻭﻟﻴﻦ ﺭﻗﻢ ﻣﺎﻧﺘﻴﺲ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﻧﺮﻣﺎﻝ ﺩﺭ )‪ FS (2, 24, −125, 128‬ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ ﻭ‬
‫ﺍﻭﻟﻴﻦ ﺭﻗﻢ ﻣﺎﻧﺘﻴﺲ ﺑﺮﺍﻱ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪ ﻧﺮﻣﺎﻝ ﺍﻳﻦ ﻣﺠﻤﻮﻋﻪ ﺑﺮﺍﺑﺮ ‪ 0‬ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﮔﺮ ﺑﺪﺍﻧﻴﻢ ﻛﻪ ﻋﺪﺩ ﻧﺮﻣﺎﻝ ﺍﺳﺖ ﻭ ﻳﺎ‬
‫ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‪ ،‬ﺁﻥﮔﺎﻩ ﺿﺮﻭﺭﺗﻲ ﻧﺪﺍﺭﺩ ﻛﻪ ﺍﻭﻟﻴﻦ ﺭﺍ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ ،‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﻗﺮﺍﺭﺩﺍﺩ ﻣﻲﻛﻨﺪ ﻛﻪ ﺍﮔﺮ ﻋﺪﺩ‬
‫‪ 0‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﻣﺤﺴﻮﺏ ﮔﺮﺩﺩ ﻭ ﺍﮔﺮ ﻳﻚ ﻋﺪﺩ ﺑﻴﻦ ‪ 1‬ﻭ ‪ 254‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ‬
‫ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﻧﺮﻣﺎﻝ ﻓﺮﺽ ﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﺍﻳﻦ ﻗﺮﺍﺭﺩﺍﺩ‪ ،‬ﻳﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﻘﺪﺍﺭ ﻧﻤﺎ ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ‬
‫ﻛﻪ ﻋﺪﺩ ﻧﺮﻣﺎﻝ ﺍﺳﺖ ﻳﺎ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ‪ .‬ﻟﺬﺍ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺫﺧﻴﺮﻩ ﺭﻗﻢ )ﺑﻴﺖ( ﺍﻭﻝ ﻣﺎﻧﺘﻴﺲ ﺻﺮﻑ ﻧﻈﺮ ﻛﺮﺩ ﻭ ﺑﺪﻳﻦ ﺗﺮﺗﻴﺐ ﺩﺭ‬
‫ﺣﺎﻓﻈﻪ ﺻﺮﻓﻪﺟﻮﻳﻲ ﻧﻤﻮﺩ‪.‬‬
‫• ﺗﺎ ﺍﻳﻦﺟﺎ ﺗﻜﻠﻴﻒ ﺍﻋﺪﺍﺩ ﻣﻮﺟﻮﺩ ﺩﺭ )‪ FS (2, 24, −125, 128‬ﻣﺸﺨﺺ ﮔﺮﺩﻳﺪ‪ ،‬ﺍﻣﺎ ﻫﻨﻮﺯ ﺩﺭ ﻣﻮﺭﺩ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺧﺎﺹ‬
‫‪ ± Inf‬ﻭ ‪ NaN‬ﺻﺤﺒﺘﻲ ﻧﺸﺪﻩ ﺍﺳﺖ‪ .‬ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺧﺎﺹ ﺑﺎﺯ ﻫﻢ ﺍﺯ ﻋﺪﺩ ﻣﻮﺟﻮﺩ ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۴‬‬

‫ﻛﻤﻚ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺗﺎﻛﻨﻮﻥ ﺍﺯ ﻣﻘﺪﺍﺭ ‪ 255‬ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻧﺸﺪﻩ ﻭ ﻟﺬﺍ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺍﻳﻦ ﻣﻘﺪﺍﺭ ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ‬
‫ﺍﻋﺪﺍﺩ ﺧﺎﺹ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﮔﺮ ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﻋﺪﺩ ‪ 255‬ﺫﺧﻴﺮﻩ ﺷﻮﺩ ﻭ ﻫﻤﭽﻨﻴﻦ ﺗﻤﺎﻣﻲ ﺍﺭﻗﺎﻡ ﻣﺎﻧﺘﻴﺲ‬
‫ﺻﻔﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺫﺧﻴﺮﻩ ﺷﺪﻩ ﺭﺍ ﺑﺴﺘﻪ ﺑﻪ ﻣﻘﺪﺍﺭ ﺑﻴﺖ ﻋﻼﻣﺖ ﺑﻪ ‪ +Inf‬ﻳﺎ ‪ −Inf‬ﺗﻌﺒﻴﺮ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﮔﺮ ﻋﺪﺩ ﺫﺧﻴﺮﻩ‬
‫ﺷﺪﻩ ﺩﺭ ‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ ﺑﺮﺍﺑﺮ ‪ 255‬ﺑﺎﺷﺪ ﻭ ﺣﺪﺍﻗﻞ ﻳﻚ ﺭﻗﻢ ﻣﺎﻧﺘﻴﺲ ﻳﻚ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﺑﻪ ﻋﻨﻮﺍﻥ ‪ NaN‬ﺗﻌﺒﻴﺮ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺟﻤﻊﺑﻨﺪﻱ‪ :‬ﺍﮔﺮ ﻣﺤﺘﻮﻱ ‪ ۳۲‬ﺑﻴﺖ ﺣﺎﻓﻈﻪﻱ ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ﻳﻚ ﻋﺪﺩ ﺩﺭ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺷﺪ‬
‫ﺑﻴﺖ ﻋﻼﻣﺖ‬

‫‪ ۸‬ﺑﻴﺖ ﻧﻤﺎ‬ ‫‪ ۲۳‬ﺑﻴﺖ ﻣﺎﻧﺘﻴﺲ‬

‫‪d−10‬‬
‫‪d−11‬‬
‫‪d−12‬‬
‫‪d−13‬‬
‫‪d−14‬‬
‫‪d−15‬‬
‫‪d−16‬‬
‫‪d−17‬‬
‫‪d−18‬‬
‫‪d−19‬‬
‫‪d−20‬‬
‫‪d−21‬‬
‫‪d−22‬‬
‫‪d−23‬‬
‫‪d−24‬‬
‫‪d−2‬‬
‫‪d−3‬‬
‫‪d−4‬‬
‫‪d−5‬‬
‫‪d−6‬‬
‫‪d−7‬‬
‫‪d−8‬‬
‫‪d−9‬‬
‫}‪ê ∈ {0, 1, . . . , 255‬‬
‫‪s‬‬

‫ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ﻋﺪﺩ ﻣﻮﺟﻮﺩ ﺩﺭ ﺣﺎﻓﻈﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺒﻴﺮ ﻣﻲﺷﻮﺩ‬

‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪.‬‬
‫‪(−1)s 0 0d−2 · · · d−24 × 2−125 ,‬‬ ‫‪ê = 0‬‬
‫‪‬‬
‫= ﻣﻘﺪﺍﺭ‬
‫‪.‬‬
‫‪(−1)s 0 1d−2 · · · d−24 × 2ê−126 ,‬‬ ‫‪1 ≤ ê ≤ 255‬‬
‫)‪(۵.۱‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪(−1)s Inf,‬‬ ‫‪.‬‬
‫‪ê = 255, 0. d−2 · · · d−24 = 0,‬‬
‫‪‬‬
‫‪NaN,‬‬ ‫‪ê = 255, 0..d‬‬ ‫‪−2‬‬ ‫‪· · · d−24 ̸= 0‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﭼﻮﻥ ﻋﺪﺩ ﺻﻔﺮ ﺟﺮﻭ ﺍﻋﺪﺍﺩ ﻧﻴﻤﻪﻧﺮﻣﺎﻝ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﺑﺮﺍﻱ ﻋﺪﺩ ‪ 0‬ﻧﻤﺎﻱ ‪ ê = 0‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻫﺮﭼﻨﺪ‬
‫ﻛﻪ ﭼﻮﻥ ﻋﺪﺩ ﺻﻔﺮ ﺩﺍﺭﺍﻱ ﻣﺎﻧﺘﻴﺲ ﺻﻔﺮ ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﻣﻘﺪﺍﺭ ﻧﻤﺎﻱ ﺁﻥ ﻫﺮ ﻋﺪﺩﻱ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺎﺷﺪ‪ .‬ﺻﻔﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻦ ﻧﻤﺎﻱ ﻋﺪﺩ ‪0‬‬
‫ﺩﺭ ﺍﻳﻦ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ‪ ،‬ﻣﻨﺠﺮ ﺑﻪ ﺳﻬﻮﻟﺖ ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻀﺎﻋﻒ ﻣﺸﺎﺑﻪ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﻣﻲﺑﺎﺷﺪ ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﺍﺯ ﺑﻴﺖﻫﺎﻱ ﺑﻴﺸﺘﺮﻱ ﺑﻪ ﻣﻨﻈﻮﺭ ﺣﺼﻮﻝ ﺑﻪ ﺩﻗﺖ‬
‫ﺑﺎﻻﺗﺮ ﺩﺭ ﻧﻤﺎﻳﺶ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﺖ‪ ۶۴ ،‬ﺑﻴﺖ )‪ ۸‬ﺑﺎﻳﺖ( ﺍﺧﺘﺼﺎﺹ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ‪ ۱‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻋﻼﻣﺖ‪ ۱۱ ،‬ﺑﻴﺖ‬
‫ﺑﺮﺍﻱ ﻧﻤﺎ ﻭ ‪ ۵۲‬ﺑﻴﺖ ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫‪۵.۱‬‬

‫ﻫﺪﻑ ﺍﺯ ﺍﻳﻦ ﺑﺨﺶ ﺁﺷﻨﺎﻳﻲ ﺑﺎ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﻭ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﺭﻳﺎﺿﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺳﺖ‪ .‬ﺍﻟﺒﺘﻪ ﻣﻄﺎﻟﻌﻪ ﺩﻗﻴﻖ‬
‫ﻧﺤﻮﻩ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﻣﺴﺘﻠﺰﻡ ﺁﺷﻨﺎﻳﻲ ﺑﺎ ﻣﻌﻤﺎﺭﻱ ﺳﺨﺖﺍﻓﺰﺍﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺳﺖ‪ ،‬ﻛﻪ ﺧﺎﺭﺝ ﺍﺯ ﺣﻮﺻﻠﻪ ﺍﻳﻦ ﺩﺭﺱ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ‬
‫ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺳﻌﻲ ﻣﻲﺷﻮﺩ ﺑﻪ ﺳﺎﺩﻩﺗﺮﻳﻦ ﺯﺑﺎﻥ ﻣﻤﻜﻦ‪ ،‬ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺗﺸﺮﻳﺢ ﺷﻮﺩ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺩﺍﻧﺶ ﻛﺎﻓﻲ‬
‫ﺑﺮﺍﻱ ﺗﺤﻠﻴﻞ ﻣﺤﺎﺳﺒﺎﺕ ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻓﺮﺍﻫﻢ ﮔﺮﺩﺩ‪.‬‬
‫ﺩﻭ ﻗﺴﻤﺖ ﺍﺻﻠﻲ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻭ ﻳﺎ ﻛﺎﻣﭙﻴﻮﺗﺮﻫﺎ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫• ﻭﺍﺣﺪ ﭘﺮﺩﺍﺯﻧﺪﻩ ﻣﺮﻛﺰﻱ ‪(CPU) ۳۵‬‬
‫• ﺣﺎﻓﻈﻪ ﺩﺍﺧﻠﻲ ﺑﺎ ﺩﺳﺘﺮﺳﻲ ﺗﺼﺎﺩﻓﻲ ‪(RAM) ۳۶‬‬
‫ﺍﺯ ‪ RAM‬ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﺑﺮﻧﺎﻣﻪﻫﺎ ﻭ ﺍﻋﺪﺍﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻓﻈﻪ ‪ RAM‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻳﻚ ﺭﺷﺘﻪ ﻃﻮﻻﻧﻲ ﺍﺯ ﺑﻴﺖﻫﺎ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‬
‫ﻛﻪ ﻫﺮ ﻫﺸﺖ ﺑﻴﺖ ﻣﺘﻮﺍﻟﻲ ﻳﻚ ﺑﺎﻳﺖ ﻧﺎﻣﮕﺬﺍﺭﻱ ﻣﻲﺷﻮﺩ‪ .‬ﻭﻗﺘﻲ ﻛﻪ ﻳﻚ ﻋﺪﺩ ﺑﺎ ﺩﻗﺖ ﻣﻌﻤﻮﻟﻲ ﺭﺍ ﺗﻌﺮﻳﻒ ﻭ ﺫﺧﻴﺮﻩ ﻣﻲﻛﻨﻴﻢ‪ ،‬ﺩﺭ‬
‫‪35 Centeral‬‬ ‫‪Processing Unit‬‬ ‫‪36 Random‬‬ ‫‪Access Memory‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۵‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻭﺍﻗﻊ ﺍﻳﻦ ﻋﺪﺩ ﺩﺭ ‪ ۳۲‬ﺑﻴﺖ ﻣﺘﻮﺍﻟﻲ ﺍﺯ ‪ RAM‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‪ CPU ،‬ﻭﻇﻴﻔﻪ ﭘﺮﺩﺍﺯﺵ ﻭ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﺑﻪ ﻋﻬﺪﻩ‬
‫ﺩﺍﺭﺩ‪ .‬ﻭﻗﺘﻲ ﺩﻭ ﻋﺪﺩ ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﺟﻤﻊ ﻣﻲﺷﻮﻧﺪ‪ ،‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻧﺠﺎﻡ ﻋﻤﻞ ﺟﻤﻊ ﺗﻮﺳﻂ ‪ CPU‬ﺑﻪ ﻭﺳﻴﻠﻪ ﻣﺪﺍﺭﻫﺎﻱ ﻣﻨﻄﻘﻲ ﺩﺍﺧﻞ‬
‫ﺁﻥ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻓﻈﻪ ‪ RAM‬ﻭ ‪ CPU‬ﺗﻮﺳﻂ ﻣﺴﻴﺮﻫﺎﻳﻲ ‪ ۳۷‬ﺑﻪ ﻳﻚﺩﻳﮕﺮ ﻣﺘﺼﻞ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺍﻣﻜﺎﻥ ﺍﻧﺘﻘﺎﻝ ﺍﻃﻼﻋﺎﺕ ﺑﻴﻦ‬
‫ﺁﻥﻫﺎ ﺭﺍ ﻓﺮﺍﻫﻢ ﻣﻲﻛﻨﺪ‪.‬‬
‫‪۳۸‬‬
‫ﺩﺭ ‪CPU‬ﻫﺎ ﺗﻌﺪﺍﺩﻱ ﺧﺎﻧﻪ ﺣﺎﻓﻈﻪ ﻣﻮﺳﻮﻡ ﺑﻪ ﺛﺒﺎﺕ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ ،‬ﻛﻪ ﺍﻃﻼﻋﺎﺕ ﻓﺮﺳﺘﺎﺩﻩ ﺷﺪﻩ ﺍﺯ ‪ RAM‬ﺩﺍﺧﻞ ﺁﻥﻫﺎ ﻗﺮﺍﺭ‬
‫ﻣﻲﮔﻴﺮﺩ‪ .‬ﺗﻌﺪﺍﺩ ﺛﺒﺎﺕﻫﺎﻱ ﺩﺍﺧﻞ ‪ CPU‬ﻛﻢ ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺍﻧﺪﺍﺯﻩ ﺁﻥﻫﺎ ﻧﺴﺒﺘﺎ ﺑﺰﺭﮒ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺛﺒﺎﺕﻫﺎ ﺑﻪ ﻣﺪﺍﺭﻫﺎﻱ ﻣﻨﻄﻘﻲ ﻣﻮﺟﻮﺩ ﺩﺭ‬
‫‪ CPU‬ﻣﺘﺼﻞ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺑﺎ ﻛﻤﻚ ﻣﺪﺍﺭﻫﺎﻱ ﻣﻨﻄﻘﻲ‪ ،‬ﺍﻣﻜﺎﻥ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺤﺘﻮﻳﺎﺕ ﺛﺒﺎﺕﻫﺎ ﻓﺮﺍﻫﻢ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫�‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻧﺎﻡﻫﺎﻱ ‪ x‬ﻭ ‪ y‬ﺩﺭ ﺣﺎﻓﻈﻪ ‪ RAM‬ﻳﻚ ﻣﺎﺷﻴﻦ ﻣﻨﻄﺒﻖ ﺑﺎ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ‬
‫ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ ﻛﺮﺩﻩ ﻭ ﺣﺎﺻﻞ ﺭﺍ ﺩﺭ ﺧﺎﻧﻪﺍﻱ ﺍﺯ ﺣﺎﻓﻈﻪ ‪ RAM‬ﺑﺎ ﻧﺎﻡ ‪ z‬ﻗﺮﺍﺭ ﺩﻫﻴﻢ‪ .‬ﻣﺮﺍﺣﻞ ﺯﻳﺮ ﺑﺮﺍﻱ ﺟﻤﻊ ﺩﻭ ﻋﺪﺩ‬
‫ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪:‬‬
‫ﺍﺑﺘﺪﺍ ﺩﻭ ﻋﺪﺩ ‪ x‬ﻭ ‪ y‬ﺑﻪ ﺩﺍﺧﻞ ‪ CPU‬ﻓﺮﺳﺘﺎﺩﻩ ﻣﻲﺷﻮﻧﺪ ﻭ ﺩﺭ ﺛﺒﺎﺕﻫﺎﻱ ‪ CPU‬ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﻧﺪ‪.‬‬ ‫‪.۱‬‬
‫ﺳﭙﺲ‪ ،‬ﻣﺪﺍﺭﻫﺎﻱ ﻣﻨﻄﻘﻲ ﺩﺍﺧﻞ ‪ ،CPU‬ﺩﻭ ﻋﺪﺩ ﻣﻮﺟﻮﺩ ﺩﺭ ﺛﺒﺎﺕﻫﺎ ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ ﻣﻲﻛﻨﻨﺪ ﻭ ﻧﺘﻴﺠﻪ ﺭﺍ ﺩﺍﺧﻞ ﻳﻚ ﺛﺒﺎﺕ‬ ‫‪.۲‬‬
‫ﻗﺮﺍﺭ ﻣﻲﺩﻫﻨﺪ‪ .‬ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ‪ ،‬ﺑﻪ ﻟﺤﺎﻅ ﺗﺌﻮﺭﻱ‪ ،‬ﺍﻳﻦ ﺟﻤﻊ ﺩﺭ ﺩﺍﺧﻞ ‪ CPU‬ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻭ ﻃﻮﻝ‬
‫ﺛﺒﺎﺕﻫﺎ ﺑﻪ ﺍﻧﺪﺍﺯﻩﺍﻱ ﺑﺰﺭﮒ ﺍﺳﺖ ﻛﻪ ﺣﺎﺻﻞﺟﻤﻊ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ ﺩﺭ ﺛﺒﺎﺕ ﺟﺎ ﺑﮕﻴﺮﺩ‪.‬‬
‫ﺣﺎﺻﻞﺟﻤﻊ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ ﺩﺭ ﺛﺒﺎﺕ ﺭﻭﻧﺪ ﻣﻲﺷﻮﺩ‪.‬‬ ‫‪.۳‬‬
‫ﻧﺘﻴﺠﻪ ﺭﻭﻧﺪ ﺷﺪﻩ ﺑﻪ ‪ RAM‬ﻣﺘﻨﻘﻞ ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﻳﻚ ﻗﺴﻤﺖ ﺍﺯ ﺁﻥ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪.‬‬ ‫‪.۴‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻫﺮﭼﻨﺪ ‪ x‬ﻭ ‪ y‬ﺍﻋﺪﺍﺩﻱ ﻣﺎﺷﻴﻨﻲ ﻫﺴﺘﻨﺪ‪ ،‬ﺍﻣﺎ ﺣﺎﺻﻞﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x‬ﻭ ‪ y‬ﻟﺰﻭﻣﺎ ﻋﺪﺩﻱ ﻣﺎﺷﻴﻨﻲ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ ،‬ﻋﻤﻞ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺍﻧﺠﺎﻡ ﻣﻲﮔﺮﺩﺩ ﺗﺎ ﺣﺎﺻﻞ ‪ x + y‬ﺑﻪ ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺗﺒﺪﻳﻞ ﮔﺮﺩﺩ ﻭ ﺳﭙﺲ ﺑﺘﻮﺍﻥ‬
‫ﺁﻥ ﺭﺍ ﺩﺭ ‪ RAM‬ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻛﺮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۹−۱‬ﺟﻤ ﺩﻭ ﻋﺪﺩ ﺩﺭ ﻣﺎﺷﯿﻨ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ )‪( FS+ (10, 3, −2, 2‬‬

‫ﻳﻚ ﻣﺎﺷﻴﻦ ﺑﺎ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪ FS+ (10, 3, −2, 2‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺩﻭ ﻋﺪﺩ ‪ x1 = 0.123 × 101‬ﻭ‬
‫‪ x2 = 0.456 × 10−1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻫﺴﺘﻨﺪ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﺳﺘﻮﺭ ‪z = x1 + x2‬‬
‫ﺭﺍ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺍﺟﺮﺍ ﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﻭ ﻋﺪﺩ ‪ x‬ﻭ ‪ y‬ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ ﻭ ﺣﺎﺻﻞ ﺭﺍ ﺩﺭ ‪ z‬ﺫﺧﻴﺮﻩ ﻧﻤﺎﻳﻴﻢ‪.‬‬

‫• ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ ،‬ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﺑﻪ ‪ CPU‬ﻣﺎﺷﻴﻦ ﻓﺮﺳﺘﺎﺩﻩ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫• ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﺣﺎﺻﻞﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﻣﺤﺎﺳﺒﻪ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺣﺎﺻﻞﺟﻤﻊ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪x1 + x2 = 0.123 × 101 + 0.456 × 10−1‬‬
‫‪= 0.123 × 101 + 0.00456 × 101‬‬
‫‪= 0.12756 × 101 ,‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻋﺪﺩ ‪ 0.12756 × 10+1‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞﺟﻤﻊ ﺩﺭ ﻳﻚ ﺛﺒﺎﺕ ﺍﺯ ‪ CPU‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫• ﺩﺭ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ ،‬ﻧﺘﻴﺠﻪ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺭﻭﻧﺪ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﺻﻞﺟﻤﻊ ﻓﻮﻕ ﻋﺪﺩﻱ ﺩﺭ ﻣﺠﻤﻮﻋﻪ )‪FS+ (10, 3, −2, 2‬‬

‫‪37 Bus‬‬ ‫‪38 Register‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۶‬‬

‫ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﻣﺎﺷﻴﻨﻲ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺍﻳﻦ ﻋﺪﺩ ﺭﺍ ﺑﻪ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ 0.128 × 10−1‬ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪.‬‬
‫• ﺩﺭ ﻣﺮﺣﻠﻪ ﭼﻬﺎﺭﻡ‪ ،‬ﻋﺪﺩ ﺭﻭﻧﺪ ﺷﺪﻩ ‪ ،0.128 × 10−1‬ﺑﻪ ﺣﺎﻓﻈﻪ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺑﺮﺍﻱ ‪ z‬ﺩﺭ ‪ RAM‬ﻣﻨﺘﻘﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﺍﺟﺮﺍﻱ ﺩﺳﺘﻮﺭ ‪ z = x + y‬ﻣﻘﺪﺍﺭ ‪ 0.128 × 101‬ﺑﻪ ﺟﺎﻱ ‪ 0.12756 × 101‬ﺩﺭ ﻣﺘﻐﻴﺮ ‪ z‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻭﺍﺿﺢ‬
‫ﺍﺳﺖ ﻛﻪ ﺣﺎﺻﻞﺟﻤﻊ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺑﺎ ﺣﺎﺻﻞﺟﻤﻊ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻛﺮﺩﻳﺪ‪ ،‬ﺍﻧﺠﺎﻡ ﺟﻤﻊ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﻪ ﺻﻮﺭﺗﻲ ﻣﺘﻔﺎﻭﺕ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﻣﻌﻤﻮﻻ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ‬
‫ﻧﺘﻴﺠﻪ ﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺑﺮﺍﻱ ﺗﻤﺎﻳﺰ ﺑﻴﻦ ﺟﻤﻊ ﻭﺍﻗﻌﻲ ﻭ ﺟﻤﻊ ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬
‫ﺍﺯ ﻭﺍﮊﻩ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺍﺯ ﻧﻤﺎﺩ ⊕ ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺩﺭ‬
‫ﻣﺜﺎﻝ ‪ ۹-۱‬ﺟﻤﻊ ﻭﺍﻗﻌﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻣﻲﺷﻮﺩ‬
‫‪x + y = 0.12756 × 101 ,‬‬
‫ﺍﻣﺎ ﺟﻤﻊ ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺗﻮﺳﻂ ﺁﻥ ﻣﺎﺷﻴﻦ )ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ( ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻣﻲﺷﻮﺩ‬
‫‪x ⊕ y = 0.128 × 101 .‬‬

‫ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﺗﻘﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺩﻗﻴﻘﺎ ﺑﻪ ﺻﻮﺭﺕ ﻣﺸﺎﺑﻪ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﻤﺎﺩﻫﺎﻱ ⊖‪ ⊛ ،‬ﻭ ⊘ ﺑﺮﺍﻱ‬
‫ﺍﺷﺎﺭﻩ ﺑﻪ ﺗﻔﺮﻳﻖ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﺿﺮﺏ ﻣﺎﺷﻴﻨﻲ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺭ ﺍﻧﺘﻬﺎ‪ ،‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻧﻤﺎﺩﻫﺎﻱ ﻣﻌﺮﻓﻲ ﺷﺪﻩ‪ ،‬ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﺑﺨﺶ ﺭﺍ ﺩﺭ ﺭﻭﺍﺑﻂ ﺯﻳﺮ ﺧﻼﺻﻪ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪x ⊕ y := rd(x + y),‬‬ ‫)‪(۶.۱‬‬
‫‪x ⊖ y := rd(x − y),‬‬ ‫)‪(۷.۱‬‬
‫‪x ⊛ y := rd(x ∗ y),‬‬ ‫)‪(۸.۱‬‬
‫‪x ⊘ y := rd(x/y).‬‬ ‫)‪(۹.۱‬‬
‫ﺭﺍﺑﻄﻪ ﺍﻭﻝ ﺩﺭ ﻓﻮﻕ ﺍﻳﻦﮔﻮﻧﻪ ﺗﻔﺴﻴﺮ ﻣﻲﺷﻮﺩ ﻛﻪ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x‬ﻭ ‪ y‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺭﻭﻧﺪﺷﺪﻩ ﺟﻤﻊ ﺩﻗﻴﻖ ‪ x‬ﻭ‬
‫‪.y‬‬

‫ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﺭﻳﺎﺿﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫�‬

‫ﭘﺮﺩﺍﺯﻧﺪﻩ ﻣﺎﺷﻴﻦﻫﺎ ﻗﺎﺩﺭ ﻧﻴﺴﺘﻨﺪ ﻛﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ ﺷﺎﻣﻞ ﺳﻪ ﻳﺎ ﭼﻨﺪ ﻋﻤﻞ ﺭﺍ ﺑﻪ ﻃﻮﺭ ﻳﻜﺠﺎ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻨﺪ‪ .‬ﺑﻠﻜﻪ ﺩﺭ ﺍﺑﺘﺪﺍ ﺑﺎﻳﺪ ﺁﻥ‬
‫ﻋﺒﺎﺭﺕ ﺑﻪ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺍﺯ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻭﺗﺎﻳﻲ ﺗﺒﺪﻳﻞ ﺷﻮﺩ ﻭ ﺳﭙﺲ ﺍﻳﻦ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻭﺗﺎﻳﻲ ﺑﻪ ﻃﻮﺭ ﭘﻲ ﺩﺭ ﭘﻲ ﺩﺭ ‪ CPU‬ﻣﺎﺷﻴﻦ ﺍﻧﺠﺎﻡ‬
‫ﺷﻮﺩ‪.‬‬

‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻋﺒﺎﺭﺕ‬

‫‪x∗y‬‬
‫=‪z‬‬ ‫)‪(۱۰.۱‬‬
‫‪1 + x2‬‬
‫ﺭﺍ ﺩﺭ ﻣﺎﺷﻴﻦ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ x‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ‪ RAM‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺩﻧﺒﺎﻟﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۷‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪:‬‬

‫;‪z1 = x ∗ y‬‬
‫;‪z2 = x ∗ x‬‬
‫; ‪z3 = 1 + z2‬‬
‫; ‪z = z1 /z3‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ z‬ﻧﻴﺎﺯ ﺑﻪ ﺍﻧﺠﺎﻡ ﻳﻚ ﺟﻤﻊ‪ ،‬ﺩﻭ ﺿﺮﺏ ﻭ ﻳﻚ ﺗﻘﺴﻴﻢ ﺩﺍﺭﻳﻢ‪ .‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‪ ،‬ﻣﺤﺎﺳﺒﻪ‬
‫ﻋﺒﺎﺭﺕ )‪ (۱۰.۱‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻧﺠﺎﻡ ﻣﻲﮔﺮﺩﺩ‬
‫;‪z1 = x ⊛ y‬‬
‫;‪z2 = x ⊛ x‬‬
‫; ‪z3 = 1 ⊕ z2‬‬
‫; ‪z = z1 ⊘ z3‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻫﺮ ﻛﺪﺍﻡ ﺍﺯ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻭﺩﻭﻳﻲ ﻣﺎﺷﻴﻨﻲ ﻓﻮﻕ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺕ‬
‫ﻓﻮﻕ ﺩﺭ ﻣﺎﺷﻴﻦ ﺑﺎ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﺁﻥ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﻣﻘﺪﺍﺭ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺗﻮﺳﻂ ﻣﺎﺷﻴﻦ ﺍﺯ ﻧﻤﺎﺩ‬
‫‬ ‫‬
‫‪x∗y‬‬
‫‪fl‬‬
‫‪1 + x2‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺭﺍﺑﻄﻪ ﺑﺮﺍﻱ ﻏﺎﻟﺐ ﺍﻋﺪﺍﺩ ﺩﺍﺭﻳﻢ‪:‬‬
‫‬ ‫‬ ‫‬ ‫‬
‫‪x∗y‬‬ ‫‪x∗y‬‬
‫‪fl‬‬ ‫‪̸= rd‬‬
‫‪1 + x2‬‬ ‫‪1 + x2‬‬
‫ﺩﺭ ﻭﺍﻗﻊ ﺩﺍﺭﻳﻢ‪:‬‬
‫‬ ‫‬
‫‪x∗y‬‬
‫‪fl‬‬ ‫‪= (x ⊛ y) ⊘ (1 ⊕ (x ⊛ x)) ,‬‬
‫‪1 + x2‬‬
‫‪=rd (rd(x ∗ y)/rd (1 + rd(x ∗ x))) .‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﺑﺎ ﺍﺭﺍﻳﻪ ﻣﺜﺎﻝﻫﺎﻳﻲ ﺑﻪ ﺗﺸﺮﻳﺢ ﺑﻴﺸﺘﺮ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ )ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ( ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۰−۱‬ﻣﺤﺎﺳﺒﻪ ﺣﺠﻢ ﮐﺮﻩ ﺑﺎ ﺷﻌﺎﻉ ‪ 1.2‬ﺩﺭ ﻣﺎﺷﯿﻦ )‪( FS+ (10, 2, −1, 1‬‬

‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺣﺠﻢ ﻛﺮﻩ ﺍﺯ ﻓﺮﻣﻮﻝ ‪ 43 πr3‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻟﺬﺍ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺣﺠﻢ ﻳﻚ ﻛﺮﻩ ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﺍﺯ ﻋﺒﺎﺭﺕ ﺯﻳﺮ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‬
‫‪4 3‬‬
‫=‪v‬‬ ‫‪πr‬‬
‫‪3‬‬
‫ﻛﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺩﻧﺒﺎﻟﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻭﺩﻭﻳﻲ ﺯﻳﺮ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‬
‫‪v1 = r ⊛ r,‬‬
‫‪v2 = v1 ⊛ r,‬‬
‫‪v3 = 4 ⊘ 3,‬‬
‫‪v4 = v3 ∗ rd(π),‬‬
‫‪v = v4 ⊛ v2 .‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۸‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﻣﺤﺎﺳﺒﻪ ﺣﺠﻢ ﻛﺮﻩﺍﻱ ﺑﺎ ﺷﻌﺎﻉ ‪ r = 1.2‬ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS+ (10, 2, −1, 1‬ﺷﺒﻴﻪﺳﺎﺯﻱ ﻣﻲﺷﻮﺩ‬
‫
‬ ‫

‬
‫‪v1 = r ⊛ r = rd(r ∗ r) = rd‬‬ ‫‪0.12 × 101 ∗ 0.12 × 101‬‬
‫
‬
‫‪= rd 0.144 × 10+1 = 0.14 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v2 = v1 ⊛ r = rd(v1 ∗ r) = rd 0.14 × 101 ∗ 0.12 × 101‬‬
‫
‬
‫‪= rd 0.168 × 10+1 = 0.17 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v3 = 4 ⊘ 3 = rd(4/3) = rd 0.4 × 101 / 0.3 × 101‬‬
‫
‬
‫‪= rd 0.13333 · · · × 10+1 = 0.13 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v4 = v3 ∗ rd(π) = rd(v3 ∗ rd(π)) = rd 0.13 × 101 ∗ 0.31 × 101‬‬
‫
‬
‫‪= rd 0.403 × 10+1 = 0.40 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v = v4 ⊛ v2 = rd(v4 ∗ v2 ) = rd 0.40 × 101 ∗ 0.17 × 101‬‬
‫
‬
‫‪= rd 0.68 × 10+1 = 0.68 × 10+1 .‬‬
‫ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‪ ،‬ﺣﺠﻢ ﻛﺮﻩ ﺑﺮﺍﺑﺮ ‪ 6.8‬ﻣﺤﺎﺳﺒﻪ ﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺩﻗﻴﻖ ﺣﺠﻢ ﻛﺮﻩ ﺑﺮﺍﺑﺮ · · · ‪ 7.23822947387088‬ﺍﺳﺖ ﻛﻪ ﻣﻼﺣﻈﻪ‬
‫ﻣﻲﺷﻮﺩ‪ ،‬ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﺍﺯ ﻣﺎﺷﻴﻦ ﺑﺎ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﺗﻔﺎﻭﺕ ﻧﺴﺒﺘﺎ ﺯﻳﺎﺩﻱ ﺩﺍﺭﺩ‪.‬‬

‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪ ،FS+ (β, m, emin , emax‬ﻫﺮﭼﻪ ‪) m‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﻣﺎﻧﺘﻴﺲ( ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ‬
‫ﺩﻗﻴﻖﺗﺮ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﺣﺠﻢ ﻛﺮﻩ ﺑﺎ ﺷﻌﺎﻉ ‪ r = 1.2‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ )‪ FS+ (10, 4, −1, 1‬ﻭ‬
‫)‪ FS+ (10, 8, −1, 1‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﺮﺍﺑﺮ ‪ 7.237‬ﻭ ‪ 7.2382295‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ ﻛﻪ ﺑﻪ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ · · · ‪7.23822947387088‬‬
‫ﻧﺰﺩﻳﻚﺗﺮﻧﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﻣﻘﺪﺍﺭ ‪ m‬ﺑﺮ ﺩﻗﺖ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺗﺎﺛﻴﺮ ﻣﻲﮔﺬﺍﺭﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺗﺎﺛﻴﺮ ‪ emax‬ﺑﺮ ﺳﺮﺭﻳﺰﻱ‬
‫ﻭ ﻧﺘﻴﺠﻪ ﺩﺍﺩﻥ ‪ Inf‬ﻭ ‪ NaN‬ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺳﺮﺭﻳﺰﻱ ﻭ ﺣﺎﻻﺕ ﻣﺒﻬﻢ ﻭ ﻧﺎﻣﻤﻜﻦ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﺭﻳﺎﺿﻲ‬

‫ﻣﻤﻜﻦ ﺍﺳﺖ‪ ،‬ﻫﻨﮕﺎﻡ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺕﻫﺎﻱ ﺭﻳﺎﺿﻲ ﺳﺮﺭﻳﺰﻱ ﻭ ﻳﺎ ﺣﺎﻟﺖﻫﺎﻱ ﻣﺒﻬﻢ ﻭ ﻧﺎﻣﻤﻜﻦ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ‪،‬‬
‫ﻣﺎﺷﻴﻦﻫﺎ ﻣﻲﺗﻮﺍﻧﻨﺪ ﺑﻪ ﺩﻭ ﺭﻭﻳﻜﺮﺩ ﻣﺘﻔﺎﻭﺕ ﺭﻓﺘﺎﺭ ﻛﻨﻨﺪ‪ .‬ﺩﺭ ﺭﻭﻳﻜﺮﺩ ﺍﻭﻝ‪ ،‬ﻣﺎﺷﻴﻦﻫﺎ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﻣﺘﻮﻗﻒ ﻣﻲﻛﻨﻨﺪ ﻭ ﻳﻚ ﭘﻴﺎﻡ ﺧﻄﺎ‬
‫ﻣﻲﺩﻫﻨﺪ‪ .‬ﺩﺭ ﺭﻭﻳﻜﺮﺩ ﺩﻭﻡ‪) ،‬ﻛﻪ ﺩﺭ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﻧﻴﺰ ﺭﻋﺎﻳﺖ ﻣﻲﺷﻮﺩ( ﻫﻨﮕﺎﻡ ﺳﺮﺭﻳﺰﻱ ﻣﻘﺪﺍﺭ ‪ ± Inf‬ﻭ ﻫﻨﻜﺎﻡ ﺣﺎﻻﺕ ﻣﺒﻬﻢ‬
‫ﻭ ﻧﺎﻣﻤﻜﻦ‪ ،‬ﻣﻘﺪﺍﺭ ‪ NaN‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﺑﺎ ﻫﻤﻴﻦ ﻣﻘﺎﺩﻳﺮ ﺍﺩﺍﻣﻪ ﭘﻴﺪﺍ ﻣﻲﻛﻨﺪ‪ .‬ﻗﻮﺍﻋﺪ ﻣﺤﺎﺳﺒﺎﺕ ﺭﻭﻱ ‪ Inf‬ﻭ ‪ NaN‬ﺑﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۹‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬

‫‪) ⊘ 0 = +Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻣﺜﺒﺖ(‬ ‫‪) ⊘ 0 = −Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻣﻨﻔﻲ(‬
‫‪0 ⊘ 0 = NaN,‬‬ ‫‪) ⊘ Inf = 0,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ(‬
‫‪) = Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ( ⊕ ‪Inf‬‬ ‫‪) ⊖ Inf = −Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ(‬
‫‪) ⊛ Inf = Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻣﺜﺒﺖ(‬ ‫‪) ⊛ Inf = −Inf,‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻣﻨﻔﻲ(‬
‫‪Inf ⊕ Inf = Inf,‬‬ ‫‪Inf ⊛ Inf = Inf,‬‬
‫‪Inf ⊖ Inf = NaN,‬‬ ‫‪Inf ⊘ Inf = NaN,‬‬
‫‪0 ⊛ Inf = NaN,‬‬ ‫‪Inf ⊘ 0 = Inf,‬‬
‫‪NaN ⊕ (±Inf) = NaN,‬‬ ‫‪NaN ⊛ (±Inf) = NaN,‬‬

‫ﻣﺜﺎﻝ ‪)۱۱−۱‬ﺳﺮﺭﯾﺰﯼ ﻫﻨﮕﺎﻡ ﻣﺤﺎﺳﺒﻪ ﺣﺠﻢ ﮐﺮﻩ ﺑﺎ ﺷﻌﺎﻉ ‪ 2.5‬ﺩﺭ ﻣﺎﺷﯿﻦ )‪( FS+ (10, 2, −1, 1‬‬

‫ﻫﻤﺎﻧﻨﺪ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﻣﺤﺎﺳﺎﺕ ﺣﺠﻢ ﻛﺮﻩﺍﻱ ﺑﺎ ﺷﻌﺎﻉ ‪ r = 2.5‬ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS+ (10, 2, −1, 1‬ﺷﺒﻴﻪﺳﺎﺯﻱ ﻣﻲﺷﻮﺩ‬
‫
‬ ‫

‬
‫‪v1 = r ⊛ r = rd(r ∗ r) = rd‬‬ ‫‪0.25 × 101 ∗ 0.25 × 101‬‬
‫
‬
‫‪= rd 0.625 × 10+1 = 0.63 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v2 = v1 ⊛ r = rd(v1 ∗ r) = rd 0.63 × 101 ∗ 0.25 × 101‬‬
‫
‬
‫‪,‬ﺳﺮﺭﻳﺰﻱ = ‪= rd 0.1625 × 10+2‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ‪ ،‬ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﺩﺭﺻﻮﺭﺗﻲ ﻛﻪ ﺭﻭﻳﻜﺮﺩ ﻣﺎﺷﻴﻦ ﻫﻨﮕﺎﻡ ﺳﺮﺭﻳﺰﻱ‪ ،‬ﺗﻮﻗﻒ‬
‫ﻣﺤﺎﺳﺒﺎﺕ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻫﻨﮕﺎﻡ ﻣﺤﺎﺳﺒﻪ ‪ v2‬ﻣﺎﺷﻴﻦ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﻣﺘﻮﻗﻒ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺭﻭﻳﻜﺮﺩ ﻣﺎﺷﻴﻦ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ‪ Inf‬ﻫﻨﮕﺎﻡ‬
‫ﺳﺮﺭﻳﺰﻱ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺘﻐﻴﺮ ‪ v2‬ﺑﺎ ‪ Inf‬ﻣﻘﺪﺍﺭﺩﻫﻲ ﻣﻲﺷﻮﺩ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺩﺍﻣﻪ ﭘﻴﺪﺍ ﻣﻲﻛﻨﺪ‪:‬‬
‫‪v2 = +Inf,‬‬
‫
‬ ‫

‬
‫‪v3 = 4 ⊘ 3 = rd(4/3) = rd‬‬ ‫‪0.4 × 101 / 0.3 × 101‬‬
‫
‬
‫‪= rd‬‬ ‫‪0.13333 · · · × 10+1 = 0.13 × 10+1 ,‬‬
‫
‬ ‫

‬
‫‪v4 = v3 ∗ rd(π) = rd(v3 ∗ rd(π)) = rd‬‬ ‫‪0.13 × 101 ∗ 0.31 × 101‬‬
‫
‬
‫‪= rd‬‬ ‫‪0.403 × 10+1 = 0.40 × 10+1 ,‬‬
‫
‬ ‫
‬
‫‪v = v4 ⊛ v2 = rd(v4 ∗ v2 ) = rd‬‬ ‫)‪0.40 × 101 ∗ (+Inf‬‬
‫‪= +Inf.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ ﺍﮔﺮ ﺍﺯ ﻣﺎﺷﻴﻦ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪ FS+ (10, 2, −1, 2‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﻳﮕﺮ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺎﺩ‬
‫ﻭ ﻣﻘﺪﺍﺭ ‪ Inf‬ﺣﺎﺻﻞ ﻧﻤﻲﺷﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۳۰‬‬

‫ﻣﺜﺎﻝ ‪ NaN)۱۲−۱‬ﺷﺪﻥ ﻧﺘﯿﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ(‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺣﺎﺻﻞ ﻋﺒﺎﺭﺕ‬
‫‪2‬‬
‫‪8.2‬‬
‫=‪z‬‬
‫‪4.12‬‬
‫ﺭﺍ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS+ (10, 2, −1, 1‬ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ‪ .‬ﺣﺎﺻﻞ ﺩﻗﻴﻖ ﺍﻳﻦ ﻋﺒﺎﺭﺕ ﺑﺮﺍﺑﺮ ‪ 4‬ﺍﺳﺖ‪ .‬ﻣﺮﺍﺣﻞ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺩﺭ ﺍﻳﻦ‬
‫ﻣﺎﺷﻴﻦ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪z1 = rd(8.2) = 0.82 × 101‬‬
‫‪z2 = rd(4.1) = 0.41 × 101‬‬
‫
‬
‫) ‪z3 = z1 ⊛ z1 = rd (0.82 × 10+1 ) ∗ (0.82 × 10+1‬‬
‫‪= rd(0.6724 × 10+2 ) = +Inf‬‬
‫
‬
‫) ‪z4 = z2 ⊛ z2 = rd (0.41 × 10+1 ) ∗ (0.41 × 10+1‬‬
‫‪= rd(0.1681 × 10+2 ) = +Inf‬‬
‫‪z = z4 ⊘ z4 = Inf ⊘ Inf = NaN‬‬
‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ﻣﺎﺷﻴﻦ ﻣﺒﺘﻨﻲ ﺑﺮ )‪ FS+ (10, 2, −1, 2‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺘﻴﺠﻪ‬
‫‪ NaN‬ﻧﺨﻮﺍﻫﺪ ﺷﺪ‪.‬‬

‫ﺗﻔﺎﻭﺕﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ‬ ‫�‬

‫ﻫﻤﺎﻧﻄﻮﺭ ﻛﻪ ﻣﻼﺣﻈﻪ ﺷﺪ‪ ،‬ﻣﻌﻤﻮﻻ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺗﻔﺎﻭﺕﻫﺎ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ‬
‫ﻛﻪ ﺑﺮﺧﻲ ﺍﺯ ﺍﺻﻮﻝ ﻭ ﻗﻮﺍﻋﺪ ﺭﻳﺎﺿﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻗﺮﺍﺭ ﻧﺒﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﺍﻳﻦ ﺗﻔﺎﻭﺕﻫﺎ ﺍﺷﺎﺭﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﻧﺒﻮﺩﻥ ﻋﻀﻮ ﺑﻲﺍﺛﺮ ﺟﻤﻊ‬

‫ﻋﺪﺩ ‪ 0‬ﺗﻨﻬﺎ ﻋﻨﺼﺮ ﺑﻲﺍﺛﺮ ﺟﻤﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ‪ x + y = x‬ﺍﮔﺮ‬ ‫ﺣﻘﻴﻘﻲ ‪،۳۹‬‬
‫ﻣﻄﺎﺑﻖ ﺑﺎ ﺍﺻﻮﻝ ﻣﻮﺿﻮﻋﻪ ﺩﺳﺘﮕﺎﻩ ﺍﻋﺪﺍﺩ‬
‫ﻭ ﻓﻘﻂ ﺍﮔﺮ ‪ .y = 0‬ﺍﻳﻦ ﺧﺎﺻﻴﺖ ﺩﺭ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﺭ )‪ FS (10, 3, −5, 5‬ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬
‫‪ x = 0.123 × 103‬ﻭ ‪ y = 0.123 × 10−3‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺩﺍﺭﻳﻢ ﻛﻪ‬
‫‪x ⊕ y = 0.123 × 103 = x.‬‬
‫ﻣﻌﻤﻮﻻ‪ ،‬ﺍﻳﻦ ﺍﺗﻔﺎﻕ ﺩﺭ ﺟﻤﻊ ﻳﻚ ﻋﺪﺩ ﺑﺎ ﺍﻧﺪﺍﺯﻩ ﺑﺰﺭﮒ ﻭ ﻳﻚ ﻋﺪﺩ ﺑﺎ ﺍﻧﺪﺍﺯﻩ ﻛﻮﭼﻚ ﺭﺥ ﻣﻲﺩﻫﺪ‪.‬‬

‫■ ﺑﺮﻗﺮﺍﺭ ﻧﺒﻮﺩﻥ ﻗﺎﻋﺪﻩ ﺷﺮﻛﺖﭘﺬﻳﺮﻱ ﺟﻤﻊ‬

‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﺩﺍﺭﺍﻱ ﺧﺎﺻﻴﺖ ﺟﺎﺑﺠﺎﻳﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﻛﺮﺩﻥ ﺩﻭ ﻋﺪﺩ ﺍﻫﻤﻴﺘﻲ ﻧﺪﺍﺭﺩ‪ .‬ﺩﺭ ﻣﻮﺭﺩ ﺳﻪ ﻳﺎ‬
‫ﺑﻴﺸﺘﺮ ﺍﺯ ﺳﻪ ﻋﺪﺩ ﻧﻴﺰ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﻛﺮﺩﻥ ﺍﻫﻤﻴﺘﻲ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻳﻦ ﺧﺎﺻﻴﺖ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺧﺎﺻﻴﺖ ﺷﺮﻛﺖﭘﺬﻳﺮﻱ ‪ ۴۰‬ﻣﻲﺷﻨﺎﺧﺘﻴﻢ‪ ،‬ﻛﻪ ﻳﻜﻲ‬
‫ﺩﻳﮕﺮ ﺍﺯ ﺍﺻﻮﻝ ﻣﻮﺿﻮﻋﻪ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺧﺎﺻﻴﺖ ﺷﺮﻛﺖﭘﺬﻳﺮﻱ ﻧﻴﺰ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﺳﻪ‬
‫ﻳﺎ ﭼﻨﺪ ﻋﺪﺩ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻪ ﻧﺘﺎﻳﺞ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻨﺠﺮ ﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪FS (10, 8, −5, 5‬‬
‫ﺳﻪ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ y = 0.33678429 × 102 ،x = 0.23371258 × 10−4‬ﻭ ‪ z = −0.33677811 × 102‬ﺭﺍ ﺩﺭ‬
‫‪39 Axioms‬‬ ‫‪for real numbers‬‬ ‫‪40 Associative‬‬ ‫‪law‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۳۱‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬

‫‪fl (x + (y + z)) = x ⊕ (y ⊕ z) = 0.64137126 × 10−3 ,‬‬
‫‪fl ((x + y) + z) = (x ⊕ y) ⊕ z = 0.64100000 × 10−3 .‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﻣﺎﺷﻴﻦ ﻓﻮﻕ‪ ،‬ﺗﺮﺗﻴﺐ ﻣﺘﻔﺎﻭﺕ ﺟﻤﻊ ﺳﻪ ﻋﺪﺩ ﻣﻨﺠﺮ ﺑﻪ ﻧﺘﺎﻳﺞ ﻣﺘﻔﺎﻭﺗﻲ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ‪.‬‬

‫■ ﺑﺮﻗﺮﺍﺭ ﻧﺒﻮﺩﻥ ﺑﺮﺧﻲ ﺍﺯ ﻗﻮﺍﻋﺪ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬

‫ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ‪ ،‬ﻗﻮﺍﻋﺪ ﺯﻳﺮ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﻨﺪ‪:‬‬

‫‪ : x ⊛ (y ⊛ z) = (x ⊛ y) ⊛ z‬ﻗﺎﻋﺪﻩ ﺷﺮﻛﺖﭘﺬﻳﺮﻱ ﺿﺮﺏ‬
‫‪ : x ⊛ y = x ⊛ z, x ̸= 0 ⇒ y = z‬ﻗﺎﻋﺪﻩ ﺣﺬﻑ ﺩﺭ ﺿﺮﺏ‬
‫)‪ : x ⊛ (y ⊕ z) = (x ⊛ y) ⊕ (x ⊛ z‬ﻗﺎﻋﺪﻩ ﭘﺨﺶﭘﺬﻳﺮﻱ‬
‫‪ : x ⊛ (y ⊘ x) = y‬ﺧﻨﺜﻲ ﺷﺪﻥ ﺗﻘﺴﻴﻢ ﺑﺎ ﺿﺮﺏ‬
‫ﺍﻟﺒﺘﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻳﻦ ﻗﻮﺍﻋﺪ ﺑﺮﺍﻱ ﺑﺮﺧﻲ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭﺳﺖ ﺑﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ﺍﻳﻦ ﻗﻮﺍﻋﺪ ﺑﺮﺍﻱ ﺗﻤﺎﻣﻲ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺭﺳﺖ ﻧﻤﻲﺑﺎﺷﻨﺪ‪.‬‬

‫■ ﺑﺮﻗﺮﺍﺭ ﻧﺒﻮﺩﻥ ﺍﺗﺤﺎﺩﻫﺎﻱ ﺟﺒﺮﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬

‫ﺩﺭ ﺭﻳﺎﺿﻴﺎﺕ ﺍﺗﺤﺎﺩﻫﺎﻱ ﺯﻳﺎﺩﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﺯ ﺁﻥﻫﺎ ﺑﺮﺍﻱ ﺗﺤﻠﻴﻞ ﻭ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﺍﺗﺤﺎﺩ )‪ a2 − b2 = (a − b)(a + b‬ﺍﺷﺎﺭﻩ ﻛﺮﺩ ﻛﻪ ﻃﺒﻖ ﺁﻥ ﻋﺒﺎﺭﺕﻫﺎﻱ ‪ a2 − b2‬ﻭ )‪ (a − b)(a + b‬ﻣﻌﺎﺩﻝ ﻫﺴﺘﻨﺪ‪ .‬ﺍﻣﺎ‬
‫ﺍﻳﻦ ﺩﻭ ﻋﺒﺎﺭﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻟﺰﻭﻣﺎ ﻣﻌﺎﺩﻝ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﻣﻘﺪﺍﺭ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ ﺍﻳﻦ ﺩﻭ ﻋﺒﺎﺭﺕ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻲﺗﻮﺍﻧﺪ‬
‫ﻣﺘﻔﺎﻭﺕ ﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﻧﺸﺎﻥ ﺩﺍﺩﻥ ﺍﻳﻦ ﺗﻔﺎﻭﺕ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 3, −2, 3‬ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﺩﻭ ﻋﺒﺎﺭﺕ ﺭﺍ ﺑﻪ ﺍﺯﺍﻱ ‪ a = 6.21‬ﻭ‬
‫‪ b = 5.67‬ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﻣﻘﺪﺍﺭ ﻣﺎﺷﻴﻨﻲ ﺩﻭ ﻋﺒﺎﺭﺕ ‪ a2 − b2‬ﻭ )‪ (a − b)(a + b‬ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫
‬
‫‪fl a2 − b‬‬ ‫‪2‬‬
‫))‪= (a ⊛ a) ⊖ (b ⊛ b) = rd (rd(a ∗ a) − rd(b ∗ b‬‬
‫))‪fl ((a − b)(a + b)) = (a ⊖ b) ⊛ (a ⊕ b) = rd (rd(a − b) ∗ rd(a + b‬‬

‫ﺩﺭ ﺯﻳﺮ ﻣﺮﺍﺣﻞ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺮﺑﻮﻁ ﺑﻪ ﺩﻭ ﻋﺒﺎﺭﺕ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬

‫
‬
‫‪fl a2 − b2‬‬ ‫))‪fl ((a − b)(a + b‬‬
‫‪z1 = a ⊛ a = 0.386 e2‬‬ ‫‪w1 = a ⊖ b = 0.540e0‬‬
‫‪z2 = b ⊛ b = 0.321 e2‬‬ ‫‪w2 = a ⊕ b = 0.119e2‬‬
‫‪z = z1 ⊖ z2 = 0.650e1‬‬ ‫‪w = w1 ⊛ w2 = 0.643e1‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‪ ،‬ﻋﺒﺎﺭﺕ ﺍﻭﻝ ﻣﻘﺪﺍﺭ ‪ 6.5‬ﻭ ﻋﺒﺎﺭﺕ ﺩﻭﻡ ﻣﻘﺪﺍﺭ ‪ 6.43‬ﺭﺍ ﻧﺘﻴﺠﻪ ﻣﻲﺩﻫﺪ‪.‬‬
‫ﻫﺮﭼﻨﺪ ﺍﻳﻦ ﺩﻭ ﻋﺒﺎﺭﺕ ﻣﻌﺎﺩﻝ ﻣﻲﺑﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ﻧﺘﻴﺠﻪ ﻣﺎﺷﻴﻨﻲ ﺁﻥﻫﺎ ﻣﺘﻔﺎﻭﺕ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺩﻭ ﻋﺒﺎﺭﺕ‬
‫ﺑﺮﺍﺑﺮ ‪ 6.4152‬ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻋﺒﺎﺭﺕ ﺩﻭﻡ ﺩﻗﻴﻖﺗﺮ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۳۲‬‬

‫■ ﻫﻤﮕﺮﺍ ﺷﺪﻥ ﺑﺮﺧﻲ ﺳﺮﻱﻫﺎﻱ ﻭﺍﮔﺮﺍ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬

‫ﺍﺯ ﺣﺴﺎﺑﺎﻥ ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺑﺮﺧﻲ ﺳﺮﻱﻫﺎ ﻛﻪ ﺟﻤﻠﻪ ﻋﻤﻮﻣﻲ ﺁﻥﻫﺎ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ ،‬ﻟﺰﻭﻣﺎ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‬
‫‪P‬‬
‫∞ ﺭﺍ ﻣﺜﺎﻝ ﺯﺩ ﻛﻪ ﺑﻪ ﺳﺮﻱ ﻫﺎﺭﻣﻮﻧﻴﻚ ‪ ۴۱‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ ﻭ ﻧﻤﻮﻧﻪﺍﻱ ﺍﺯ ﻳﻚ ﺳﺮﻱ ﻭﺍﮔﺮﺍ ﺷﻨﺎﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬‫ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺳﺮﻱ ‪n=1 n‬‬
‫‪1‬‬

‫ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ﺍﻳﻦ ﺳﺮﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﻪ ‪ +Inf‬ﻣﻨﺠﺮ ﮔﺮﺩﺩ‪ .‬ﺍﻣﺎ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ ν ∈ N‬ﻣﻮﺟﻮﺩ ﺍﺳﺖ‪ ،‬ﺑﻪ‬
‫ﻃﻮﺭﻱ ﻛﻪ ﺑﺎ ﺍﺯﺍﻱ ﻫﺮ ‪ n > ν‬ﻣﻘﺪﺍﺭ ‪ n1‬ﺩﺭ ﻣﺎﺷﻴﻦ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﺍﻳﻦ ﺳﺮﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﻪ ﻳﻚ ﻣﻘﺪﺍﺭ‬
‫ﻣﺘﻨﺎﻫﻲ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺩﺭ ]؟ ‪ ،‬ﺹ ‪ ،[۳۹‬ﺑﺮﻧﺎﻣﻪﺍﻱ ﺩﺭ ‪) MATLAB‬ﺑﺎ ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻀﺎﻋﻒ( ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺳﺮﻱ ﻫﺎﺭﻣﻮﻧﻴﻚ ﻧﻮﺷﺘﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﺍﻳﻦ ﺑﺮﻧﺎﻣﻪ ﺑﺮﺍﺑﺮ ‪ +Inf‬ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﺑﺮﻧﺎﻣﻪ ﻣﻘﺪﺍﺭ ‪ 14.39272672286478‬ﺭﺍ ﻧﺘﻴﺠﻪ ﻣﻲﺩﻫﺪ‪.‬‬

‫■ ﻣﺘﻔﺎﻭﺕ ﺑﻮﺩﻥ ﺭﻓﺘﺎﺭ ﺣﺪﻱ ﺑﺮﺧﻲ ﺗﻮﺍﺑﻊ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﺣﺪ ﻭﺍﻗﻌﻲ ﺗﻮﺍﺑﻊ‬

‫ﭼﻨﻴﻦ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ ﻛﻪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺣﺪ ﻳﻚ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﻄﻪ ﺻﻔﺮ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻛﺎﻣﭙﻴﻮﺗﺮ ﻛﻤﻚ ﮔﺮﻓﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ‬
‫ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻣﻘﺎﺩﻳﺮ ﻛﻮﭼﻚ ﻧﺰﺩﻳﻚ ﺑﻪ ﺻﻔﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ﺍﺯ ﺍﻳﻦ ﻣﻘﺎﺩﻳﺮ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺣﺪ ﺗﺎﺑﻊ ﺭﺍ ﺣﺪﺱ‬
‫ﻣﻲﺯﻧﻴﻢ‪ .‬ﺑﺎ ﻳﻚ ﻣﺜﺎﻝ‪ ،‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﺍﻳﻦ ﺭﻭﺵ ﻧﻤﻲﺗﻮﺍﻧﺪ ﻫﻤﻮﺍﺭﻩ ﺩﺭﺳﺖ ﺑﺎﺷﺪ‪.‬‬
‫‪x‬‬ ‫‪f (x) := e −e‬‬
‫‪x‬‬ ‫‪−x‬‬
‫ﺗﺎﺑﻊ‬
‫‪x‬‬
‫‪10−6‬‬ ‫‪1.999999999946489‬‬ ‫‪ex − e−x‬‬
‫=‪f (x) :‬‬
‫‪10−7‬‬ ‫‪1.999999998947288‬‬ ‫‪x‬‬
‫‪10−8‬‬ ‫‪1.999999987845058‬‬ ‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺎ ﻛﻤﻚ ﻣﺎﺷﻴﻦ ﺣﺪ ﺗﺎﺑﻊ ﺩﺭ‬
‫‪10−9‬‬ ‫‪2.000000054458440‬‬ ‫ﻧﻘﻄﻪ ‪ x = 0‬ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺩﺭ ‪ MATLAB‬ﻣﻘﺪﺍﺭ‬
‫‪10−10‬‬ ‫‪2.000000165480742‬‬
‫ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻧﻘﺎﻁ‬
‫‪10−11‬‬ ‫‪2.000000165480742‬‬
‫‪10−12‬‬ ‫‪2.000066778862220‬‬ ‫‪x = 10−6 , 10−7 , . . . , 10−18‬‬
‫‪10−13‬‬ ‫‪1.999511667349907‬‬
‫ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﻳﻦ ﻣﻘﺎﺩﻳﺮ ﺩﺭ ﺭﻭﺑﺮﻭ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺑﺎ‬
‫‪10−14‬‬ ‫‪1.998401444325282‬‬
‫‪10−15‬‬ ‫‪2.109423746787797‬‬ ‫ﺗﻮﺟﻪ ﺑﻪ ﺍﻃﻼﻋﺎﺕ ﺍﻳﻦ ﺟﺪﻭﻝ‪ ،‬ﭼﻨﻴﻦ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ ﻛﻪ ﺣﺪ‬
‫‪10−16‬‬ ‫‪1.110223024625157‬‬ ‫ﺗﺎﺑﻊ ﺩﺭ ﺍﻃﺮﺍﻑ ﺻﻔﺮ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﺎﻋﺪﻩ‬
‫‪10−17‬‬ ‫‪0‬‬ ‫ﻫﻮﭘﻴﺘﺎﻝ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﺭﻳﺎﻓﺖ ﻛﻪ ﺣﺪ ﺍﻳﻦ ﺗﺎﺑﻪ ﺩﺭ ﻧﻘﻄﻪ ‪ 0‬ﺑﺮﺍﺑﺮ ‪2‬‬
‫‪10−18‬‬ ‫‪0‬‬
‫‪..‬‬ ‫‪..‬‬
‫ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻛﺎﻣﭙﻴﻮﺗﺮﻱ ﻗﺎﺩﺭ ﺑﻪ ﻳﺎﻓﺘﻦ ﻭ ﻳﺎ ﺣﺪﺱ‬
‫‪.‬‬ ‫‪.‬‬ ‫ﺣﺪ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﻄﻪ ﺻﻔﺮ ﻧﻤﻲﺑﺎﺷﻴﻢ‪.‬‬

‫■ ﺳﺨﻦ ﺁﺧﺮ‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻔﺎﻭﺕﻫﺎﻱ ﺫﻛﺮ ﺷﺪﻩ ﺩﺭ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺩﻋﺎ ﻛﺮﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﻫﺮ ﮔﻮﻧﻪ ﻣﺤﺎﺳﺒﺎﺕ ﻛﻪ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‪ ،‬ﻟﺰﻭﻣﺎ ﺑﺎ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ‬
‫ﻳﻜﺴﺎﻥ ﻧﻴﺴﺖ‪ .‬ﻫﺮﭼﻨﺪ ﻛﻪ ﺑﺎ ﺍﺿﺎﻓﻪ ﻛﺮﺩﻥ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﻣﺎﻧﺘﻴﺲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﺭﺍ ﻛﺎﻫﺶ ﺩﺍﺩ ﺍﻣﺎ ﺣﺬﻑ ﻛﺎﻣﻞ ﺍﻳﻦ ﺗﻔﺎﻭﺕ‬
‫ﻣﻤﻜﻦ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﺳﻮﺍﻻﺕ ﺯﻳﺮ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪ .۱‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺍﻧﺪﺍﺯﻩ ﺗﻔﺎﻭﺕ ﻧﺘﺎﻳﺞ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻧﺘﺎﻳﺞ ﻭﺍﻗﻌﻲ ﺭﺍ ﺗﺨﻤﻴﻦ ﺯﺩ؟‬
‫‪ .۲‬ﺁﻳﺎ ﻫﻤﻮﺍﺭﻩ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺍﻋﺘﻤﺎﺩ ﻛﺮﺩ؟ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻳﻚ ﻋﺒﺎﺭﺕ‪ ،‬ﺁﻳﺎ ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺘﻴﺠﻪ ﻣﺎﺷﻴﻨﻲ‬
‫ﺑﺎ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ ﺗﻔﺎﻭﺕ ﺑﺴﻴﺎﺭ ﺯﻳﺎﺩﻱ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ؟ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﺍﺯ ﻣﺎﺷﻴﻦ‪ ،‬ﺑﻪ ﻫﻴﺞ ﻭﺟﻪ ﻗﺎﺑﻞ ﻗﺒﻮﻝ ﻧﺒﺎﺷﺪ‪.‬‬
‫‪ .۳‬ﺁﻳﺎ ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﻃﻮﺭﻱ ﺗﻐﻴﻴﺮ ﺩﺍﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺑﻪ ﻧﺘﻴﺠﻪ ﺩﻗﻴﻖ ﻧﺰﺩﻳﻚﺗﺮ ﺷﻮﺩ؟‬
‫‪41 Harmonic‬‬ ‫‪series‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۳۳‬‬ ‫ﻧﻤﺎﯾﺶ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨﯽ‬ ‫ﻓﺼﻞ ﺍﻭﻝ‪:‬‬

‫‪ .۴‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ ﻛﻪ ﺍﺯ ﺑﻴﻦ ﻋﺒﺎﺭﺕﻫﺎﻱ ﻣﻌﺎﺩﻝ ﻛﺪﺍﻡﻳﻚ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻨﺠﺮ ﺑﻪ ﻧﺘﻴﺠﻪ ﺩﻗﻴﻖﺗﺮﻱ ﻣﻲﺷﻮﺩ؟‬

‫ﺩﺭ ﻓﺼﻞ ﺑﻌﺪ‪ ،‬ﺍﺑﺰﺍﺭﻱ ﺑﺮﺍﻱ ﭘﺎﺳﺦ ﺑﻪ ﺳﻮﺍﻻﺕ ﻓﻮﻕ ﺍﺭﺍﻳﻪ ﺧﻮﺍﻫﺪ ﺷﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۳۴‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۲‬‬
‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ‬
‫ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬
‫ﻓﺼﻞ ﺩﻭﻡ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۳۶‬‬ ‫ﻣﺒﺎﻧﻲ ﺳﻨﺠﺶ ﺧﻄﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲‬‬
‫‪۳۶‬‬ ‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۲‬‬
‫‪۳۹‬‬ ‫∗ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ ‪. . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۲‬‬
‫‪۴۱‬‬ ‫ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ )ﺭﻭﻧﺪﻛﺮﺩﻥ( ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲‬‬
‫‪۴۲‬‬ ‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪. . FS (β, m, min, emax‬‬ ‫‪۱.۲.۲‬‬
‫‪۴۲‬‬ ‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪. . . FS (β, m, emin , emax‬‬ ‫‪۲.۲.۲‬‬
‫‪۴۷‬‬ ‫ﺧﻄﺎﻱ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ)ﺧﻄﺎﻱ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ( ‪. . . . . . . .‬‬ ‫‪۳.۲‬‬
‫‪۴۸‬‬ ‫ﺧﻄﺎﻱ ﻧﺎﺷﻲ ﺍﺯ ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۲‬‬
‫‪۴۹‬‬ ‫ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﻭ ﺗﻮﺍﺑﻊ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۲‬‬
‫‪۵۳‬‬ ‫ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ)ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ( ‪. . . . . .‬‬ ‫‪۲.۴.۲‬‬
‫‪۵۴‬‬ ‫ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۴.۲‬‬
‫‪۵۵‬‬ ‫ﺗﺮﻓﻨﺪﻫﺎ ﻭ ﺗﻮﺻﻴﻪﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۲‬‬

‫‪۳۵‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۳۶‬‬

‫ﺩﺭ ﻓﺼﻞ ﻗﺒﻞ ﺩﻳﺪﻳﻢ ﻛﻪ‪ ،‬ﻭﻗﺘﻲ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻋﺪﺩﻱ ﺭﺍ ﺩﺭ ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ ،‬ﻣﻌﻤﻮﻻ ﻧﻴﺎﺯ ﺍﺳﺖ ﻛﻪ ﻋﺪﺩ ﺭﻭﻧﺪ ﺷﻮﺩ ﻭ ﺳﭙﺲ‬
‫ﺫﺧﻴﺮﻩ ﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺧﻮﺩ ﻋﺪﺩ ﺫﺧﻴﺮﻩ ﻧﻤﻲﺷﻮﺩ ﻭ ﻋﺪﺩﻱ ﺩﻳﮕﺮﻱ ﻛﻪ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﭼﻨﻴﻦ ﺣﺎﻟﺘﻲ‬
‫ﻣﻲﮔﻮﻳﻴﻢ ﻛﻪ ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ﻋﺪﺩ‪ ،‬ﺧﻄﺎ ‪ ۱‬ﺭﺥ ﻣﻲﺩﻫﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﻳﺪﻳﻢ ﻛﻪ‪ ،‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﺩﺭ ﻣﺎﺷﻴﻦ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﺟﻤﻊ‬
‫ﻣﻲﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﺑﺎ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﻮﺭﺩ ﻧﻴﺰ ﻣﻲﮔﻮﻳﻴﻢ ﻛﻪ ﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﺩﺭ ﻣﺎﺷﻴﻦ ﺑﺎ ﺧﻄﺎ‬
‫ﻫﻤﺮﺍﻩ ﺍﺳﺖ‪.‬‬
‫‪۲‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﺩﺑﻴﺎﺕ ﺭﻭﺯﻣﺮﻩ ﻭﺍﮊﻩﻫﺎﻱ ﺧﻄﺎ ﻭ ﺍﺷﺘﺒﺎﻩ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﻌﺎﺩﻝ ﻫﺴﺘﻨﺪ‪ ،‬ﺍﻣﺎ ﺩﺭ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ ﺍﻳﻦ ﺩﻭ ﻭﺍﮊﻩ ﻛﺎﻣﻼ‬
‫ﻣﺘﻔﺎﻭﺕ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺷﺘﺒﺎﻫﻲ ﺭﺥ ﻧﻤﻲﺩﻫﺪ ﻭﻟﻲ ﺑﻪ ﻟﺤﺎﻅ ﻣﺤﺪﻭﺩﻳﺖ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‪ ،‬ﺑﺮﺧﻲ ﻋﻤﻠﻴﺎﺕ ﻭ ﻓﺮﺍﻳﻨﺪﻫﺎ‬
‫ﻃﻮﺭﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﺁﻥﻫﺎ ﺑﺎ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ ﺗﻔﺎﻭﺕ ﺩﺍﺭﺩ‪ .‬ﻭﺍﮊﻩ ﺧﻄﺎ ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﺍﻳﻦ ﺗﻔﺎﻭﺕﻫﺎ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻭﺍﮊﻩﻫﺎﻱ ﺧﻄﺎ ﻭ ﺗﻘﺮﻳﺐ ﺩﺭ ﻛﻨﺎﺭ ﻫﻢ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﻧﺪ‪ .‬ﻭﺟﻮﺩ ﺧﻄﺎ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺟﺎﻱ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﻳﻚ ﻋﺪﺩ‪،‬‬
‫ﻣﻘﺪﺍﺭ ﺗﻘﺮﻳﺒﻲ ﺁﻥ ﻋﺪﺩ ﺩﺭ ﺩﺳﺘﺮﺱ ﺑﺎﺷﺪ‪ .‬ﻣﺜﻼ‪ ،‬ﺩﺭ ﺫﺧﻴﺮﻩ ﻳﻚ ﻋﺪﺩ‪ ،‬ﻣﻘﺪﺍﺭ ﺭﻭﻧﺪ ﺷﺪﻩ ﻋﺪﺩ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺍﺻﻄﻼﺡ ﻣﻲﮔﻮﻳﻴﻢ‬
‫ﺑﻪ ﺟﺎﻱ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﻋﺪﺩ‪ ،‬ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﻋﺪﺩ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻝ ﻫﺮﭼﻪ ﻓﺎﺻﻠﻪ ﻣﻘﺪﺍﺭ ﺭﻭﻧﺪ ﺷﺪﻩ ﺑﺎ ﻣﻘﺪﺍﺭ ﺍﺻﻠﻲ ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺗﻘﺮﻳﺐ ﺩﻗﻴﻖﺗﺮ ﺍﺳﺖ‪.‬‬
‫ﻳﻚ ﺩﺳﺘﻪﺑﻨﺪﻱ ﻛﻠﻲ ﺍﺯ ﺧﻄﺎﻫﺎﻳﻲ ﻛﻪ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺭﺥ ﻣﻲﺩﻫﺪ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫‪.۱‬‬
‫ﺧﻄﺎﻱ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ‬ ‫‪.۲‬‬
‫ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻧﺎﺷﻲ ﺍﺯ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ‬ ‫‪.۳‬‬
‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ)ﺧﻄﺎﻱ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ(‬ ‫‪.۴‬‬
‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺧﻄﺎﻫﺎ ﺭﺍ ﻧﻤﻲﺗﻮﺍﻥ ﺍﺯ ﺑﻴﻦ ﺑﺮﺩ‪ .‬ﻟﺬﺍ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺁﻥﻫﺎ ﺭﺍ ﻣﻮﺭﺩ ﺁﻧﺎﻟﻴﺰ ﻗﺮﺍﺭ ﺩﺍﺩ ﻭ ﺗﺨﻤﻴﻨﻲ‬
‫ﺑﺮﺍﻱ ﺁﻥﻫﺎ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ‪ ،‬ﺑﻪ ﻣﻌﺮﻓﻲ ﻭ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻫﺎﻱ ﻓﻮﻕ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻣﻨﺘﻬﻲ ﻗﺒﻞ ﺍﺯ ﺁﻥ‬
‫ﺍﺑﺰﺍﺭ ﺳﻨﺠﺶ ﺧﻄﺎ ﻣﻌﺮﻓﻲ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺍﻧﺘﻬﺎ ﻧﻴﺰ ﺑﻪ ﺗﺮﻓﻨﺪﻫﺎﻳﻲ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎﻋﺚ ﻛﺎﻫﺶ ﺍﺛﺮ ﺧﻄﺎ ﺩﺭ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﺒﺎﻧﻲ ﺳﻨﺠﺶ ﺧﻄﺎ‬ ‫‪۱.۲‬‬

‫ﺑﺮﺍﻱ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺧﻄﺎ ﺩﻭ ﺳﺒﻚ ﻳﺎ ﺭﻭﻳﻜﺮﺩ ﻣﺘﻔﺎﻭﺕ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﺳﺒﻚ ﺍﻭﻝ ﺍﺯ ﺍﺑﺰﺍﺭﻫﺎﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ‪ ۳‬ﻭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ‪ ۴‬ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﺳﺒﻚ ﺩﻭﻡ ﺍﺯ ﺍﺑﺰﺍﺭﻫﺎﻱ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ‪ ۵‬ﻭ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﺎﻱ ﺩﺭﺳﺖ ‪ ۶‬ﺑﺮﺍﻱ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺧﻄﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﺍﻳﻦ ﺩﻭ ﺳﺒﻚ ﻣﻌﺮﻓﻲ ﻣﻲﮔﺮﺩﻧﺪ‪.‬‬

‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ‬ ‫�‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x‬ﻃﻮﻝ ﻳﻚ ﺟﺴﻢ‪ ،‬ﻭﺯﻥ ﻳﻚ ﺟﺴﻢ‪ ،‬ﻣﻘﺪﺍﺭ ﻳﻚ ﻋﺒﺎﺭﺕ ﻭ ﻳﺎ ﻫﺮ ﭼﻴﺰ ﺩﻳﮕﺮ ﺑﺎﺷﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭ ﺁﻥ ﺭﺍ ﺑﻪ ﺩﺳﺖ‬
‫ﺁﻭﺭﻳﻢ‪ .‬ﺍﺻﻄﻼﺣﺎ ﻣﻲﮔﻮﻳﻴﻢ ﻛﻪ ‪ x‬ﻳﻚ ﻛﻤﻴﺖ ﺍﺳﺖ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭﺵ ﺭﺍ ﺑﺪﺍﻧﻴﻢ‪ .‬ﻋﻤﻮﻣﺎ‪ ،‬ﺑﻪ ﺩﻻﻳﻞ ﻭﺟﻮﺩ ﺧﻄﺎ‪ ،‬ﻧﻤﻲﺗﻮﺍﻥ‬
‫ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﻛﻤﻴﺖ ‪ x‬ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺑﻠﻜﻪ ﻣﻘﺪﺍﺭ ﺩﻳﮕﺮﻱ ﻣﺎﻧﻨﺪ ̃‪ x‬ﺭﺍ ﺑﻪ ﺟﺎﻱ ‪ x‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﻳﻢ‪ x̃ .‬ﺭﺍ ﻳﻚ ﺗﻘﺮﻳﺐ ‪ ۷‬ﺑﺮﺍﻱ ‪x‬‬
‫ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﻣﺮﺗﻜﺐ ﺧﻄﺎ ﺷﺪﻩﺍﻳﻢ ﻭ ﺍﻧﺪﺍﺯﻩ ﺍﻳﻦ ﺧﻄﺎ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ |̃‪ .|x − x‬ﺍﻳﻦ ﻣﻘﺪﺍﺭ ﺑﻪ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻣﻮﺳﻮﻡ‬
‫ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺯﻳﺮ ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖ ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪1 Error‬‬ ‫‪5 Number‬‬ ‫)‪of correct decimal digits(figures‬‬
‫‪2 Mistake‬‬ ‫‪6 Number‬‬ ‫)‪of correct significant digits(figures‬‬
‫‪3 Absolute‬‬ ‫‪Error‬‬ ‫‪7 Approximation‬‬
‫‪4 Relative‬‬ ‫‪error‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۳۷‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺧﻄﺎﯼ ﻣﻄﻠﻖ‬ ‫ﺗﻌﺮﯾﻒ ‪۱−۲‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ̃‪ x‬ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ﻛﻤﻴﺖ ‪ x‬ﺑﺎﺷﺪ‪ .‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺑﻪﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪EA (x̃) = |x − x̃| .‬‬ ‫)‪(۱.۲‬‬

‫= ‪ x‬ﺭﺍ ﺑﺎ ‪ x̃ = 0.24‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ‬ ‫‪1‬‬
‫‪4‬‬ ‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺍﮔﺮ‬
‫‪1‬‬
‫= )̃‪EA (x‬‬ ‫‪− 0.24 = 0.01‬‬
‫‪4‬‬
‫ﺣﺎﻝ ﺍﮔﺮ ﻋﺪﺩ ‪ π‬ﺭﺍ ﺑﺎ ‪ π̃ = 3.14‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ |‪.EA (π̃) = |π − 3.14‬‬
‫ﺍﺻﻮﻻ ﺩﺭ ﻏﺎﻟﺐ ﻣﻮﺍﺭﺩ‪ ،‬ﻧﻤﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﮔﺮ ﺍﻳﻦ ﻛﺎﺭ ﻣﻤﻜﻦ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫ﻣﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺧﻮﺩ ﻛﻤﻴﺖ ﺭﺍ ﻧﻴﺰ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﻏﺎﻟﺒﺎ ﻣﻲﺗﻮﺍﻥ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ ﻭ ﺍﺯ ﺁﻥ‬
‫ﺑﺮﺍﻱ ﺗﺤﻠﻴﻞ ﺧﻄﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﻣﺜﻼ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ π‬ﺑﺎ ‪ 3.14‬ﻣﻲﺗﻮﺍﻥ ‪ 0.002‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﮔﺰﺍﺭﺵ‬
‫ﺩﺍﺩ )ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ‪ .(π = 3.141592 . . .‬ﺍﻟﺒﺘﻪ ‪ 0.0016‬ﻧﻴﺰ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‬
‫ﻣﻲﺗﻮﺍﻥ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ ﮔﺰﺍﺭﺵ ﻛﺮﺩ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﻫﺮﭼﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎ‬
‫ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﻣﻨﺎﺳﺐﺗﺮ ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﻃﻼﻋﺎﺕ ﺩﻗﻴﻖﺗﺮﻱ ﺩﺭ ﻣﻮﺭﺩ ﺧﻄﺎ ﺍﺭﺍﻳﻪ ﻣﻲﺩﻫﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺩﺭ ﺗﻘﺮﻳﺐ ‪ π‬ﺑﺎ‬
‫‪ ،3.14‬ﺩﻳﺪﻳﻢ ﻛﻪ ‪ 0.0016‬ﻭ ‪ 0.002‬ﻫﺮ ﺩﻭ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻣﻲﺑﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ‪ 0.0016‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﻣﻨﺎﺳﺐﺗﺮﻱ‬
‫ﻧﺴﺒﺖ ﺑﻪ ‪ 0.002‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ‪:‬‬

‫» ﺁﻳﺎ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﺑﺰﺍﺭ ﺧﻮﺑﻲ ﺑﺮﺍﻱ ﺳﻨﺠﺶ ﺧﻄﺎ ﺍﺳﺖ؟ ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺁﻳﺎ ﺑﺎ ﻛﻤﻚ ﻣﻘﺪﺍﺭ ﻭ ﻳﺎ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‬
‫ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻛﻴﻔﻴﺖ ﺁﻥ ﺗﻘﺮﻳﺐ ﭘﻲ ﺑﺮﺩ؟ «‬

‫ﺟﻮﺍﺏ ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺗﻘﺮﯾﺐ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺧﻄﺎﯼ ﻣﻄﻠﻖ(‬ ‫ﻣﺜﺎﻝ ‪)۱−۲‬ﻋﺪﻡ ﺗﺸﺨﯿﺺ ﻣﻨﺎﺳﺐ ﺑﻮﺩﻥ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻃﻮﻝ ﻳﻚ ﭘﻞ ﻭ ﻃﻮﻝ ﻳﻚ ﻣﺪﺍﺩ ﺭﺍ ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﺎ ﻳﻚ ﻣﺘﺮ ﻧﻮﺍﺭﻱ ﻭ ﻳﻚ ﺧﻂﻛﺶ ﺍﻧﺪﺍﺯﻩ ﺑﮕﻴﺮﻳﻢ‪.‬‬
‫ﻫﻤﭽﻨﻴﻦ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻃﻮﻝ ﺩﻗﻴﻖ ﭘﻞ ﺑﺮﺍﺑﺮ ‪ x = 1000 cm‬ﻭ ﻃﻮﻝ ﺩﻗﻴﻖ ﻣﺪﺍﺩ ﺑﺮﺍﺑﺮ ‪ y = 10 cm‬ﺍﺳﺖ‪ .‬ﺍﮔﺮ ﺷﺨﺼﻲ‬
‫ﻃﻮﻝ ﭘﻞ ﺭﺍ ‪ x̃ = 999 cm‬ﻭ ﻃﻮﻝ ﻣﺪﺍﺩ ﺭﺍ ‪ ỹ = 9 cm‬ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻧﻤﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﻫﺮ ﺩﻭ‬
‫ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺑﺮﺍﺑﺮ ‪ 1 cm‬ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‪ ،‬ﺩﻗﺖ ﺩﻭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻳﻜﺴﺎﻥ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﻧﻜﺘﻪ ﻣﻌﻘﻮﻝ ﺑﻪ ﻧﻈﺮ‬
‫ﻧﻤﻲﺭﺳﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺩﺭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻃﻮﻝ ﭘﻞ‪ ،‬ﺩﺭ ‪ 1000 cm‬ﻳﻚ ﺳﺎﻧﺘﻲﻣﺘﺮ ﺧﻄﺎ ﻣﺮﺗﻜﺐ ﺷﺪﻩﺍﻳﻢ ﻭﻟﻲ ﺩﺭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻃﻮﻝ ﻣﺪﺍﺩ‪،‬‬
‫ﺩﺭ ‪ 10 cm‬ﻳﻚ ﺳﺎﻧﺘﻲﻣﺘﺮ ﺧﻄﺎ ﻣﺮﺗﻜﺐ ﺷﺪﻩﺍﻳﻢ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﮔﺮ ﻓﻘﻂ ﺑﻪ ﻣﻘﺪﺍﺭ ﺧﻄﺎﻱ ﻣﻄﻠﻖ)ﻳﺎ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺁﻥ( ﺗﻮﺟﻪ ﻛﻨﻴﻢ ﻭ ﺑﻪ ﻣﻘﺪﺍﺭ ﺧﻮﺩ ﻛﻤﻴﺖ ﺗﻮﺟﻪ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫ﻧﻤﻲﺗﻮﺍﻥ ﺩﺭ ﻣﻮﺭﺩ ﻣﻨﺎﺳﺐ ﺑﻮﺩﻥ ﺗﻘﺮﻳﺐ ﺍﻇﻬﺎﺭ ﻧﻈﺮ ﻛﺮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺑﻪ ﺗﻨﻬﺎﻳﻲ ﺍﻃﻼﻋﺎﺗﻲ ﺩﺭ ﻣﻮﺭﺩ ﻛﻴﻔﻴﺖ‬
‫ﺗﻘﺮﻳﺐ ﺑﻪ ﻣﺎ ﻧﻤﻲﺩﻫﺪ‪ .‬ﺑﻪ ﻣﻨﻈﻮﺭ ﺑﺮﻃﺮﻑ ﻛﺮﺩﻥ ﺍﻳﻦ ﻣﺸﻜﻞ ﺍﺯ ﺍﺑﺰﺍﺭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۳۸‬‬

‫ﺧﻄﺎﯼ ﻧﺴﺒﯽ ﻭ ﺩﺭﺻﺪ ﺧﻄﺎﯼ ﻧﺴﺒﯽ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۲‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ̃‪ x‬ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ﻛﻤﻴﺖ ‪ x‬ﺑﺎﺷﺪ‪ ،‬ﻛﻪ ‪ .x ̸= 0‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ x‬ﺑﺎ ̃‪ x‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫|̃‪|x − x‬‬
‫= )̃‪ER (x‬‬ ‫‪.‬‬ ‫)‪(۲.۲‬‬
‫|‪|x‬‬
‫ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻭ ﻳﺎ ﺩﺭﺻﺪ ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫|̃‪|x − x‬‬
‫= ‪ER (x̃) × 100 %‬‬ ‫‪× 100 %.‬‬ ‫)‪(۳.۲‬‬
‫|‪|x‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ‪ ،‬ﺩﻭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱-۲‬ﺭﺍ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻘﺎﻳﺴﻪ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۲−۲‬ﻣﺜﺎﻟ ﺍﺯ ﺧﻄﺎﯼ ﻧﺴﺒﯽ(‬

‫ﺩﺭ ﻣﺜﺎﻝ ‪ ۱-۲‬ﺩﻳﺪﻳﻢ ﻛﻪ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﻫﺮ ﺩﻭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺑﺮﺍﺑﺮ ‪ 1 cm‬ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‪ ،‬ﺩﻗﺖ ﺩﻭ‬
‫ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻳﻜﺴﺎﻥ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺍﻳﻦ ﺩﻭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺭﺍ ﺑﺎ ﻣﻌﻴﺎﺭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻲﺳﻨﺠﻴﻢ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫|̃‪|x − x‬‬ ‫‪1‬‬
‫= )̃‪ER (x‬‬ ‫=‬ ‫‪= 0.001 (= 0.1 %),‬‬
‫|‪|x‬‬ ‫‪1000‬‬
‫|̃‪|y − y‬‬ ‫‪1‬‬
‫= )̃‪ER (y‬‬ ‫=‬ ‫‪= 0.1 (= 10 %).‬‬
‫|‪|y‬‬ ‫‪10‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ‪ ،‬ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‪ ،‬ﺩﺭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﭘﻞ ﺧﻄﺎ ‪) 0.001‬ﻳﻚ ﺩﻫﻢ ﺩﺭﺻﺪ( ﺍﺳﺖ ﻭ ﺩﺭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻣﺪﺍﺩ‬
‫ﺧﻄﺎ ‪) 0.1‬ﺩﻩ ﺩﺭﺻﺪ( ﻣﻲﺑﺎﺷﺪ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‪ ،‬ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﭘﻞ ﺩﻗﻴﻖﺗﺮ ﺻﻮﺭﺕ ﮔﺮﻓﺘﻪ ﺍﺳﺖ‪.‬‬

‫ﻫﻤﺎﻧﻨﺪ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‪ ،‬ﻏﺎﻟﺒﺎ ﻧﻤﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﻟﺬﺍ ﺩﺭ ﻋﻤﻞ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‬
‫|‪ER (π̃) = |π−3.14‬‬
‫‪π‬‬ ‫ﮔﺰﺍﺭﺵ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺍﮔﺮ ‪ π‬ﺭﺍ ﺑﺎ ‪ ˜π = 3.14‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﺑﺮ‬
‫ﺍﺳﺖ ﻭ ﭼﻮﻥ ‪ π‬ﻋﺪﺩﻱ ﻧﺎﻣﺨﺘﻮﻡ ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﻧﻤﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺗﻘﺮﻳﺐ ﺧﻄﺎﻱ‬
‫ﻧﺴﺒﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‬
‫|‪|π − 3.14‬‬ ‫‪0.0016‬‬ ‫‪0.0016‬‬
‫= )̃‪ER (π‬‬ ‫<‬ ‫<‬ ‫‪≃ 5.12 e − 4.‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪3.13‬‬

‫■ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺗﻘﺮﻳﺒﻲ‬

‫ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﺍﻃﻼﻋﺎﺗﻲ ﺍﺯ ﻛﻤﻴﺖ ﺍﺻﻠﻲ ‪ x‬ﻭ ﺣﺪﻭﺩ ﺁﻥ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ ﻭ ﻳﺎ ﺑﻪ ﺳﺎﺩﮔﻲ ﺩﺭ ﺩﺳﺘﺮﺱ ﻧﻴﺴﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ‬
‫|‪ ER (x̃) = |x̃−x‬ﺍﺯ‬
‫ﺳﺎﺩﻩﺗﺮ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺟﺎﻱ |‪|x‬‬
‫|‪|x̃ − x‬‬
‫=‪ẼR (x̃) :‬‬ ‫)‪(۴.۲‬‬
‫|̃‪|x‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪ ،‬ﻛﻪ ﺑﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺗﻘﺮﻳﺒﻲ ‪ ۸‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ )̃‪ ẼR (x‬ﻣﺘﻔﺎﻭﺕ ﺍﺯ )̃‪ ER (x‬ﺍﺳﺖ ﻭ ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ‬
‫ﻫﺮ ﭼﻪ )̃‪ ẼR (x‬ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ )̃‪ ER (x‬ﻧﺰﺩﻳﻚﺗﺮ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ )̃‪ ẼR (x‬ﻛﻮﭼﻚ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺮﺍﻱ‬
‫ﺳﺎﺩﮔﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺁﻥ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ )̃‪ ER (x‬ﻣﻌﺮﻓﻲ ﻛﺮﺩ‪.‬‬

‫‪8 Relative‬‬ ‫‪Approximate Error‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۳۹‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫∗ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ‬ ‫�‬

‫ﻋﻼﻭﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ‪ ،‬ﺳﺒﻚ ﻭ ﺭﻭﻳﻜﺮﺩ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﮔﺰﺍﺭﺵ ﺧﻄﺎ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺍﻳﻦ ﺳﺒﻚ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻣﺸﺨﺺ‬
‫ﻛﺮﺩﻥ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ‪ ۹‬ﻭ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﺎﻱ ﺩﺭﺳﺖ ‪ ۱۰‬ﺩﺭ ﻳﻚ ﺗﻘﺮﻳﺐ‪.‬‬

‫ﺗﻘﺮﯾﺐ‬ ‫ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺩﺭ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۲‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ̃‪ x‬ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ﻛﻤﻴﺖ ‪ x‬ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ‪ k‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﺻﺤﻴﺢ ﻧﺎﻣﻨﻔﻲ ﺑﺎﺷﺪ ﻛﻪ‬
‫< |̃‪EA (x̃) := |x − x‬‬ ‫‪1‬‬
‫‪2‬‬ ‫‪× 10−k ,‬‬ ‫)‪(۵.۲‬‬
‫ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺗﻘﺮﻳﺐ ̃‪ x‬ﺩﺍﺭﺍﻱ ‪ k‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﮔﻮﻳﻴﻢ ﻛﻪ ̃‪ x‬ﻭ ‪ x‬ﺗﺎ ‪ k‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﻄﺎﺑﻘﺖ‬
‫ﺩﺍﺭﻧﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۳−۲‬ﻣﺤﺎﺳﺒﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ(‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ‪ x = 1.23456‬ﺑﺎ ‪ x̃ = 1.23‬ﺗﻘﺮﻳﺐ ﺯﺩﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪1‬‬
‫= ‪|x − x̃| = |1.23456 − 1.23| = 0.00456 < 0.5‬‬ ‫‪× 100‬‬
‫‪2‬‬
‫‪1‬‬
‫‪< 0.05 = × 10−1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪< 0.005 = × 10−2‬‬
‫‪2‬‬
‫‪1‬‬
‫‪≮ 0.0005 = × 10−3‬‬
‫‪2‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ .2‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺗﻘﺮﻳﺐ ‪ x̃ = 1.23‬ﺩﺍﺭﺍﻱ ‪ 2‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ‬
‫ﺩﺭﺳﺖ ﺍﺳﺖ‪.‬‬
‫ﭼﻨﻴﻦ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺑﺮﺍﺑﺮ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﻳﻜﺴﺎﻥ ﺩﺭ ﺧﻮﺩ ﻋﺪﺩ ﻭ ﺗﻘﺮﻳﺐ ﺁﻥ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺍﻳﻦ‬
‫ﺑﺮﺩﺍﺷﺖ ﺍﺷﺘﺒﺎﻫﻲ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ y = 2‬ﻭ ‪ .ỹ = 1.9999‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ y‬ﻭ ̃‪ y‬ﺩﺍﺭﺍﻱ ﻫﻴﭻ‬
‫ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﻳﻜﺴﺎﻧﻲ ﻧﻴﺴﺘﻨﺪ‪ .‬ﺍﻣﺎ ̃‪ y‬ﺗﻘﺮﻳﺒﻲ ﺑﺎ ﭼﻬﺎﺭ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﭼﺮﺍ ﻛﻪ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪1‬‬
‫= ‪|y − ỹ| = |2 − 1.9999| = 0.0001 < 0.0005‬‬ ‫‪× 10−3‬‬
‫‪2‬‬
‫‪1‬‬
‫‪≮ 0.00005 = × 10−4 .‬‬
‫‪2‬‬

‫ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ‪ ،‬ﺑﻪ ﻧﻮﻋﻲ ﻳﻚ ﻛﺮﺍﻥ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ‪ ،‬ﺑﺎ ﻣﻔﻬﻮﻡ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ‬
‫ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ‪ ،‬ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﺸﺨﺾ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪9 Number‬‬ ‫)‪of correct decimal digits(figures‬‬ ‫‪10 Number‬‬ ‫)‪of correct significant digits(figures‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴۰‬‬

‫ﺗﻘﺮﯾﺐ‬ ‫ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨ ﺩﺭﺳﺖ ﺩﺭ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۴−۲‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ̃‪ x‬ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ﻛﻤﻴﺖ ‪ x ̸= 0‬ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ‪ s‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﺻﺤﻴﺢ ﻧﺎﻣﻨﻔﻲ ﺑﺎﺷﺪ ﻛﻪ‬
‫|̃‪|x − x‬‬
‫=‪ER (x̃) :‬‬ ‫‪< 5 × 10−s ,‬‬ ‫)‪(۶.۲‬‬
‫|‪|x‬‬
‫ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺗﻘﺮﻳﺐ ̃‪ x‬ﺩﺍﺭﺍﻱ ‪ s‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﮔﻮﻳﻴﻢ ﻛﻪ ̃‪ x‬ﻭ ‪ x‬ﺗﺎ ‪ s‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺑﺎ ﻳﻚﺩﻳﮕﺮ‬
‫ﻣﻄﺎﺑﻘﺖ ﺩﺍﺭﻧﺪ‪.‬‬
‫ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ x‬ﺑﺎ ̃‪ x‬ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ )̃‪ NCSD(x‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﻋﻤﻞ ﺑﻴﺸﺘﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﺍﺑﺰﺍﺭ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎ ﻣﻌﻨﻲ ﺩﺭﺳﺖ ﻧﻴﺰ ﺑﻴﺸﺘﺮ ﺍﺯ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ‬
‫ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺑﺮﺍﻱ ﺳﻨﺠﺶ ﺧﻄﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩﻥ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ x‬ﺑﺎ ̃‪ ،x‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺗﻌﺮﻳﻒ ‪ ۴-۲‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺍﺯ ﺭﺍﺑﻄﻪ ﺯﻳﺮ‬
‫ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ x‬ﺑﺎ ̃‪ x‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‬
‫‬ ‫‬ ‫‬
‫|‪5 |x‬‬
‫‪NCSD(x̃) = log10‬‬ ‫)‪(۷.۲‬‬
‫|̃‪|x − x‬‬

‫ﻣﺜﺎﻝ ‪)۴−۲‬ﻣﺤﺎﺳﺒﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨ ﺩﺭﺳﺖ(‬

‫ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ ﺩﻳﺪﻳﻢ ﻛﻪ ‪ x̃ = 1.23‬ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ‪ x = 1.23456‬ﺑﺎ ﺩﻭ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ‬
‫ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫|̃‪|x − x‬‬ ‫|‪|1.23456 − 1.23‬‬
‫=‬ ‫‪= 0.003693 . . . < 0.5 = 5 × 10−1‬‬
‫|‪|x‬‬ ‫‪1.23456‬‬
‫‪< 0.05 = 5 × 10−2‬‬
‫‪< 0.005 = 5 × 10−3‬‬
‫‪1‬‬
‫= ‪≮ 0.0005‬‬ ‫‪× 10−4 .‬‬
‫‪2‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ .3‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ‪ ،‬ﻣﻲﺗﻮﺍﻧﺴﺘﻴﻢ ﺍﺯ‬
‫ﻓﺮﻣﻮﻝ )‪ (۷.۲‬ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ .‬ﺑﺎ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺍﻳﻦ ﻓﺮﻣﻮﻝ ﻧﻴﺰ ﻣﺸﺨﺺ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﻘﺮﻳﺐ ̃‪ x‬ﺩﺍﺭﺍﻱ ﺳﻪ ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ‬
‫ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‬ ‫‬ ‫‬ ‫‬ ‫‬ ‫‬
‫|‪5 |x‬‬ ‫‪5 ∗ 1.23456‬‬
‫‪NCSD(x̃) := log10‬‬ ‫‪= log10‬‬ ‫‪=3‬‬
‫|̃‪|x − x‬‬ ‫‪0.00456‬‬
‫ﻣﻤﻜﻦ ﺍﺳﺖ ﭼﻨﻴﻦ ﺗﺼﻮﺭ ﺷﻮﺩ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺑﺮﺍﺑﺮ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ )ﺻﺤﻴﺢ ﻭ ﺍﻋﺸﺎﺭ( ﻳﻜﺴﺎﻥ ﺩﺭ ﺧﻮﺩ ﻋﺪﺩ ﻭ ﺗﻘﺮﻳﺐ ﺁﻥ‬
‫ﺍﺳﺖ‪ .‬ﺍﻣﺎ‪ ،‬ﻫﻤﺎﻧﻨﺪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ‪ ،‬ﺍﻳﻦ ﺑﺮﺩﺍﺷﺖ ﺍﺷﺘﺒﺎﻩ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ‪ y = 2‬ﺑﺎ ‪.ỹ = 1.9999‬‬
‫ﺗﻘﺮﻳﺐ ﺯﺩﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‬ ‫‬ ‫ ‬ ‫‬ ‫‬
‫|‪5 |y‬‬ ‫‪2∗5‬‬
‫‪NCSD(ỹ) := log10‬‬ ‫‪= log10‬‬ ‫‪=5‬‬
‫|̃‪|y − y‬‬ ‫‪0.0001‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ̃‪ y‬ﺗﻘﺮﻳﺒﻲ ﺑﺎ ﭘﻨﺞ ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺑﺮﺍﻱ ‪ y‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺩﺭﺣﺎﻟﻲ ﻛﻪ ‪ y‬ﻭ ̃‪ y‬ﺩﺍﺭﺍﻱ ﻫﻴﭻ ﺭﻗﻢ ﺻﺤﻴﺢ ﻭ ﺍﻋﺸﺎﺭ‬
‫ﻳﻜﺴﺎﻧﻲ ﻧﻴﺴﺘﻨﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۴۱‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﻣﺜﺎﻝ ‪)۵−۲‬ﻣﺜﺎﻝﻫﺎﯾﯽ ﺍﺯ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨ ﺩﺭﺳﺖ ﺩﺭ ﺗﻘﺮﯾﺐ ‪ π‬ﺑﺎ ﮐﺴﺮﻫﺎ(‬

‫‪ 22‬ﺗﻘﺮﻳﺐ ﺯﺩﻩﺍﻳﻢ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺪﺍﻧﻴﻢ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﭼﻘﺪﺭ ﺍﺳﺖ‪ .‬ﺩﺍﺭﻳﻢ‬
‫◀ ﻓﺮﺽ ﻛﻨﻴﻢ ﻋﺪﺩ ‪ π‬ﺭﺍ ﺑﺎ ‪7‬‬
‫‪ π = 3.141592654 . . .‬ﻭ ‪ . 7 = 3.142857142 . . .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬
‫‪22‬‬

‫‪π−‬‬ ‫‪22‬‬
‫‪7‬‬ ‫‪= 0.00126449 · · · < 0.005 = 0.5 × 10−2 .‬‬
‫ﺩﺍﺭﺍﻱ ﺩﻭ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ‬ ‫‪22‬‬
‫‪7‬‬ ‫‪ ، 22‬ﺑﻪ ﺗﻌﺪﺍﺩ ﺩﻭ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺗﻘﺮﻳﺐ‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ π‬ﺑﺎ ‪7‬‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪π − 22‬‬ ‫‪0.00126449 . . .‬‬
‫‪7‬‬
‫=‬ ‫‪≃ 0.0004025 < 0.0005 = 5 × 10−4 .‬‬
‫|‪|π‬‬ ‫‪3.141592654 . . .‬‬
‫‪ 22‬ﺩﺍﺭﺍﻱ ﭼﻬﺎﺭ ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺗﻘﺮﻳﺐ ‪ π‬ﺑﺎ ‪ ، 7‬ﺑﻪ ﺗﻌﺪﺍﺩ ﭼﻬﺎﺭ ﺭﻗﻢ ﺑﺎ ﻣﻌﻨﻲ ﺩﺭﺳﺖ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺑﻪ ﻋﺒﺎﺭﺗﻲ‪ ،‬ﺗﻘﺮﻳﺐ ‪7‬‬
‫‪22‬‬

‫ﺩﺭﺳﺖ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫= ̃‪ x‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬ ‫‪355‬‬
‫‪113‬‬ ‫◀ ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻋﺪﺩ ‪ x = π‬ﺭﺍ ﺑﺎ‬
‫‪π−‬‬ ‫‪355‬‬
‫‪113‬‬ ‫|‪= |3.141592654 · · · − 3.14159292035 . . .‬‬
‫‪= 0.00000026676 . . .‬‬
‫‪< 0.0000005 = 0.5 × 10−6 .‬‬
‫ﻫﻤﭽﻨﻴﻦ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪π−‬‬ ‫‪355‬‬
‫‪0.00000026676 . . .‬‬
‫‪113‬‬
‫=‬ ‫‪≃ 0.0000000848 < 0.00000005 = 5 × 10−7 .‬‬
‫|‪|π‬‬ ‫‪3.141592654 . . .‬‬
‫‪ 355‬ﻋﺪﺩ ‪ π‬ﺭﺍ ﺑﺎ ﺷﺶ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺑﺎ ﻫﻔﺖ ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﺪ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻋﺪﺩ ‪113‬‬
‫ﺩﺍﺭﺍﻱ‬ ‫‪22‬‬
‫‪7‬‬ ‫ﻧﺴﺒﺖ ﺑﻪ‬ ‫‪355‬‬
‫‪113‬‬ ‫ﺑﺮﺍﻱ ‪ π‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﭼﺮﺍ ﻛﻪ‬ ‫‪22‬‬
‫‪7‬‬ ‫‪ 355‬ﺗﻘﺮﻳﺐ ﺑﻬﺘﺮﻱ ﻧﺴﺒﺖ ﺑﻪ‬
‫◀ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﺷﺪ‪ ،‬ﻋﺪﺩ ‪113‬‬
‫ﭼﻬﺎﺭ ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﺑﻴﺸﺘﺮ ﻭ ﺳﻪ ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺑﻴﺸﺘﺮ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﺍﻧﺘﻬﺎ ﺗﺎﻛﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺸﺎﺭ ﺩﺭﺳﺖ ﻭ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﻣﻔﺎﻫﻴﻤﻲ ﻣﺮﺗﺒﻂ ﺑﺎ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ‬
‫ﻣﻄﻠﻖ ﻭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺑﺎ ﺯﺑﺎﻥ ﺩﻳﮕﺮﻱ ﺑﻪ ﺑﻴﺎﻥ ﺧﻄﺎ ﻣﻲﭘﺮﺩﺍﺯﻧﺪ‪ .‬ﺑﺮﺧﻲ ﻣﻨﺎﺑﻊ ﺍﺯ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺑﺮﺍﻱ‬
‫ﺳﻨﺠﺶ ﺧﻄﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻨﺪ ﻭ ﺑﺮﺧﻲ ﺩﻳﮕﺮ ﺍﺯ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺍﻋﺸﺎﺭ ﻭ ﺩﺭﺳﺖ‪.‬‬

‫ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ )ﺭﻭﻧﺪﻛﺮﺩﻥ( ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬ ‫‪۲.۲‬‬

‫ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺩﺭ ﻣﺎﺷﻴﻦ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﻴﺎﺯ ﺑﻪ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦﺻﻮﺭﺕ ﺧﻮﺩ ﻋﺪﺩ ﺫﺧﻴﺮﻩ ﻧﻤﻲﺷﻮﺩ‬
‫ﺑﻠﻜﻪ ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻪ ﺟﺎﻱ ﻋﺪﺩ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﺧﻄﺎ ﺭﺥ ﺩﺍﺩﻩ ﺍﺳﺖ ﻭ ﺍﻳﻦ ﺧﻄﺎ ﺑﻪ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ‪ ۱۱‬ﻳﺎ‬
‫ﺧﻄﺎﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ‪ ۱۲‬ﻭ ﻳﺎ ﺧﻄﺎﻱ ﻧﻤﺎﻳﺶ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ‪ ،|x| > xmax‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﻣﻌﻨﻲ ﻧﺪﺍﺭﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ‪ ،‬ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ‪ x‬ﺳﺮﺭﻳﺰﻱ‬
‫ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ ﻭ ‪ ±Inf‬ﺑﻪ ﺟﺎﻱ ‪ x‬ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻗﺮﺍﺭ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﺣﻴﺎﻧﺎ ‪x‬‬
‫ﺑﻌﺪ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺫﺧﻴﺮﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺑﺮﺍﻱ ‪x‬ﻫﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﻣﻌﻨﻲ ﭘﻴﺪﺍ‬
‫ﻣﻲﻛﻨﺪ‪.‬‬

‫‪11 Saving‬‬ ‫‪Error‬‬ ‫‪12 Rounding‬‬ ‫??‪error‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴۲‬‬

‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪FS (β, m, min, emax‬‬ ‫�‬
‫ﻳﻚ ﻣﺎﺷﻴﻦ‪ ،‬ﺑﺎ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ) ‪ FS (β, m, emin , emax‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬
‫] ‪x ∈ [−xmax , +xmax‬‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ ﺩﺭ ﻓﺼﻞ ﺍﻭﻝ ﮔﻔﺘﻪ ﺷﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺮﺍﻱ ‪ x‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪:‬‬
‫‪.‬‬
‫‪x = ±0 d−1 · · · d−m d−m−1 · · · × β e ,‬‬ ‫)‪(۸.۲‬‬
‫ﺑﻪ ﻃﻮﺭﻱﻛﻪ ‪ .emin ≤ e ≤ emax‬ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ‪ x‬ﻏﻴﺮﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ‪ ،x‬ﺍﺑﺘﺪﺍ ﺑﺎﻳﺪ ﺁﻥﺭﺍ ﺭﻭﻧﺪ ﻛﻨﻴﻢ‪ .‬ﺩﺭ‬
‫ﻓﺮﺍﻳﻨﺪ ﺭﻭﻧﺪ ﻛﺮﺩﻥ‪ x ،‬ﻳﺎ ﺑﻪ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻼﻓﺎﺻﻠﻪ ﻗﺒﻞ ﺍﺯ ﺧﻮﺩ ﻳﺎ ﺑﻪ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪ ﺍﺯ ﺧﻮﺩ ﺭﻭﻧﺪ ﻣﻲﺷﻮﺩ‪ .‬ﺍﮔﺮ ﺍﻳﻦ‬
‫ﺍﻋﺪﺍﺩ ﺭﺍ ﺑﺎ ‪ x−‬ﻭ ‪ x+‬ﻧﺸﺎﻥ ﺩﻫﻴﻢ ﺁﻥﮔﺎﻩ ‪ x‬ﺩﺍﺧﻞ ﺑﺎﺯﻩ ] ‪ [x− , x+‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﻃﻮﻝ ﺍﻳﻦ ﺑﺎﺯﻩ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‬
‫ﺑﺎ‪:‬‬
‫‪x+ − x− = β e−m .‬‬
‫ﺣﺎﻝ ﭼﻮﻥ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [x− , x+‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ ،‬ﻟﺬﺍ ﻓﺎﺻﻠﻪ ‪ x‬ﺗﺎ ‪ x−‬ﻭ ﻳﺎ ‪ x+‬ﻧﻤﻲﺗﻮﺍﻧﺪ ﺍﺯ ﻃﻮﻝ ﺑﺎﺯﻩ ﻳﻌﻨﻲ ‪ β e−m‬ﺑﻴﺸﺘﺮ ﺑﺎﺷﺪ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ‪/‬ﭘﺎﻳﻴﻦ‪ ،‬ﺣﺪﺍﻛﺜﺮ ﺍﺧﺘﻼﻑ ‪ x‬ﻭ ﺭﻭﻧﺪ ﺷﺪﻩ ﺁﻥ ﺑﺮﺍﺑﺮ ‪β e−m‬‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‬
‫‪|x − rd0 (x)| < β e−m ,‬‬ ‫)‪(۹.۲‬‬
‫‪|x − rdd (x)| < β e−m ,‬‬ ‫)‪(۱۰.۲‬‬
‫‪|x − rdu (x)| < β e−m .‬‬ ‫)‪(۱۱.۲‬‬
‫ﺍﻣﺎ ﺩﺭ ﻣﻮﺭﺩ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﮔﺮ ‪ x‬ﺩﺭ ﻧﻴﻤﻪ ﺍﻭﻝ ﺑﺎﺯﻩ ﺑﺎﺷﺪ ﺑﻪ ‪ x‬ﻭ ﺍﮔﺮ ‪ x‬ﺩﺭ ﻧﻴﻤﻪ ﺩﻭﻡ ﺑﺎﺯﻩ ﺑﺎﺷﺪ‬
‫‪−‬‬

‫ﺑﻪ ‪ x+‬ﺭﻭﻧﺪ ﻣﻲﺷﻮﺩ‪ .‬ﻟﺬﺍ ﻓﺎﺻﻠﻪ ﻋﺪﺩ ﺗﺎ ﺭﻭﻧﺪ ﺷﺪﻩﺍﺵ ﺣﺪﺍﻛﺜﺮ ﺑﺮﺍﺑﺮ ﻧﺼﻒ ﻃﻮﻝ ﺯﻳﺮ ﺑﺎﺯﻩ ﺍﺳﺖ‪ .‬ﺑﻪﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪1 e−m‬‬
‫≤ |)‪|x − rdr (x‬‬ ‫‪β‬‬
‫‪2‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺩﺍﺭﺍﻱ ﻛﻤﺘﺮﻳﻦ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺩﺭ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﻤﺎﻳﺶ ‪ x = ±0.d−1 · · · d−m d−m−1 · · · × β e‬ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻧﻤﺎﻱ‬
‫ﻋﺪﺩ ‪ ،x‬ﻳﻌﻨﻲ ‪ ،e‬ﺩﺭ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻧﻤﺎﻳﺶ ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪ .‬ﻟﺬﺍ ﻫﺮﭼﻪ ﻋﺪﺩ ﺍﺯ ﻟﺤﺎﻅ ﻗﺪﺭ ﻣﻄﻠﻖ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ)‪e‬‬
‫ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ(‪ ،‬ﺁﻥﮔﺎﻩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻧﻤﺎﻳﺶ ﻧﻴﺰ ﺑﺰﺭﮒﺗﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﻣﻮﺿﻮﻉ ﺑﻪﺧﺎﻃﺮ ﺗﻮﺯﻳﻊ ﻏﻴﺮﻳﻜﻨﻮﺍﺧﺖ ﺍﻋﺪﺍﺩ‬
‫ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﻣﻲﺑﺎﺷﺪ ﻭ ﺍﻟﺒﺘﻪ ﻳﻚ ﺧﺼﻮﺻﻴﺖ ﻣﻨﻔﻲ ﻣﺤﺴﻮﺏ ﻧﻤﻲﺷﻮﺩ‪.‬‬

‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪FS (β, m, emin , emax‬‬ ‫�‬
‫ﻳﻚ ﻣﺎﺷﻴﻦ‪ ،‬ﺑﺎ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ) ‪ FS (β, m, emin , emax‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬
‫] ‪x ∈ [−xmax , +xmax‬‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ x‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [−xmax , +xmax‬ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﻣﻲﺗﻮﺍﻥ ﻧﻤﺎﻳﺸﻲ ﺑﺮﺍﻱ ‪ x‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ‬
‫ﻧﻈﺮ ﮔﺮﻓﺖ‪:‬‬
‫‪.‬‬
‫‪x = ±0 d−1 · · · d−m d−m−1 · · · × β e‬‬ ‫)‪(۱۲.۲‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ .emin ≤ e ≤ emax‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺍﻳﻦ ﻧﻤﺎﻳﺶ‪ ،‬ﺍﮔﺮ ‪ ،|x| ≥ xmin‬ﺁﻥﮔﺎﻩ ‪ d−1 > 0‬ﻭ ﺍﮔﺮ ‪،|x| < xmin‬‬
‫ﺁﻥﮔﺎﻩ ‪.d−1 = 0‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۴۳‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺍﻛﻨﻮﻥ ﺑﻪ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ ﻭ )‪ .x̃ := rd0 (x‬ﺩﻳﺪﻳﻢ ﻛﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪EA (x̃) = |x − x̃| < β e−m .‬‬
‫ﺍﻛﻨﻮﻥ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﺭﺍ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪،‬‬
‫ﺩﺍﺭﻳﻢ‪:‬‬
‫|)‪|x − rd0 (x‬‬ ‫‪β e−m‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ )‪(۹.۲‬‬
‫= )̃‪ER (x‬‬
‫|‪|x‬‬
‫<‬
‫‪.‬‬
‫‪0 d−1 · · · d−m d−m−1 · · · × β e‬‬
‫‪β −m‬‬
‫=‬
‫‪.‬‬
‫· · · ‪0 d−1 · · · d−m d−m−1‬‬
‫‪.‬‬

‫‪.‬‬
‫ﺍﻛﻨﻮﻥ ﺑﺮﺍﻱ ﺁﻥ ﻛﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﻛﺴﺮ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻛﻤﺘﺮﻳﻦ ﻣﻘﺪﺍﺭ ﻣﻤﻜﻦ ﺑﺮﺍﻱ ﻣﺎﻧﺘﻴﺲ · · · ‪0 d−1 · · · d−m d−m−1‬‬
‫ﺭﺍ ﺩﺭ ﻣﺨﺮﺝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﻛﻨﻴﻢ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺍﮔﺮ ‪ ،|x| ≥ xmin‬ﺁﻥﮔﺎﻩ ‪ d−1 > 0‬ﻭ ﻟﺬﺍ ﻣﺎﻧﺘﻴﺲ ﺣﺪﺍﻗﻞ ﺑﺮﺍﺑﺮ‬
‫‪ (0.1)β = β −1‬ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺍﮔﺮ ‪ ،|x| < xmin‬ﺁﻥﮔﺎﻩ ‪ d−1 = 0‬ﻭ ﻟﺬﺍ ﻣﺎﻧﺘﻴﺲ ﺍﺯ ‪ (0.1)β = β −1‬ﻧﺎﺑﻴﺸﺘﺮ ﺍﺳﺖ ﻭ‬
‫ﺣﺪﺍﻗﻞ ﺑﺮﺍﺑﺮ ‪ 0‬ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﺩﻭ ﺣﺎﻟﺖ ﺯﻳﺮ ﺭﺍ ﺑﺮﺍﻱ ﺗﻌﻴﻴﻦ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬

‫ﺣﺎﻟﺖ ﺍﻭﻝ‪ |x| ≥ xmin :‬ﻭ ﺑﺎ ﺑﻪ ﻃﻮﺭ ﻣﻌﺎﺩﻝ ] ‪.x ∈ [−xmax , −xmin ] ∪ [+xmin , +xmax‬‬

‫ﺣﺎﻟﺖ ﺩﻭﻡ‪ ،|x| < xmin :‬ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ] ‪.x ∈ [−xmin , +xmax‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﺍ ﺩﺭ ﺍﻳﻦ ﺩﻭ ﺣﺎﻟﺖ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫• ﺣﺎﻟﺖ ﺍﻭﻝ‪ :‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﻱ ] ‪x ∈ [−xmax , −xmin ] ∪ [+xmin , +xmax‬‬
‫ﻭﻗﺘﻲ ‪ x‬ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺸﻲ ﺑﻪ ﺻﻮﺭﺕ )‪ (۱۲.۲‬ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺑﺎﺯﻩ ] ‪[−xmax , −xmin ] ∪ [+xmin , +xmax‬‬
‫ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ .0.d−1 · · · d−m d−m−1 · · · ≥ b−1‬ﻟﺬﺍ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫|)‪|x − rd0 (x‬‬ ‫‪β −m‬‬
‫= )̃‪ER (x‬‬
‫|‪|x‬‬
‫<‬
‫‪.‬‬
‫· · · ‪0 d−1 · · · d−m d−m−1‬‬
‫‪β −m‬‬
‫‪< −1 = β 1−m‬‬
‫‪β‬‬
‫ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‪ ،‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ‪ ،‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫|)‪|x − rd0 (x‬‬
‫‪< β 1−m‬‬
‫|‪|x‬‬
‫ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ‪ ،‬ﺍﮔﺮ ] ‪ x ∈ [−xmax , −xmin ] ∪ [+xmin , +xmax‬ﺁﻥﮔﺎﻩ‪ ،‬ﺩﺭ ﻣﻮﺭﺩ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ‪/‬ﭘﺎﻳﻴﻦ‬
‫ﻭ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫|)‪|x − rdd (x‬‬ ‫|)‪|x − rdu (x‬‬ ‫|)‪|x − rdr (x‬‬ ‫‪1‬‬
‫‪< β 1−m ,‬‬ ‫‪< β 1−m ,‬‬ ‫‪< β 1−m‬‬
‫|‪|x‬‬ ‫|‪|x‬‬ ‫|‪|x‬‬ ‫‪2‬‬
‫• ﺣﺎﻟﺖ ﺩﻭﻡ‪ :‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﻱ ) ‪x ∈ (−xmin , +xmin‬‬
‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ) ‪ x ∈ (−xmin , +xmin‬ﻧﻤﻲﺗﻮﺍﻥ ﺑﻪ ﻧﺘﺎﻳﺞ ﺣﺎﻟﺖ ﺍﻭﻝ ﺭﺳﻴﺪ‪ .‬ﭼﺮﺍ ﻛﻪ‬
‫ﻣﺎﻧﺘﻴﺲ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ x‬ﺩﻳﮕﺮ ﺍﺯ ﭘﺎﻳﻴﻦ ﺑﻪ ‪ β −1‬ﻛﺮﺍﻧﺪﺍﺭ ﻧﻴﺴﺖ ﻭ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺴﻴﺎﺭ ﻛﻮﭼﻚ ﺷﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ‬
‫ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻣﻲﺗﻮﺍﻧﺪ ﺗﺎ ‪ 1‬ﺍﻓﺰﺍﻳﺶ ﻳﺎﺑﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺍﮔﺮ ﺑﺮﺧﻲ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ )ﻭ ﻧﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴۴‬‬

‫ﻫﻤﻪ ﺁﻥﻫﺎ( ﻣﻮﺟﻮﺩ ﺩﺭ ) ‪ x ∈ (−xmin , +xmin‬ﺭﺍ ﺭﻭﻧﺪ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺮﺍﺑﺮ ‪1‬‬
‫ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦ ) ‪ ،FS (β, m, emin , emax‬ﺍﮔﺮ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺻﻔﺮ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬
‫‪ x = 12 β emin −m‬ﺑﻪ ‪ 0‬ﺭﻭﻧﺪ ﻣﻲﺷﻮﺩ‪ .‬ﻳﻌﻨﻲ ‪ .x̃ := rd0 (x) = 0‬ﺍﻛﻨﻮﻥ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪1 emin −m‬‬
‫|̃‪|x − x‬‬ ‫‪2β‬‬ ‫‪−‬‬ ‫‪0‬‬
‫= )̃‪ER (x‬‬ ‫=‬ ‫‪=1‬‬
‫|‪|x‬‬ ‫‪1 emin −m‬‬
‫‪2β‬‬
‫= ‪ x‬ﺭﺥ ﻣﻲﺩﻫﺪ‪.‬‬ ‫‪1 emin −m‬‬
‫‪2β‬‬ ‫ﻳﻌﻨﻲ ﺻﺪ ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ‬

‫■ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ )ﺩﻗﺖ ﻣﺎﺷﻴﻦ(‬

‫‪β‬‬ ‫‪1−m‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ] ‪ [−xmax , −xmin ] ∪ [+xmin , +xmax‬ﺣﺪﺍﻛﺜﺮ ﺑﺮﺍﺑﺮ‬
‫ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻭﻟﻲ ﺩﺭ ﺑﺎﺯﻩ ) ‪ (−xmin , +xmin‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺗﻲ‪ ،‬ﺩﺭ ﻗﺴﻤﺖ ﺑﺰﺭﮔﻲ ﺍﺯ ﺑﺎﺯﻩ‬
‫] ‪ [−xmin , +xmax‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﺑﺮ ‪ β 1−m‬ﺍﺳﺖ ﻭ ﺩﺭ ﻗﺴﻤﺖ ﻛﻮﭼﻜﻲ ﺍﺯ ﺍﻳﻦ ﺑﺎﺯﻩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‬
‫ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ‪ ،‬ﻋﺪﺩ ‪ β m−1‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻛﻪ ﺑﺮﺍﻱ ﻗﺴﻤﺖ ﺑﺰﺭﮔﻲ‬
‫‪ [−x‬ﻣﺤﻘﻖ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﺩﺑﻴﺎﺕ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﻋﺪﺩ ‪ β 1−m‬ﺑﻪ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ‪ ۱۳‬ﻭ ﻳﺎ ﺩﻗﺖ ﻣﺎﺷﻴﻦ ‪۱۴‬‬
‫ﺍﺯ ﺑﺎﺯﻩ ] ‪min , +xmax‬‬

‫ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﺍﭘﺴﯿﻠﻮﻥ ﻣﺎﺷﯿﻦ)ﺩﻗﺖ ﻣﺎﺷﯿﻦ(‬ ‫ﺗﻌﺮﯾﻒ ‪۵−۲‬‬
‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ) ‪ ،FS (β, m, emin , emax‬ﻋﺪﺩ ‪ β 1−m‬ﺑﻪ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻭ ﻳﺎ ﺩﻗﺖ ﻣﺎﺷﻴﻦ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺍﭘﺴﻴﻠﻮﻥ‬
‫ﻣﺎﺷﻴﻦ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﺳﺖ‪.‬‬
‫ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ eps‬ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ‪ ،‬ﻳﻌﻨﻲ ‪ ،β 1−m‬ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ‪ .‬ﻫﻤﻴﻨﻄﻮﺭ ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ ﻓﺎﺻﻠﻪ ﻋﺪﺩ ‪ 1‬ﺗﺎ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‬
‫ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪ ﺍﺯ ﺁﻥ ﺑﺮﺍﺑﺮ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺍﺳﺖ‪ .‬ﻟﺬﺍ‪ ،‬ﺑﺮﺧﻲ ﻣﻨﺎﺑﻊ ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﻳﻚ ﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪ ﺍﺯ ﺁﻥ ﺭﺍ ﺑﻪ‬
‫ﻋﻨﻮﺍﻥ ﺗﻌﺮﻳﻒ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻧﺪ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﻣﺎﺷﯿﻦ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﯿﻨ )‪(FN (2, 3, −2, 2‬‬ ‫ﻣﺜﺎﻝ ‪ )۶−۲‬ﺍﭘﺴﯿﻠﻮﻥ ﻣﺎﺷﯿﻦ ﺩﺭ ﯾ‬
‫ﻳﻚ ﻣﺎﺷﻴﻦ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪ FN (2, 3, −2, 2‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻃﺒﻖ ﺗﻌﺮﻳﻒ‪ ،‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺮﺍﺑﺮ‬
‫‪ β 1−m = 21−3 = 0.25‬ﺍﺳﺖ‪ .‬ﺍﺯ ﺳﻮﻱ ﺩﻳﮕﺮ‪ ،‬ﻋﺪﺩ ‪ 1‬ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶ ‪ 0.100 × 21‬ﻣﻲﺑﺎﺷﺪ ﻭ ﻋﺪﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﺑﻼﻓﺎﺻﻠﻪ ﺑﻌﺪ ﺍﺯ ﺁﻥ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ .0.101 × 21 = 1.25‬ﻓﺎﺻﻠﻪ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﺑﺮﺍﺑﺮ ‪ 0.001 × 21 = 0.25‬ﺍﺳﺖ‪،‬‬
‫ﻛﻪ ﻫﻤﺎﻥ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ‪ ،‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ )‪ FN (2, 3, −2, 2‬ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ‬
‫‪eps‬‬
‫)‪FN (2, 3, −2, 2‬‬
‫‪− 0.125‬‬

‫‪+ 0.125‬‬

‫‪− 3.5‬‬ ‫‪+ 3.5‬‬

‫‪1.00 1.25‬‬

‫ﻣﻌﻤﻮﻻ‪ ،‬ﺍﺯ ﻭﺍﮊﻩ ”ﺍﭘﺴﻴﻠﻮﻥ“ ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﺍﻋﺪﺍﺩ ﻛﻮﭼﻚ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ‪ ،‬ﺑﺮﺧﻼﻑ ﺗﺼﻮﺭﻱ ﻛﻪ ﺍﺯ ﻧﺎﻣﮕﺬﺍﺭﻱ ﺑﺮﺩﺍﺷﺖ‬
‫ﻣﻲﺷﻮﺩ‪ ،‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ﻣﻮﺟﻮﺩ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ﻧﻴﺴﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺩﺭ ﻣﺜﺎﻝ ﻓﻮﻕ‪،‬‬
‫‪13 Machine‬‬ ‫‪Epsilon‬‬ ‫‪14 Machine‬‬ ‫‪precision‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۴۵‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺮﺍﺑﺮ ‪ eps = 0.25‬ﺍﺳﺖ ﻭ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺜﺒﺖ ‪ xmin = 0.125‬ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ‬
‫ﻣﺎﺷﻴﻦ ‪.eps > xmin‬‬
‫ﺩﻳﺪﻳﻢ ﻛﻪ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻣﺜﺒﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﭼﺮﺍ‬
‫ﻭﺍﮊﻩ ﺍﭘﺴﻴﻠﻮﻥ ﺩﺭ ﺍﻳﻦ ﻧﺎﻣﮕﺬﺍﺭﻱ ﺍﺳﺘﻔﺎﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦ ) ‪ ،FS (β, m, emin , emax‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ‬
‫ﻛﻮﭼﻚﺗﺮﻳﻦ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﺳﺖ ﻭ ﺍﻧﺘﻈﺎﺭ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﻛﻤﺘﺮ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‬
‫ﻧﺎﻣﻌﻘﻮﻝ ﻭ ﻧﺸﺪﻧﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﻣﺎﻫﻴﺖ ﺍﻋﺪﺍﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ) ‪ FS (β, m, emin , emax‬ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﺳﺖ ﻛﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ‬
‫ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﻧﻤﻲﺗﻮﺍﻧﺪ ﻛﻤﺘﺮ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺩﺭ ﺗﺤﻠﻴﻞ ﺧﻄﺎ‪ ،‬ﺍﮔﺮ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﻣﺎﺷﻴﻦ ﺗﻘﺮﻳﺒﺎ ﺑﻪ‬
‫ﺍﻧﺪﺍﺯﻩ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺁﻥ ﺧﻄﺎ ﻛﻮﭼﻚ )ﺍﭘﺴﻴﻠﻮﻥ( ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺗﻲ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻﻱ‬
‫ﺍﻳﺪﻩﺁﻝ ﻭ ﻣﻌﻘﻮﻝ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﺗﻮﺿﻴﺤﺎﺕ‪ ،‬ﻭﺍﮊﻩ ﺩﻗﺖ ﻣﺎﺷﻴﻦ ﻧﻴﺰ ﻣﻨﺎﺳﺐ ﺑﻪ ﻧﻈﺮ‬
‫ﻣﻲﺭﺳﺪ‪ .‬ﺍﻣﺎ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﻴﺸﺘﺮ ﺭﺍﻳﺞ ﻣﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﺩﺭ ﺍﻳﻦ ﻛﺘﺎﺏ ﺍﺯ ﺍﻳﻦ ﻭﺍﮊﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ‬
‫ﺩﺭ ﻗﺴﻤﺖ ﺑﺰﺭﮔﻲ ﺍﺯ ﺑﺎﺯﻩ ] ‪ ،[−xmax , +xmax‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺍﺳﺖ‪ ،‬ﻛﻪ ﺍﻳﺪﻩﺁﻝ ﻭ ﻣﻌﻘﻮﻝ‬
‫ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺎﺯﻩ ) ‪ (−xmin , +xmin‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺷﻮﺩ ﻭ ﺣﺘﻲ ﺑﺮﺍﺑﺮ ‪ 1‬ﺷﻮﺩ )ﺻﺪ‬
‫ﺩﺭﺻﺪ ﺧﻄﺎ( ﻛﻪ ﺍﺗﻔﺎﻕ ﻧﺎﻣﻨﺎﺳﺐ ﻭ ﻏﻴﺮﺍﻳﺪﻩﺁﻝ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻣﺎﺷﻴﻦﻫﺎ ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ﻳﻚ ﻋﺪﺩ ﺩﺭ ﺑﺎﺯﻩ ) ‪(−xmin , +xmin‬‬
‫ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ‪ ۱۵‬ﺭﺍ ﮔﺰﺍﺭﺵ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ‪ ،‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﺰﺭﮒ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﻫﺮﮔﺎﻩ‬
‫ﻋﺪﺩﻱ ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ ﺭﻭﻧﺪ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﮔﺰﺍﺭﺵ ﻣﻲﺷﻮﺩ ﻭ ﺍﻳﻦ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﺑﻪ ﻋﻨﻮﺍﻥ ﻫﺸﺪﺍﺭﻱ ﺍﺳﺖ ﻛﻪ‬
‫ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻧﺪﺍﺯﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺰﺭﮒ ﺑﺎﺷﺪ ﻭ ﺩﻗﺖ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺩﺭ ﺍﻧﺪﺍﺯﻩ ﻭ ﻣﺮﺗﺒﻪ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻧﺒﺎﺷﺪ‪.‬‬

‫■ ﻣﺤﺪﻭﺩﻩﻫﺎﻱ ﺳﺮﺭﻳﺰﻱ‪ ،‬ﺯﻳﺮﺭﻳﺰﻱ ﻭ ﻧﺮﻣﺎﻝ ﺑﺮﺍﻱ ﻳﻚ ﻣﺎﺷﻴﻦ‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻮﺿﻴﺤﺎﺕ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ R‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺳﻪ ﺑﺨﺶ ﺗﻘﺴﻴﻢ ﻛﺮﺩ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻣﺎﻫﻴﺖ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺩﺭ ﺍﻳﻦ‬
‫ﺑﺨﺶﻫﺎ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﻣﺠﻤﻮﻋﻪ ‪x‬ﻫﺎﻳﻲ ﻛﻪ ‪ |x| > xmax‬ﺑﻪ ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ ‪ ۱۶‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺑﺎﺯﻩ ) ‪(−xmin , +xmin‬‬
‫ﺑﺎ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ‪ ۱۷‬ﻧﺎﻣﮕﺬﺍﺭﻱ ﻣﻲﺷﻮﺩ‪ .‬ﻣﺠﻤﻮﻋﻪ ] ‪ [−xmax , −xmin ] ∪ [+xmin , +xmax‬ﺑﻪ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ‪ ۱۸‬ﻣﻮﺳﻮﻡ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ ﺍﻳﻦ ﺳﻪ ﻧﺎﺣﻴﻪ ﺑﻪ ﺗﺼﻮﻳﺮ ﻛﺸﻴﺪﻩ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۷−۲‬ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‪ ،‬ﺯﯾﺮﺭﯾﺰﯼ ﻭ ﺳﺮﺭﯾﺰﯼ ﺑﺮﺍﯼ ﻣﺎﺷﯿﻦ ﻣﺒﺘﻨ ﺑﺮ )‪( FS (2, 3, −1, 1‬‬
‫ﻣﺎﺷﻴﻦ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪ FS (2, 3, −1, 1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ‪ xmin = 0.25‬ﻭ ‪ .xmax = 1.75‬ﺑﺮﺍﻱ ﺍﻳﻦ‬
‫ﻣﺎﺷﻴﻦ‪ ،‬ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺑﻪ ﺳﻪ ﺩﺳﺘﻪ ﺯﻳﺮ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﻧﺪ‬
‫‪ :‬ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‬ ‫‪[−1.75, −0.25] ∪ [+0.25, +1.75],‬‬
‫‪ :‬ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ‬ ‫‪(−0.25, +0.25),‬‬
‫‪ :‬ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ‬ ‫‪(−∞, −1.75) ∪ (1.75, +∞).‬‬
‫ﺍﻳﻦ ﻣﺤﺪﻭﺩﻩﻫﺎ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺭﺳﻢ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫‪15 Underflow‬‬ ‫‪17 Underflow‬‬ ‫‪region‬‬

‫‪16 Overflow‬‬ ‫‪region‬‬ ‫‪18 Normal‬‬ ‫‪region‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴۶‬‬

‫‪−xmax‬‬ ‫‪xmax‬‬
‫‪−xmin‬‬ ‫‪xmin‬‬
‫)‪FS (2, 3, −1, 1‬‬
‫‪− 1.75‬‬

‫‪+ 1.75‬‬
‫‪− 1.0‬‬

‫‪− 0.5‬‬

‫‪− 0.25‬‬

‫‪+ 0.25‬‬

‫‪+ 0.5‬‬

‫‪+ 1.0‬‬
‫ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‬ ‫ﻣﺤﺪﻭﺩﻩ‬ ‫ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‬
‫ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ‬ ‫)ﻗﺴﻤﺖ ﻣﻨﻔﻲ(‬ ‫ﺯﻳﺮﺭﻳﺰﻱ‬ ‫)ﻗﺴﻤﺖ ﻣﺜﺒﺖ(‬ ‫ﻣﺤﺪﻭﺩﻩ ﺳﺮﺭﻳﺰﻱ‬
‫)ﻗﺴﻤﺖ ﻣﻨﻔﻲ(‬ ‫)ﻗﺴﻤﺖ ﻣﺜﺒﺖ(‬

‫◀ ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻋﺪﺩ ‪ a = 3.14‬ﺭﺍ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ .‬ﭼﻮﻥ ‪ 3.14 > x max‬ﻟﺬﺍ ﺩﺭ ﺧﻴﺮﻩ ﺍﻳﻦ‬
‫ﻋﺪﺩ ﺧﻄﺎﻱ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪.‬‬
‫◀ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ b = 1.2‬ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺍﺯ ﺳﻮﻱ ﺩﻳﮕﺮ‪ ،‬ﺍﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﻧﻴﺴﺖ‪ .‬ﻟﺬﺍ ﺍﻳﻦ ﻋﺪﺩ‬
‫ﺑﻌﺪ ﺍﺯ ﺭﻭﻧﺪ ﺷﺪﻥ ﺩﺭ ﻣﺎﺷﻴﻦ ﻗﺎﺑﻞ ﺫﺧﻴﺮﻩ ﺧﻮﺍﻫﺪ ﺑﻮﺩ ﻭ ﺑﺎ ﺍﻃﻤﻴﻨﺎﻥ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻧﺎﺷﻲ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺍﻳﻦ ﻋﺪﺩ‬
‫ﺣﺪﺍﻛﺜﺮ ﺑﺮﺍﺑﺮ ‪ eps = 21−3 = 0.25‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫◀ ﺣﺎﻝ ﺩﻭ ﻋﺪﺩ ‪ c = 0.245‬ﻭ ‪ d = 0.01‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﺮ ﺩﻭﻱ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ‪ .‬ﻟﺬﺍ‬
‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻧﺎﺷﻲ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺁﻥﻫﺎ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺧﺼﻮﺹ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ‬
‫ﺍﮔﺮ ﺍﺯ ﺭﻭﻧﺪ ﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫|)‪|c − rdr (c‬‬ ‫|‪|0.245 − 0.25‬‬
‫=‬ ‫‪= 0.020408 < eps,‬‬
‫|‪|c‬‬ ‫|‪|0.245‬‬
‫|)‪|d − rdr (d‬‬ ‫|‪|0.01 − 0.0‬‬
‫=‬ ‫‪= 1 > eps.‬‬
‫|‪|d‬‬ ‫|‪|0.01‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻧﺎﺷﻲ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ‪ c‬ﻛﻤﺘﺮ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺩﺭ ﻣﻮﺭﺩ ‪ d‬ﺍﻳﻦ ﺧﻄﺎ‬
‫ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺍﺳﺖ ﻭ ﺍﺻﻮﻻ ﺧﻴﻠﻲ ﺑﺰﺭﮒ ﺍﺳﺖ‪.‬‬

‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﻣﺎﺷﻴﻦ ) ‪ ،FS (β, m, emin , emax‬ﻫﺮﭼﻪ ‪ emax‬ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ﺑﺰﺭﮒﺗﺮ ﺧﻮﺍﻫﺪ‬
‫ﺑﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻫﺮ ﭼﻪ ‪ emin‬ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﻛﺎﻩ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻛﻮﭼﻚﺗﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬

‫■ ﺟﻤﻊﺑﻨﺪﻱ‬
‫ﺩﺭ ﻣﺎﺷﻴﻦ ) ‪ ،FS (β, m, emin , emax‬ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺑﺮﺍﺑﺮ ‪ β 1−m‬ﺍﺳﺖ ﻭ ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺍﻳﺪﻩﺁﻝ ﺑﺮﺍﻱ ﺧﻄﺎ ﺩﺭ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‬
‫ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﻮﺭﺩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ‪ ،‬ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺭﺍ ﺩﺍﺭﻳﻢ‪.‬‬

‫ﻧﺘﯿﺠﻪ ‪۶−۲‬‬
‫ﮐﺮﺍﻥ ﺑﺎﻻﯼ ﺧﻄﺎﯼ ﻧﺴﺒﯽ ﺭﻭﻧﺪﮐﺮﺩﻥ‬
‫• ﺍﮔﺮ ‪ x‬ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ] ‪ [−xmax , −xmin ] ∪ [+xmin , +xmax‬ﻭﺍﻗﻊ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﻧﺪ ﺷﺪﻩ ﻋﺪﺩ ‪ x‬ﻳﻚ ﻋﺪﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﻧﺮﻣﺎﻝ ﺍﺳﺖ ﻭ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺮﺍﺑﺮ ‪ 21 eps‬ﻣﻲﺑﺎﺷﺪ‬
‫ﻭ ﺩﺭ ﺳﻪ ﻧﻮﻉ ﺩﻳﮕﺮ ﺭﻭﻧﺪ ﻛﺮﺩﻥ‪،‬ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﺑﺮ ‪ eps‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• ﺍﮔﺮ ‪ x‬ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ) ‪ (−xmin , +xmin‬ﻭﺍﻗﻊ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﻭﻧﺪﻛﺮﺩﻥ ‪x‬‬
‫ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﺷﻮﺩ‪ .‬ﻟﺬﺍ ﻫﻨﮕﺎﻡ ﺫﺧﻴﺮﻩ ‪ x‬ﺩﺭ ﻣﺎﺷﻴﻦ‪ ،‬ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۴۷‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﻳﻚ ﻋﺪﺩ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‪ ،‬ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻋﺪﺩ ﺑﺴﺘﮕﻲ‬
‫ﻧﺪﺍﺭﺩ ﻭ ﻓﻘﻂ ﺑﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﻣﺎﻧﺘﻴﺲ ﻭﺍﺑﺴﺘﻪ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﻣﻮﺿﻮﻉ ﺍﺯ ﻣﺰﺍﻳﺎﻱ ﻣﺎﺷﻴﻦﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺳﻴﺴﺘﻢ ﺍﻋﺪﺍﺩ ﺑﺎ ﻣﻤﻴﺰ‬
‫ﺷﻨﺎﻭﺭ ﻣﻲﺑﺎﺷﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺑﻪ ﻟﺤﺎﻅ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ‪ ،‬ﺑﻴﻦ ﻛﻠﻴﻪ ﺍﻋﺪﺍﺩ ﻣﻮﺟﻮﺩ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ﻋﺪﺍﻟﺖ ﻭ‬
‫ﻳﻚﭘﺎﺭﭼﮕﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬

‫ﺧﻄﺎﻱ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ)ﺧﻄﺎﻱ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ(‬ ‫‪۳.۲‬‬

‫ﺩﺭ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ‪ ،‬ﻃﻮﻝ ﻣﺎﻧﺘﻴﺲ ﻭ ﻣﺤﺪﻭﺩﻩﻱ ﻧﻤﺎ ﻧﺎﻣﺘﻨﺎﻫﻲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻓﺼﻞ ﺍﻭﻝ‬
‫ﺫﻛﺮ ﺷﺪ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺮﺍﻱ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﻣﺎﻧﺘﻴﺲ ﻭ ﻧﻤﺎ ﻣﺤﺪﻭﺩﻳﺖ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻭ ﺍﻳﻦ ﻣﺤﺪﻭﺩﻳﺖ ﻣﻨﺠﺮ ﺑﻪ ﺁﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺘﻴﺠﻪ‬
‫ﺟﻤﻊ‪/‬ﺗﻔﺮﻳﻖ‪/‬ﺿﺮﺏ‪/‬ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﻧﻮﻉ ﺧﻄﺎ ﺑﻪ ﺧﻄﺎﻱ ﺟﻤﻊ‪/‬ﺗﻔﺮﻳﻖ‪/‬ﺿﺮﺏ‪/‬ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ ﻣﻮﺳﻮﻡ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﺧﻄﺎﯼ ﺟﻤ ̸ﺗﻔﺮﯾﻖ̸ﺿﺮﺏ̸ﺗﻘﺴﯿﻢ ﻣﺎﺷﯿﻨ )ﺧﻄﺎﯼ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ(‬ ‫ﺗﻌﺮﯾﻒ ‪۷−۲‬‬
‫ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﮔﺮ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ‪/‬ﺗﻔﺮﻳﻖ‪/‬ﺿﺮﺏ‪/‬ﺗﻘﺴﻴﻢ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺘﻴﺠﻪ ﺩﻗﻴﻖ ﻧﺒﺎﺷﺪ ﻭ ﺑﺎ‬
‫ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﺑﺎﺷﺪ‪ .‬ﭼﻨﻴﻦ ﺧﻄﺎﻳﻲ ﻛﻪ ﻧﺎﺷﻲ ﺍﺯ ﻣﺤﺪﻭﺩﻳﺖ ﺣﺎﻓﻈﻪ ﺍﺳﺖ‪ ،‬ﺑﻪ ﺧﻄﺎﻱ ﺟﻤﻊ‪/‬ﺗﻔﺮﻳﻖ‪/‬ﺿﺮﺏ‪/‬ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ ﻣﻮﺳﻮﻡ‬
‫ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x‬ﻭ ‪ y‬ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﻨﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺍﻳﻦ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ x ⊕ y‬ﻧﻤﺎﻳﺶ‬
‫ﻣﻲﺩﻫﻴﻢ‪ .‬ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ‪ x + y‬ﻣﻲﺗﻮﺍﻧﺪ ﺑﺎ ‪ x ⊕ y‬ﻣﺘﻔﺎﻭﺕ ﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﺭﺍ ﺑﺎ ﺧﻄﺎﻱ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬
‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺩﻭ ﻋﺪﺩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫|)‪|(x + y) − (x ⊕ y)| = |(x + y) − rd(x + y‬‬
‫ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x‬ﻭ ‪ y‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫|)‪|(x + y) − (x ⊕ y‬‬ ‫|)‪|(x + y) − rd(x + y‬‬
‫=‬
‫|‪|x + y‬‬ ‫|‪|x + y‬‬
‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﻣﺮﺑﻮﻁ ﺑﻪ ﺗﻘﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﺎﺷﻴﻨﻲ ﻧﻴﺰ ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺳﻨﺠﺶ ﺧﻄﺎﻱ ﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﻛﺎﻓﻲ ﺍﺳﺖ ﺑﻪ ﺍﻳﻦ ﻧﻜﺘﻪ ﺗﻮﺟﻪ ﻛﻨﻴﻢ ﻛﻪ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞﺟﻤﻊ ﻣﺎﺷﻴﻨﻲ‬
‫ﺑﺮﺍﺑﺮ ﺭﻭﻧﺪ ﺷﺪﻩ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺣﺎﺻﻞﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﻗﺎﻋﺪﻩ ﺩﺭ ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻫﻢ ﺻﺎﺩﻕ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬
‫ﺧﻄﺎﻱ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﻣﺎﺷﻴﻨﻲ ﻫﻤﺎﻥ ﺧﻄﺎﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﻭ ﻳﺎ ﺧﻄﺎﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻧﺘﻴﺠﻪ ‪ ۶-۲‬ﺩﻳﺪﻳﻢ ﻛﻪ ﺍﮔﺮ‬
‫ﻋﺪﺩﻱ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ 12 eps‬ﺍﺳﺖ ﻭ ﺩﺭ‬
‫ﺑﻘﻴﻪ ﺭﻭﻧﺪﻛﺮﺩﻥﻫﺎ‪ ،‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﺑﺮ ‪ eps‬ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﻪ ﺳﺎﺩﮔﻲ ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺩﺭ ﻣﻮﺭﺩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ‬
‫ﻧﺴﺒﻲ ﺭﻭﻧﺪﺁﻑ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺧﻄﺎﯼ ﻧﺴﺒﯽ ﺟﻤ ‪ ،‬ﺗﻔﺮﯾﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﯿﻢ ﻣﺎﺷﯿﻨ‬ ‫ﻧﺘﯿﺠﻪ ‪۸−۲‬‬

‫ﺩﺭ ﻳﻚ ﻣﺎﺷﻴﻦ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ ) ‪ FS (β, m, emin , emax‬ﻛﻪ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﻧﺰﺩﻳﻚﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪،‬‬

‫• ﺍﮔﺮ ‪ x‬ﻭ ‪ y‬ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﺪ ﻛﻪ ﺣﺎﺻﻞﺟﻤﻊ ﺁﻥﻫﺎ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ] ‪[−xmax , −xmin ] ∪ [xmin , xmax‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۴۸‬‬

‫ﻭﺍﻗﻊ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬

‫|)‪|(x + y) − (x ⊕ y‬‬
‫=‪ER (x ⊕ y) :‬‬ ‫‪≤ 12 eps.‬‬ ‫)‪(۱۳.۲‬‬
‫|‪|x + y‬‬
‫• ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺩﻭ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ x‬ﻭ ‪ ،y‬ﺍﮔﺮ ﻧﺘﻴﺠﻪ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ‬
‫] ‪ [−xmax , −xmin ] ∪ [xmin , xmax‬ﻭﺍﻗﻊ ﺑﺎﺷﺪ‪ ،‬ﺩﺍﺭﻳﻢ‪:‬‬
‫|)‪|(x − y) − (x ⊖ y‬‬
‫=‪ER (x ⊖ y) :‬‬ ‫‪≤ 12 eps,‬‬ ‫)‪(۱۴.۲‬‬
‫|‪|x − y‬‬
‫|)‪|(x ∗ y) − (x ⊛ y‬‬
‫=‪ER (x ⊛ y) :‬‬ ‫‪≤ 12 eps,‬‬ ‫)‪(۱۵.۲‬‬
‫|‪|x ∗ y‬‬
‫|)‪|(x/y) − (x ⊘ y‬‬
‫=‪ER (x ⊘ y) :‬‬ ‫‪≤ 21 eps.‬‬ ‫)‪(۱۶.۲‬‬
‫|‪|x/y‬‬
‫• ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ﺩﺭ ﻣﺎﺷﻴﻦ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ‪ ،‬ﭘﺎﻳﻴﻦ ﻭ ﺻﻔﺮ ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ‬
‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﺑﺮ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ﺩﺭ ﻣﺎﺷﻴﻦ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﺑﻪ ﺳﻤﺖ ﺑﺎﻻ‪ ،‬ﭘﺎﻳﻴﻦ ﻭ‬
‫ﺻﻔﺮ ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺳﻤﺖ ﺭﺍﺳﺖ ﻧﺎﻣﺴﺎﻭﻱﻫﺎﻱ ﻓﻮﻕ ﺍﺯ ‪ 12 eps‬ﺑﻪ ‪ eps‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ ﻣﻨﺪﺭﺝ ﺩﺭ ﻧﺘﻴﺠﻪ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺭﻋﺎﻳﺖ ﻧﻤﻲﮔﺮﺩﻳﺪ‪ .‬ﺍﻣﺎ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪IEEE‬‬
‫ﺭﻋﺎﻳﺖ ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ )‪ (۱۶.۲)-(۱۳.۲‬ﺭﺍ ﺍﻟﺰﺍﻣﻲ ﻧﻤﻮﺩﻩ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺍﻣﺮﻭﺯﻱ ﻛﻪ ﺍﺯ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ‪ IEEE‬ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﻛﻨﻨﺪ‪ ،‬ﻧﺘﻴﺠﻪ ‪ ۸-۲‬ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ‪.‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ﺣﺎﺻﻞ ﺟﻤﻊ‪/‬ﺗﻔﺮﻳﻖ‪/‬ﺿﺮﺏ‪/‬ﺗﻘﺴﻴﻢ ﺩﻭ ﻋﺪﺩ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ) ‪ (−xmin , +xmin‬ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ‬
‫ﻧﺴﺒﻲ ﺣﺎﺻﻞ ﻋﻤﻞ ﻟﺰﻭﻣﺎ ﺍﺯ ﺍﭘﺴﻴﻠﻮﻥ ﻣﺎﺷﻴﻦ ﻛﻮﭼﻚﺗﺮ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﺍﺗﻔﺎﻕ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺗﺤﻠﻴﻞ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺧﻮﺷﺎﻳﻨﺪ ﻭ ﻣﻄﻠﻮﺏ‬
‫ﻧﻴﺴﺖ‪ ،‬ﻟﺬﺍ ﻫﺮﻣﻮﻗﻊ ﻛﻪ ﻧﺘﻴﺠﻪ ﻳﻚ ﻣﺤﺎﺳﺒﻪ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ) ‪ (−xmin , +xmin‬ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱﮔﺰﺍﺭﺵ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﺑﻪ ﻧﻮﻋﻲ ﻳﻚ ﻫﺸﺪﺍﺭ ﺍﺳﺖ ﻛﻪ ﺍﻋﻼﻡ ﻣﻲﻛﻨﺪ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺘﺎﻳﺞ ﺑﺎ ﺧﻄﺎﻱ ﻏﻴﺮﻣﺘﻌﺎﺭﻑ ﻫﻤﺮﺍﻩ‬
‫ﺑﺎﺷﺪ‪.‬‬

‫ﺧﻄﺎﻱ ﻧﺎﺷﻲ ﺍﺯ ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ‬ ‫‪۴.۲‬‬

‫ﺩﺭ ﺑﺮﺧﻲ ﻣﺴﺎﻳﻞ‪ ،‬ﻳﻚ ﻳﺎ ﭼﻨﺪ ﺩﺍﺩﻩ ﻭﺭﻭﺩﻱ ﺩﺍﺭﻳﻢ ﻭ ﺭﻭﻱ ﺍﻳﻦ ﺩﺍﺩﻩﻫﺎ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻭ ﻳﻚ ﻳﺎ ﭼﻨﺪ ﻧﺘﻴﺠﻪ ﺧﺮﻭﺟﻲ‬
‫ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﻋﻤﻮﻣﺎ‪ ،‬ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﻛﻤﻴﺖﻫﺎﻳﻲ ﻫﺴﺘﻨﺪ ﻛﻪ ﺍﺯ ﺭﻭﻧﺪﻛﺮﺩﻥ ﻭ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺣﺎﺻﻞ ﻣﻲﺷﻮﻧﺪ ﻭ ﻟﺬﺍ ﺁﻟﻮﺩﻩ‬
‫ﺑﻪ ﺧﻄﺎ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺗﺒﻌﺎ ﺧﻄﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ﺍﻳﻦ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺑﻪ ﺧﺮﻭﺟﻲ ﻣﺴﺎﻟﻪ ﺳﺮﺍﻳﺖ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻳﻦ ﻧﻮﻉ ﺧﻄﺎ ﺑﻪ‬
‫ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻧﺎﺷﻲ ﺍﺯ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﻭ ﻳﺎ ﺑﻪ ﺍﺧﺘﺼﺎﺭ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺴﺎﺣﺖ ﻳﻚ ﻛﺮﻩ ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﻣﺴﺎﺣﺖ ﻳﻚ ﻛﺮﻩ ﺍﺯ‬
‫ﻓﺮﻣﻮﻝ ‪ s = 4πr2‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ‪ r‬ﺷﻌﺎﻉ ﻛﺮﻩ ﺍﺳﺖ‪ .‬ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﻋﺪﺩ ‪ π‬ﻳﻚ ﻋﺪﺩ ﻧﺎﻣﺨﺘﻮﻡ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻧﻤﻲﺗﻮﺍﻥ‬
‫ﺁﻥ ﺭﺍ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻣﺴﺎﺣﺖ ﻛﺮﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ ،‬ﺑﻠﻜﻪ ﻣﻲﺗﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺁﻥ ﻣﺎﻧﻨﺪ ̃‪ π‬ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺷﻌﺎﻉ ﻛﺮﻩ ﺭﺍ ﺑﺎﻳﺪ ﺑﺎ‬
‫ﻳﻚ ﻭﺳﻴﻠﻪ ﻣﺎﻧﻨﺪ ﺧﻂﻛﺶ ﺍﻧﺪﺍﺯﻩ ﺑﮕﻴﺮﻳﻢ ﻭ ﻃﺒﻴﻌﺘﺎ ﺷﻌﺎﻉ ﻛﺮﻩ ﺭﺍ ﻧﻤﻲﺗﻮﺍﻧﻴﻢ ﺩﻗﻴﻖ ﺍﻧﺪﺍﺯﻩ ﺑﮕﻴﺮﻳﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺷﻌﺎﻉ ﻛﺮﻩ‬
‫ﻣﺎﻧﻨﺪ ̃‪ r‬ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻣﺴﺎﺣﺖ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻣﺴﺎﺣﺖ ﻛﺮﻩ‪ ،‬ﺍﺯ ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ π‬ﻭ ‪ ،r‬ﻳﻌﻨﻲ ̃‪ π‬ﻭ ̃‪r‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﻧﻪ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺁﻥﻫﺎ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﻪ ﺟﺎﻱ ﻣﺤﺎﺳﺒﻪ ‪ s = 4πr2‬ﻣﻘﺪﺍﺭ ‪ s̃ = 4π̃r̃2‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺗﺒﻌﺎ ﺩﺭ‬
‫ﺍﻳﻦ ﻣﺤﺎﺳﺒﻪ ﺧﻄﺎ ﻣﺮﺗﻜﺐ ﺷﺪﻩﺍﻳﻢ ﻭ ﺍﻳﻦ ﻧﻮﻉ ﺧﻄﺎ ﻫﻤﺎﻥ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۴۹‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪|s − s̃| = 4πr2 − 4π̃r̃3‬‬

‫ﻭ ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫|̃‪|s − s‬‬ ‫‪4πr2 − 4π̃r̃3‬‬
‫=‬
‫|‪|s‬‬ ‫| ‪|4πr2‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻣﺘﻔﺎﻭﺕ ﺍﺯ ﺧﻄﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﻣﺴﺎﺣﺖ ﻛﺮﻩ‪ ،‬ﺍﮔﺮ ﻣﺤﺎﺳﺒﺎﺕ ﺭﻭﻱ ﻳﻚ‬
‫ﻣﺎﺷﻴﻦ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﺿﺮﺏ ﻣﺎﺷﻴﻨﻲ ﻧﻴﺰ ﻋﻼﻭﻩ ﺑﺮ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺭﺥ ﻣﻲﺩﻫﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺿﺮﺏﻫﺎﻱ ﻣﻮﺟﻮﺩ‬
‫ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻣﺴﺎﺣﺖ ﻛﺮﻩ ﺑﻪ ﺻﻮﺭﺕ ﻭﺍﻗﻌﻲ ﻭ ﺑﺪﻭﻥ ﺧﻄﺎ ﺍﻧﺠﺎﻡ ﺷﻮﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﻧﺪﺍﺭﻳﻢ‪ ،‬ﺍﻣﺎ ﺑﺎﺯ ﻫﻢ ﺧﻄﺎﻱ‬
‫ﺍﻧﺘﺸﺎﺭ ﺭﺥ ﻣﻲﺩﻫﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺑﻪ ﺧﻄﺎ ﺁﻟﻮﺩﻩﺍﻧﺪ ﻭ ﺍﻳﻦ ﺁﻟﻮﺩﮔﻲ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺑﻪ ﻧﺘﻴﺠﻪ ﺳﺮﺍﻳﺖ ﻣﻲﻛﻨﺪ‪ ،‬ﻭ ﻫﺮ‬
‫ﭼﻨﺪ ﻛﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺑﻪ ﺻﻮﺭﺕ ﺩﻗﻴﻖ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺻﺮﻓﺎ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺭﺍ ﻣﻄﺎﻟﻌﻪ ﻣﻲﻛﻨﻴﻢ ﻭ ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺑﻪ ﺻﻮﺭﺕ ﻭﺍﻗﻌﻲ )ﻭ ﻧﻪ ﺭﻭﻱ ﻳﻚ‬
‫ﻣﺎﺷﻴﻦ( ﻭ ﺑﺪﻭﻥ ﺧﻄﺎ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﻧﺪ ﻭ ﻓﻘﻂ ﺩﺭ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺧﻄﺎ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﻫﺮ ﺩﻭ ﻧﻮﻉ ﺧﻄﺎ ﺭﺍ ﺧﻮﺍﻫﻴﻢ‬
‫ﺩﺍﺷﺖ ﻭ ﺑﺎﻳﺪ ﺗﺎﺛﻴﺮ ﻫﺮ ﺩﻭ ﺧﻄﺎ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﻢ‪ .‬ﻭﻟﻲ ﻓﻌﻼ ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺗﻨﻬﺎ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻣﻄﺎﻟﻌﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻫﺮﮔﺎﻩ ﻭﺭﻭﺩﻱﻫﺎﻱ ﻣﺴﺎﻟﻪ ﺑﺎ ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﺮﻭﺟﻲ ﻣﺴﺎﻟﻪ ﻧﻴﺰ ﺑﺎ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻫﻤﺮﺍﻩ ﺍﺳﺖ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻫﺮﮔﺎﻩ ﻧﺘﻮﺍﻧﻴﻢ ﻭﺭﻭﺩﻱﻫﺎﻱ ﺩﻗﻴﻖ ﺭﺍ ﻓﺮﺍﻫﻢ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺍﺟﺘﻨﺎﺏﻧﺎﭘﺬﻳﺮ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﻻﺯﻡ ﺍﺳﺖ‬
‫ﻛﻪ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺭﺍ ﺗﺤﻠﻴﻞ ﻭ ﺁﻧﺎﻟﻴﺰ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‪ ،‬ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺗﺨﻤﻴﻦ ﺯﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺩﺭ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺍﮔﺮ ﻣﺘﻐﻴﺮﻫﺎﻱ ﻭﺭﻭﺩﻱ ﻳﻚ ﻣﺴﺎﻟﻪ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻣﺸﺨﺼﻲ ﺧﻄﺎ‬
‫ﺁﻟﻮﺩﻩ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺣﺪﺍﻛﺜﺮ ﺧﻄﺎ ﺩﺭ ﺧﺮﻭﺟﻲ)ﻫﺎﻱ( ﻣﺴﺎﻟﻪ ﭼﻘﺪﺭ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬

‫ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻋﺒﺎﺭﺍﺕ ﻭ ﺗﻮﺍﺑﻊ‬ ‫�‬

‫ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺑﺮﺍﻱ ﻫﺮ ﻣﺴﺎﻟﻪﺍﻱ ﺳﺎﺩﻩ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﻣﺴﺎﻟﻪ ﻋﺒﺎﺭﺕ ﺑﺎﺷﺪ ﺍﺯ ﻣﺤﺎﺳﺒﻪ ﻣﻘﺪﺍﺭ ﻳﻚ ﺗﺎﺑﻊ )ﻳﺎ ﻋﺒﺎﺭﺕ(‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺳﺎﺩﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺩﺭ ﺍﻳﻦ ﻣﺴﺎﻳﻞ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺸﺨﺺ ﻛﻨﻴﻢ ﻛﻪ ﺧﻄﺎﻱ ﻣﺘﻐﻴﺮﻫﺎﻱ ﻭﺭﻭﺩﻱ‬
‫ﺗﺎﺑﻊ ﺗﺎ ﭼﻪ ﺍﻧﺪﺍﺯﻩ ﺑﻪ ﻣﺘﻐﻴﺮ ﺧﺮﻭﺟﻲ ﺗﺎﺑﻊ ﺳﺮﺍﻳﺖ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﻌﺎﺩﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺣﺴﺎﺳﻴﺖ ﺗﻮﺍﺑﻊ ﺭﺍ ﻧﺴﺒﺖ ﺑﻪ ﺗﻐﻴﻴﺮ ﺩﺭ‬
‫ﺁﺭﮔﻮﻣﺎﻥﻫﺎﻱ ﻭﺭﻭﺩﻱﺍﺵ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﺗﻮﺍﺑﻊ ﻳﻚﻣﺘﻐﻴﺮﻩ ﺭﺍ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‬
‫ﺩﺭ ﺗﻮﺍﺑﻊ ﭼﻨﺪﻣﺘﻐﻴﺮﻩ ﺣﻘﻴﻘﻲ ﺭﺍ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫■ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﻮﺑﻊ ﻳﻚﻣﺘﻐﻴﺮﻩ ﺣﻘﻴﻘﻲ‬

‫ﺗﺎﺑﻊ ﻳﻚ ﻣﺘﻐﻴﺮﻩ ﺣﻘﻴﻘﻲ ‪ f : R → R‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻳﻦ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻧﻘﻄﻪ ‪ x = a‬ﺑﻪ‬
‫ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﺍﻣﺎ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ‪ a‬ﺩﺭ ﺩﺳﺖ ﻧﻴﺴﺖ‪ ،‬ﺑﻠﻜﻪ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ a‬ﻣﺎﻧﻨﺪ ̃‪ a‬ﺩﺭ ﺩﺳﺖ ﺍﺳﺖ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺮﺭﺳﻲ ﻛﻨﻴﻢ ﻛﻪ ﺩﺭ‬
‫ﻣﺤﺎﺳﺒﻪ ‪ f‬ﺍﮔﺮ ﺑﻪ ﺟﺎﻱ ‪ a‬ﺍﺯ ̃‪ a‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﭼﻘﺪﺭ ﺧﻄﺎ ﻣﺮﺗﻜﺐ ﻣﻲﺷﻮﻳﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‬
‫ﺩﺭ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻘﺮﻳﺐ ̃‪ a‬ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪.‬‬
‫ﺑﻪ ﺯﺑﺎﻥ ﺭﻳﺎﺿﻲ‪ ،‬ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫|)̃‪|f (a) − f (a‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۵۰‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ﺍﮔﺮ ﻣﺸﺘﻖ ﺍﻭﻝ ‪ f‬ﺩﺭ ﺑﻴﻦ ‪ a‬ﻭ ̃‪ a‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﺗﺨﻤﻴﻦ ﺧﻄﺎﻱ ﻓﻮﻕ‪ ،‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﻲ ﺩﺍﺭﻳﻢ‪:‬‬
‫)̃‪f (a) − f (ã) = f ′ (η)(a − a‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺩﺭ ﺑﻴﻦ ‪ a‬ﻭ ̃‪ a‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ η‬ﻧﺎﻣﻌﻠﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻟﺬﺍ ‪ η‬ﺭﺍ ﺑﺎ ‪ a‬ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﻴﻢ ﻭ ﺧﻮﺍﻫﻴﻢ‬
‫ﺩﺍﺷﺖ‪:‬‬
‫)̃‪f (a) − f (ã) ≃ f ′ (a)(a − a‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻫﺮﭼﻪ ̃‪ a‬ﺑﻪ ‪ a‬ﻧﺰﺩﻳﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﺩﻗﻴﻖﺗﺮ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺑﺎ ﮔﺮﻓﺘﻦ ﻗﺪﺭﻣﻄﻠﻖ ﺍﺯ ﻃﺮﻓﻴﻦ ﺭﺍﺑﻄﻪ ﻓﻮﻕ‪،‬‬
‫ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫|̃‪|f (a) − f (ã)| ≃ |f ′ (a)| |a − a‬‬
‫)̃‪= |f ′ (a)| EA (a‬‬
‫ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪′‬‬
‫|)̃‪|f (a) − f (a‬‬ ‫|)‪|f (a‬‬
‫≃‬ ‫)̃‪EA (a‬‬
‫|)‪|f (a‬‬ ‫)‪f (a‬‬
‫|)‪|f ′ (a‬‬ ‫)̃‪EA (a‬‬
‫=‬ ‫|‪|a‬‬
‫)‪f (a‬‬ ‫|‪|a‬‬
‫‪′‬‬
‫)‪f (a‬‬
‫‪= a‬‬ ‫)̃‪ER (a‬‬
‫)‪f (a‬‬
‫ﻣﻄﺎﻟﺐ ﻓﻮﻕ ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ ﺧﻼﺻﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻣﺘﻐﯿﺮﻩ‬ ‫ﺧﻄﺎﯼ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ﯾ‬ ‫ﻗﻀﯿﻪ ‪۹−۲‬‬

‫ﺍﮔﺮ ‪ f : R → R‬ﺩﺭ ﺍﻃﺮﺍﻑ ‪ a‬ﺑﻪ ﻃﻮﺭ ﭘﻴﻮﺳﺘﻪ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺑﺎﺷﺪ ﻭ ̃‪ a‬ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ a‬ﺑﺎ ﺧﻄﺎﻱ ﻣﻄﻠﻖ )̃‪ EA (a‬ﻭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‬
‫)̃‪ ER (a‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﺨﻤﻴﻨﻲ ﺍﺯ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ )‪ f (a‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫)̃‪|f (a) − f (ã)| ≃ |f ′ (a)| EA (a‬‬
‫|)̃‪|f (a) − f (a‬‬ ‫)‪f ′ (a‬‬
‫‪≃ a‬‬ ‫)̃‪ER (a‬‬
‫|)‪|f (a‬‬ ‫)‪f (a‬‬

‫ﻃﺒﻖ ﻗﻀﻴﻪ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ ،x = a‬ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ‪:‬‬
‫• ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‪ ،‬ﺍﮔﺮ ﺧﻄﺎﻱ ﺩﺭ ﺩﺍﺩﻩﻱ ﻭﺭﻭﺩﻱ ﺑﺮﺍﺑﺮ )̃‪ EA (a‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺧﻄﺎ ﺗﻘﺮﻳﺒﺎ ﺩﺭ |)‪ |f ′ (a‬ﺿﺮﺏ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﻣﻘﺪﺍﺭ ﺧﺮﻭﺟﻲ ﺗﺎﺑﻊ ﺍﻧﺘﺸﺎﺭ ﻣﻲﻳﺎﺑﺪ‪.‬‬
‫‪′‬‬
‫)‪ a ff (a‬ﺿﺮﺏ‬
‫)‪(a‬‬
‫• ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ‪ ،‬ﺍﮔﺮ ﺧﻄﺎ ﺩﺭ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺑﺮﺍﺑﺮ )̃‪ ER (a‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺧﻄﺎ ﺗﻘﺮﻳﺒﺎ ﺩﺭ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﻣﻘﺪﺍﺭ ﺧﺮﻭﺟﻲ ﺗﺎﺑﻊ ﺍﻧﺘﺸﺎﺭ ﻣﻲﻳﺎﺑﺪ‪.‬‬
‫‪′‬‬
‫)‪ a ff (a‬ﻣﺸﺨﺺ ﻛﻨﻨﺪﻩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺍﺩﻩ ﻭﺭﻭﺩﻱ‬ ‫)‪(a‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ ،x = a‬ﻓﺎﻛﺘﻮﺭ‬
‫‪۱۹‬‬
‫ﺗﻘﻮﻳﺖ ﻭ ﻳﺎ ﺗﻀﻌﻴﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻫﻤﻴﻦ ﺩﻟﻴﻞ‪ ،‬ﺍﻳﻦ ﻓﺎﻛﺘﻮﺭ ﺑﻪ ﻋﺪﺩ ﺣﺎﻟﺖ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ a‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﻫﺮﭼﻪ‬
‫ﻋﺪﺩ ﺣﺎﻟﺖ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ‪ 1‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﺑﺎ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺍﺩﻩﻱ ﻭﺭﻭﺩﻱ ﺑﺎﻻﺗﺮ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻭﻟﻲ ﺍﮔﺮ‬
‫ﻋﺪﺩ ﺣﺎﻟﺖ ﻛﻮﭼﻚﺗﺮ ﺍﺯ ﻳﻚ ﻭ ﻧﺰﺩﻳﻚ ﺑﻪ ﺻﻔﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﺑﺎ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺍﺩﻩﻱ ﻭﺭﻭﺩﻱ‬
‫ﻛﻮﭼﻚ ﺍﺳﺖ‪.‬‬
‫‪19 Condition‬‬ ‫‪number‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵۱‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﻧﻜﺘﻪ‪ :‬ﺩﺭ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ‪ ،‬ﻣﺘﻐﻴﺮ ‪ a‬ﻇﺎﻫﺮ ﺷﺪﻩ ﺍﺳﺖ ﻛﻪ ﻏﺎﻟﺒﺎ ﺩﺭ ﺩﺳﺘﺮﺱ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ‪،‬‬
‫ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ‪ ،‬ﺍﺯ ﺗﻘﺮﻳﺐ ̃‪ a‬ﺑﻪ ﺟﺎﻱ ‪ a‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻟﺒﺘﻪ‪ ،‬ﺍﻳﻦ ﻛﺎﺭ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﻗﺖ ﺗﺨﻤﻴﻦ ﺧﻄﺎ ﺍﻧﺪﻛﻲ ﻛﺎﻫﺶ‬
‫ﻳﺎﺑﺪ‪.‬‬

‫ﻣﺘﻐﯿﺮﻩ(‬ ‫ﺗﺎﺑﻊ ﯾ‬ ‫ﻣﺜﺎﻝ ‪)۸−۲‬ﺗﺨﻤﯿﻦ ﺧﻄﺎﯼ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ a = 2.49‬ﺭﺍ ﺑﺎ ‪ ã = 2.5‬ﺗﻘﺮﻳﺐ ﺯﺩﻩ ﺑﺎﺷﻴﻢ ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ‪ .EA (ã) = 0.01‬ﻫﻤﻴﻦﻃﻮﺭ ﻓﺮﺽ ﻛﻨﻴﺪ‬
‫‪ .f (x) = x3‬ﺍﮔﺮ )‪ f (a‬ﺭﺍ ﺑﺎ )̃‪ f (a‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺨﻤﻴﻦ ﺯﺩﻩ ﻣﻲﺷﻮﺩ‬
‫‪|f (a) − f (ã)| ≃ |f ′ (ã)| EA (ã) = 3(2.5)2 (0.01) = 0.1875.‬‬
‫ﺍﻟﺒﺘﻪ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪:‬‬
‫‪|f (a) − f (ã)| = (2.49)3 − (2.5)3 = 0.186751‬‬
‫ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﺑﺎ ﻛﻤﻚ ﻓﺮﻣﻮﻝ‪ ،‬ﺗﻮﺍﻧﺴﺘﻪﺍﻳﻢ ﺗﺨﻤﻴﻦ ﺧﻮﺑﻲ ﺍﺯ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪.‬‬

‫■ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﻮﺑﻊ ﭼﻨﺪﻣﺘﻐﻴﺮﻩ ﺣﻘﻴﻘﻲ‬

‫ﺗﺎﺑﻊ ‪ n‬ﻣﺘﻐﻴﺮﻩ ‪ f : Rn → R‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ) ‪ f (a1 , . . . , an‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ‬
‫ﻛﻪ ‪ ã1 , · · · , ãn‬ﺗﻘﺮﻳﺐﻫﺎﻳﻲ ﺑﺮﺍﻱ ‪ a1 , · · · , an‬ﺑﺎﺷﺪ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺮﺭﺳﻲ ﻛﻨﻴﻢ ﻛﻪ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ‪ ،f‬ﺍﮔﺮ ﺑﻪ ﺟﺎﻱ ‪ai‬ﻫﺎ ﺍﺯ‬
‫ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ ãi‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺑﻪ ﭼﻪ ﺍﻧﺪﺍﺯﻩ ﺗﻐﻴﻴﺮ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺸﺨﺺ ﻛﻨﻴﻢ ﻛﻪ‬
‫ﺍﮔﺮ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ) ‪ f (a1 , · · · , an‬ﺑﻪ ﺟﺎﻱ ‪xi‬ﻫﺎ ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ ãi‬ﺭﺍ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ‪ ãi‬ﺗﺎ ﭼﻪ ﺍﻧﺪﺍﺯﻩ ﺑﻪ‬
‫ﻣﻘﺪﺍﺭ ﺧﺮﻭﺟﻲ ﺗﺎﺑﻊ ﺍﻧﺘﺸﺎﺭ ﻣﻲﻳﺎﺑﺪ‪ .‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫|) ‪|f (a1 , · · · , an ) − f (ã1 , · · · , ãn‬‬
‫ﺍﮔﺮ ‪ f‬ﺩﺍﺭﺍﻱ ﻣﺸﺘﻘﺎﺕ ﻧﺴﺒﻲ ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫) ‪∂f (η1 , . . . , ηn‬‬ ‫) ‪∂f (η1 , . . . , ηn‬‬
‫= ) ‪f (a1 , · · · , an ) − f (ã1 , · · · , ãn‬‬ ‫‪(a1 − ã1 ) + · · · +‬‬ ‫) ‪(an − ãn‬‬
‫‪∂x1‬‬ ‫‪∂xn‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ) ‪ (η1 , . . . , ηn‬ﻧﻘﻄﻪﺍﻱ ﺭﻭﻱ ﭘﺎﺭﻩﺧﻂ ﻭﺍﺻﻞ ) ‪ (a1 , . . . , an‬ﻭ ) ‪ (ã1 , . . . , ãn‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺣﺎﻝ‪ ηi ،‬ﺭﺍ ﺑﺎ ‪ai‬‬
‫ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﻴﻢ ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫) ‪∂f (a1 , . . . , an‬‬ ‫) ‪∂f (a1 , . . . , an‬‬
‫≃ ) ‪f (a1 , · · · , an ) − f (ã1 , · · · , ãn‬‬ ‫‪(a1 − ã1 ) + · · · +‬‬ ‫) ‪(an − ãn‬‬
‫‪∂x1‬‬ ‫‪∂xn‬‬
‫ﻫﺮﭼﻪ ‪ãi‬ﻫﺎ ﺗﻘﺮﻳﺐﻫﺎﻱ ﺩﻗﻴﻖﺗﺮﻱ ﺑﺮﺍﻱ ‪ai‬ﻫﺎ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﻧﻴﺰ ﺩﻗﻴﻖﺗﺮ ﺍﺳﺖ‪ .‬ﺣﺎﻝ‪ ،‬ﺑﺎ ﻗﺪﺭﻣﻄﻠﻖ ﮔﺮﻓﺘﻦ ﺍﺯ ﻃﺮﻓﻴﻦ‬
‫ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﻭ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻧﺎﻣﺴﺎﻭﻱ ﻣﺜﻠﺚ ﺩﺍﺭﻳﻢ‪:‬‬
‫) ‪∂f (a1 , . . . , an‬‬ ‫) ‪∂f (a1 , . . . , an‬‬
‫≲ |) ‪|f (a1 , · · · , an ) − f (ã1 , · · · , ãn‬‬ ‫‪EA (ã1 ) + · · · +‬‬ ‫) ‪EA (ãn‬‬
‫‪∂x1‬‬ ‫‪∂xn‬‬
‫ﺑﺎ ﺗﻘﺴﻴﻢ ﻃﺮﻓﻴﻦ ﻧﺎﻣﺴﺎﻭﻱ ﻓﻮﻕ ﺑﺮ |) ‪ |f (a1 , · · · , an‬ﻭ ﻛﻤﻲ ﺩﺳﺘﻜﺎﺭﻱ ﺟﺒﺮﻱ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ‬
‫ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪:‬‬
‫) ‪∂f (a1 ,...,an‬‬ ‫) ‪∂f (a1 ,...,an‬‬
‫|) ‪|f (a1 , · · · , an ) − f (ã1 , · · · , ãn‬‬ ‫‪∂x1‬‬ ‫‪∂xn‬‬
‫≲‬ ‫‪EA (ã1 ) + · · · +‬‬ ‫) ‪EA (ãn‬‬
‫|) ‪|f (a1 , · · · , an‬‬ ‫|) ‪|f (a1 , · · · , an‬‬ ‫|) ‪|f (a1 , · · · , an‬‬
‫) ‪∂f (a1 ,...,an‬‬ ‫) ‪∂f (a1 ,...,an‬‬
‫‪∂x1‬‬ ‫‪∂xn‬‬
‫‪= a1‬‬ ‫‪ER (ã1 ) + · · · + an‬‬ ‫) ‪ER (ãn‬‬
‫) ‪f (a1 , · · · , an‬‬ ‫) ‪f (a1 , · · · , an‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۵۲‬‬

‫ﻣﻮﺍﺭﺩ ﻓﻮﻕ ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ ﺟﻤﻊﺑﻨﺪﻱ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺧﻄﺎﯼ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ﭼﻨﺪﻣﺘﻐﯿﺮﻩ‬ ‫ﻗﻀﯿﻪ ‪۱۰−۲‬‬

‫ﺍﮔﺮ ‪ f : Rn → R‬ﺑﻪ ﻃﻮﺭ ﭘﻴﻮﺳﺘﻪ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺑﺎﺷﺪ ﻭ ‪ ã1 , . . . , ãn‬ﺗﻘﺮﻳﺐﻫﺎﻳﻲ ﺍﺯ ‪ a1 , . . . , an‬ﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺗﻘﺮﻳﺐ ‪ ai‬ﺑﺎ ‪ ãi‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﺮﺍﺑﺮ ) ‪ EA (ãi‬ﻭ ) ‪ ER (ãi‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﺨﻤﻴﻨﻲ ﺍﺯ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‬
‫ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪X‬‬‫‪n‬‬
‫) ‪∂f (a1 , . . . , an‬‬
‫≲ |) ‪|f (a1 , . . . , an ) − f (ã1 , . . . , ãn‬‬ ‫) ‪EA (ãi‬‬
‫‪i=1‬‬
‫‪∂xi‬‬
‫‪|f (a1 , . . . , an ) − f (ã1 , . . . , ãn )| X‬‬
‫‪n‬‬
‫‪ai‬‬ ‫) ‪∂f (a1 , . . . , an‬‬
‫≲‬ ‫) ‪ER (ãi‬‬
‫|) ‪|f (a1 , . . . , an‬‬ ‫‪i=1‬‬
‫‪f‬‬ ‫‪(a‬‬ ‫‪1‬‬ ‫‪,‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪,‬‬ ‫‪a‬‬ ‫‪n‬‬ ‫)‬ ‫‪∂xi‬‬

‫ﺩﺭ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﻀﻴﻪ ﻓﻮﻕ‪ ،‬ﻣﻌﻤﻮﻻ‪ ،‬ﻣﻘﺎﺩﻳﺮ ‪ai‬ﻫﺎ ﺩﺭ ﺩﺳﺘﺮﺱ ﻧﻤﻲﺑﺎﺷﺪ ﻭﻟﻲ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ ãi‬ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﺑﻪ‬
‫ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺩﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﻧﺎﻣﺴﺎﻭﻱﻫﺎﻱ ﻓﻮﻕ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺟﺎﻱ ‪ ai‬ﺍﺯ ‪ ãi‬ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ ﭼﻨﺪﻣﺘﻐﯿﺮﻩ(‬ ‫ﻣﺜﺎﻝ ‪)۹−۲‬ﺗﺨﻤﯿﻦ ﺧﻄﺎﯼ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭ ﻋﺒﺎﺭﺕ ﺯﻳﺮ ﺭﺍ ﺣﺴﺎﺏ ﻛﻨﻴﻢ‬
‫‪f l4‬‬
‫=‪y(f, l, e, i) :‬‬
‫‪8ei‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﻛﻤﻴﺖﻫﺎﻱ ‪ e ،l ،f‬ﻭ ‪ i‬ﺩﺭ ﺩﺳﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺍﻣﺎ ﺗﻘﺮﻳﺐ ﺁﻥﻫﺎ ﺑﻪ ﻋﻼﻭﻩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ‬
‫ﺁﻥﻫﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﺩﺳﺖ ﺍﺳﺖ‪:‬‬
‫‪f˜ = 750‬‬ ‫‪EA (f˜) ≤ 30‬‬
‫‪˜l = 9‬‬ ‫‪EA (˜l) ≤ 0.03‬‬
‫‪ẽ = 7.5 × 109‬‬ ‫‪EA (ẽ) ≤ 5 × 107‬‬
‫‪ĩ = 0.0005‬‬ ‫‪EA (ĩ) ≤ 0.000005‬‬
‫ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﻘﺪﺍﺭﻫﺎﻱ ﺗﻘﺮﻳﺒﻲ ﻓﻮﻕ‪ ،‬ﻣﻘﺪﺍﺭ ‪ y‬ﺑﺮﺍﺑﺮ ‪ y = 0.164025‬ﺗﻘﺮﻳﺐ ﺯﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺭﺍ‬
‫ﺗﺨﻤﻴﻦ ﺑﺰﻧﻴﻢ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫)̃‪∂y(f˜, ˜l, ẽ, i‬‬ ‫)̃‪∂y(f˜, ˜l, ẽ, i‬‬
‫≲ )̃‪y(f, l, e, i) − y(f˜, ˜l, ẽ, i‬‬ ‫‪EA (f˜) +‬‬ ‫‪EA (˜l)+‬‬
‫‪∂f‬‬ ‫‪∂l‬‬

‫)̃‪∂y(f˜, ˜l, ẽ, i‬‬ ‫)̃‪∂y(f˜, ˜l, ẽ, i‬‬

‫‪EA (ẽ) +‬‬ ‫)̃‪EA (i‬‬
‫‪∂e‬‬ ‫‪∂i‬‬
‫ﻭ ﻟﺬﺍ‬
‫‪˜l4‬‬ ‫‪f˜ĩ3‬‬ ‫‪f˜ĩ4‬‬ ‫‪f˜ĩ4‬‬
‫≲ )̃‪y(f, l, e, i) − y(f˜, ˜l, ẽ, i‬‬ ‫‪EA (f ) +‬‬ ‫‪EA (l) + 2 EA (e) +‬‬ ‫)‪EA (i‬‬
‫̃‪8ẽi‬‬ ‫̃‪2ẽi‬‬ ‫̃‪8ẽ i‬‬ ‫‪8ẽĩ2‬‬
‫ﻭ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪y(f, l, e, i) − y(f˜, ˜l, ẽ, ĩ) ≲ 0.006561 + 0.002187 + 0.001094 + 0.00164 = 0.011482‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵۳‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ)ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ(‬ ‫�‬

‫ﺩﺭ ﺍﻳﻦ ﺯﻳﺮﺑﺨﺶ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺭﺍ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﻟﺒﺘﻪ ﻧﻜﺘﻪ ﺟﺪﻳﺪﻱ ﺩﺭ ﺍﻳﻦ ﺯﻳﺮﺑﺨﺶ ﺍﺭﺍﻳﻪ ﻧﻤﻲﺷﻮﺩ ﻭ‬
‫ﺗﻨﻬﺎ ﻣﻄﺎﻟﺐ ﻗﺒﻞ ﺭﺍ ﺑﺮﺍﻱ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺗﻜﺮﺍﺭ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺍﺑﺘﺪﺍ‪ ،‬ﺗﻮﺿﻴﺤﺎﺕ ﺭﺍ ﻓﻘﻂ ﺑﺮﺍﻱ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﻋﻤﻞ ﺟﻤﻊ ﺍﺭﺍﻳﻪ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺍﻋﻤﺎﻝ ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ ﻗﺎﺑﻞ ﺍﺳﺘﺨﺮﺍﺝ ﺍﺳﺖ‪.‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ x‬ﻭ ‪ y‬ﺩﻭ ﻋﺪﺩ ﺑﺎﺷﻨﺪ ﻭ ̃‪ x‬ﻭ ̃‪ y‬ﺗﻘﺮﻳﺐﻫﺎﻳﻲ ﺍﺯ ﺁﻥ ﺩﻭ ﻋﺪﺩ ﺑﺎﺷﻨﺪ‪ .‬ﻣﻲﺗﻮﺍﻥ ̃‪ z̃ := x̃ + y‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫ﺗﻘﺮﻳﺒﻲ ﺑﺮﺍﻱ ‪ z := x + y‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ̃‪ x̃ + y‬ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖ ﺍﻧﺠﺎﻡ ﺷﻮﺩ ﻭ ﺧﻄﺎﻳﻲ ﺍﺯ ﺑﺎﺑﺖ ﺟﻤﻊ‬
‫ﺩﻭ ﻋﺪﺩ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ .‬ﺍﻣﺎ ﺧﻄﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ﻫﺮ ﻛﺪﺍﻡ ﺍﺯ ﺍﻋﺪﺍﺩ ̃‪ x‬ﻭ ̃‪ y‬ﺑﻪ ﺣﺎﺻﻞﺟﻤﻊ ̃‪ x̃ + y‬ﺳﺮﺍﻳﺖ ﻣﻲﻛﻨﺪ ﻭ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ‬
‫ﻛﻪ ̃‪ x̃ + y‬ﻣﺘﻔﺎﻭﺕ ﺍﺯ ‪ x + y‬ﺑﺎﺷﺪ‪ .‬ﺗﻔﺎﻭﺕ ‪ x + y‬ﺑﺎ ̃‪ x̃ + y‬ﺑﻪ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻋﻤﻞ ﺟﻤﻊ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺣﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﻪ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺟﻤﻊ ﺑﭙﺮﺩﺍﺯﻳﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺑﻪ ﺍﻳﻦ ﻧﻜﺘﻪ ﺗﻮﺟﻪ ﻣﻲﻛﻨﻴﻢ ﻛﻪ‪ :‬ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‬
‫ﺩﺭ ﺟﻤﻊ ‪ x‬ﻭ ‪ y‬ﻣﻌﺎﺩﻝ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﺗﺎﺑﻊ ﺩﻭﻣﺘﻐﻴﺮﻩ ‪ f (x, y) := x + y‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻛﻪ ﻗﺒﻼ ﺑﺮﺭﺳﻲ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪،‬‬
‫ﺑﺮﺍﻱ ﺗﻌﻴﻴﻦ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻋﻤﻞ ﺟﻤﻊ‪ ،‬ﻛﺎﻓﻲ ﺍﺳﺖ ﺩﺭ ﻗﻀﻴﻪ ‪ ۱۰-۲‬ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪ f (x, y) := x + y‬ﺗﻌﺮﻳﻒ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫|)̃‪|f (x, y) − f (x̃, ỹ)| = |(x − y) − (x̃ − y‬‬
‫)‪∂f (x, y‬‬ ‫)‪∂f (x, y‬‬
‫≲‬ ‫‪EA (x̃) +‬‬ ‫)̃‪EA (y‬‬
‫‪∂x‬‬ ‫‪∂y‬‬
‫)̃‪= EA (x̃) + EA (y‬‬
‫ﻣﻄﺎﺑﻖ ﺑﺎ ﻧﺎﻣﺴﺎﻭﻱ ﻓﻮﻕ‪ ،‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﺟﻤﻊ ﻛﻮﭼﻚﺗﺮ ﺍﺯ ﻣﺠﻤﻮﻉ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺭ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‬
‫ﺩﺍﺭﻳﻢ‪:‬‬
‫|)̃‪|f (x, y) − f (x̃, y‬‬ ‫|)̃‪|(x − y) − (x̃ − y‬‬
‫=‬
‫|)‪|f (x, y‬‬ ‫|)‪|f (x, y‬‬
‫)‪x ∂f (x, y‬‬ ‫)‪x ∂f (x, y‬‬
‫≲‬ ‫‪ER (x̃) +‬‬ ‫)̃‪ER (y‬‬
‫‪f (x, y) ∂x‬‬ ‫‪f (x, y) ∂y‬‬
‫|‪|x‬‬ ‫|‪|y‬‬
‫=‬ ‫‪ER (x̃) +‬‬ ‫)̃‪ER (y‬‬
‫|‪|x + y‬‬ ‫|‪|x + y‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ‪ ،‬ﻏﺎﻟﺒﺎ‪ ،‬ﻣﻘﺎﺩﻳﺮ ‪ x‬ﻭ ‪ y‬ﺩﺭ ﺩﺳﺘﺮﺱ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﻟﺬﺍ ﺩﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﻧﺎﻣﺴﺎﻭﻱ ﻓﻮﻕ‪ ،‬ﺍﺯ ﺗﻘﺮﻳﺐﻫﺎﻱ ̃‪ x‬ﻭ ̃‪ y‬ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﻛﻨﻴﻢ ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫|)̃‪|f (x, y) − f (x̃, y‬‬ ‫|̃‪|x‬‬ ‫|̃‪|y‬‬
‫≲‬ ‫‪ER (x̃) +‬‬ ‫)̃‪ER (y‬‬
‫|)‪|f (x, y‬‬ ‫|̃‪|x̃ + y‬‬ ‫|̃‪|x̃ + y‬‬
‫ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ﺍﻋﺪﺍﺩ ‪ x‬ﻭ ‪ y‬ﻫﻢﻋﻼﻣﺖ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪ER (z̃) ≤ max {ER (x̃), ER (ỹ)} .‬‬ ‫)‪(۱۷.۲‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ‪ x‬ﻭ ‪ y‬ﻫﻢﻋﻼﻣﺖ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻧﺎﺷﻲ ﺍﺯ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺩﺭ ﺟﻤﻊ‪ ،‬ﺍﺯ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﻭ ﻋﺪﺩ‬
‫ﻛﻮﭼﻚﺗﺮ ﺍﺳﺖ‪.‬‬
‫ﺗﺎ ﺍﻳﻨﺠﺎ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻋﻤﻞ ﺟﻤﻊ ﺍﺳﺘﺨﺮﺍﺝ ﮔﺮﺩﻳﺪ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ ﻣﻲﺗﻮﺍﻥ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ‬
‫ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻧﻴﺰ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ‪ ،‬ﻛﺮﺍﻥﻫﺎﻱ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﻭ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺑﺮﺍﻱ ﭼﻬﺎﺭ‬
‫ﻋﻤﻞ ﺍﺻﻠﻲ ﺧﻼﺻﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۵۴‬‬

‫ﻋﻤﻞ‬ ‫ﺩﺍﺩﻩﻫﺎ‬ ‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺍﻧﺘﺸﺎﺭ‬ ‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ‬

‫‪z := x + y‬‬ ‫|̃‪|x‬‬ ‫|̃‪|y‬‬
‫ﺟﻤﻊ‬ ‫)̃‪EA (z̃) ≲ EA (x̃) + EA (y‬‬ ‫≲ )̃‪ER (z‬‬ ‫)̃‪|x̃+ỹ| ER (x‬‬ ‫‪+‬‬ ‫)̃‪|x̃+ỹ| ER (y‬‬
‫̃‪z̃ := x̃ + y‬‬
‫)‪ER (z̃) ≲ max {ER (x̃), ER (ỹ)} , (xy ≥ 0‬‬
‫‪z := x − y‬‬ ‫|̃‪|x‬‬ ‫|̃‪|y‬‬
‫ﺗﻔﺮﻳﻖ‬ ‫)̃‪EA (z̃) ≲ EA (x̃) + EA (y‬‬ ‫≲ )̃‪ER (z‬‬ ‫)̃‪|x̃−ỹ| ER (x‬‬ ‫‪+‬‬ ‫)̃‪|x̃−ỹ| ER (y‬‬
‫̃‪z̃ := x̃ − y‬‬
‫‪z := xy‬‬
‫ﺿﺮﺏ‬ ‫)̃‪EA (z̃) ≲ |ỹ| EA (x̃) + |x̃| EA (y‬‬ ‫)̃‪ER (z̃) ≲ ER (x̃) + ER (y‬‬
‫̃‪z̃ := x̃y‬‬
‫‪z := x/y,‬‬ ‫|̃‪|x‬‬
‫ﺗﻘﺴﻴﻢ‬ ‫≲ )̃‪EA (z‬‬ ‫)̃‪ỹ 2 EA (y‬‬ ‫‪+‬‬ ‫‪1‬‬
‫)̃‪|ỹ| EA (x‬‬ ‫)̃‪ER (z̃) ≲ ER (x̃) + ER (y‬‬
‫̃‪z̃ := x̃/y‬‬

‫ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ‬ ‫�‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺟﺪﻭﻝ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ‪ ،‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺭ ﻋﻤﻞ ﺟﻤﻊ ﺩﻭ ﻋﺪﺩ ﻫﻢﻋﻼﻣﺖ‪ ،‬ﺿﺮﺏ ﻭ‬
‫ﺗﻘﺴﻴﻢ ﺗﻘﻮﻳﺖ ﻧﻤﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﺍﻧﺘﺸﺎﺭ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎ ﺩﺭ ﺍﻳﻦ ﻋﻤﻠﻴﺎﺕ ﻧﮕﺮﺍﻥ ﻛﻨﻨﺪﻩ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻋﻤﻠﻴﺎﺕ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ‬
‫ﺗﻘﺮﻳﺒﺎ ﺍﺯ ﻣﺮﺗﺒﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﻣﻮﺭﺩ ﻋﻤﻞ ﺗﻔﺮﻳﻖ‪ ،‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺩﻭ ﻋﺪﺩ ‪ x‬ﻭ ‪ y‬ﺑﻪ ﻫﻢ ﻧﺰﺩﻳﻚ‬
‫‪ x−y‬ﺑﺰﺭﮒ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﺍﻳﻦ‬‫‪y‬‬
‫‪ x−y‬ﻭ‬
‫‪x‬‬
‫ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ x − y‬ﻣﻘﺪﺍﺭﻱ ﻛﻮﭼﻚ ﻣﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﻓﺎﻛﺘﻮﺭﻫﺎﻱ‬
‫ﻓﺎﻛﺘﻮﺭﻫﺎ ﺩﺭ )̃‪ ER (x‬ﻭ )̃‪ ER (y‬ﺿﺮﺏ ﻣﻲﺷﻮﻧﺪ‪ ،‬ﻟﺬﺍ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺑﺰﺭﮒ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻧﻜﺘﻪ ﺑﻪ ﺍﻳﻦ‬
‫ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﺧﻄﺎﻱ ﺍﻧﺘﺸﺎﺭ ﺩﺍﺩﻩﻫﺎ ﺩﺭ ﺗﻔﺮﻳﻖ ﺩﻭ ﻋﺪﺩ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﻢ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺴﻴﺎﺭ ﺑﺰﺭﮒ ﻭ ﻣﺨﺮﺏ ﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ‬
‫ﭼﻨﻴﻦ ﻣﺸﻜﻠﻲ ﺍﺯ ﻭﺍﮊﻩ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ‪ ۲۰‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻫﺮﮔﺎﻩ ﺩﻭ ﻋﺪﺩ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﻢ ﺭﺍ ﺗﻔﺮﻳﻖ ﻣﻲﻛﻨﻴﻢ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺑﺰﺭﮔﻲ ﻣﻮﺳﻮﻡ ﺑﻪ ﻛﻨﺴﻠﻴﺸﻦ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪.‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ‪،‬ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻫﻨﮕﺎﻡ ﺟﻤﻊ ﺩﺭ ﻋﺪﺩ ﻛﻪ ﻋﻼﻣﺖ ﻣﺘﻔﺎﻭﺕ ﺩﺍﺭﻧﺪ ﻭﻟﻲ ﻗﺪﺭﻣﻄﻠﻖ ﺁﻥﻫﺎ ﺑﻪ ﻳﻜﺪﻳﮕﺮ ﻧﺰﺩﻳﻚ‬
‫ﺍﺳﺖ ﻧﻴﺰ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۰−۲‬ﻣﺜﺎﻟ ﺍﺯ ﺧﻄﺎﯼ ﮐﻨﺴﻠﯿﺸﻦ(‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x̃ = 1.000001‬ﻭ ‪ ỹ = 1.000009‬ﺗﻘﺮﻳﺐﻫﺎﻳﻲ ﺑﺮﺍﻱ ‪ x = 1.000000‬ﻭ ‪ y = 1.000003‬ﺑﺎﺷﻨﺪ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫|‪|1.000000 − 1.000001‬‬
‫= )̃‪ER (x‬‬ ‫‪= 1 e − 6,‬‬
‫‪1.000000‬‬
‫|‪|1.000003 − 1.000009‬‬
‫= )̃‪ER (y‬‬ ‫‪≃ 6e − 6‬‬
‫‪1.000003‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ﺗﻘﺮﻳﺐﻫﺎﻱ ﻓﻮﻕ ﺗﻘﺮﻳﺒﺎ ﺑﺮﺍﺑﺮ ‪ 1 e − 6‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﺻﻄﻼﺡ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺍﺩﻩﻫﺎﻱ‬
‫ﻭﺭﻭﺩﻱ ﺍﺯ ﻣﺮﺗﺒﻪ ‪ 1 e − 6‬ﺍﺳﺖ‪.‬‬
‫ﺣﺎﻝ ﻣﻲﺗﻮﺍﻥ ̃‪ x̃ − y‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ x − y‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪x − y = −0.000003,‬‬
‫‪x̃ − ỹ = −0.000008‬‬

‫‪20 Cancellation‬‬ ‫‪error‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵۵‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﻭ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ‪:‬‬

‫|)̃‪|(x − y) − (x̃ − y‬‬ ‫‪0.000005‬‬
‫= )̃‪ER (x̃ − y‬‬ ‫=‬ ‫‪≃ 1.66‬‬
‫|‪|x − y‬‬ ‫‪0.000003‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻫﺮﭼﻨﺪ ﻛﻪ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺩﺭ ﺣﺪﻭﺩ ‪ 1e − 6‬ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺍﻧﺘﺸﺎﺭ ﺗﻘﺮﻳﺒﺎ ﺑﺮﺍﺑﺮ ‪1.66‬‬
‫ﺍﺳﺖ ﻛﻪ ﺣﺪﻭﺩﺍ ‪ 1000000‬ﺑﺮﺍﺑﺮ ﺧﻄﺎﻱ ﺩﺍﺩﻩﻫﺎﻱ ﻭﺭﻭﺩﻱ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺧﻄﺎﻱ ﭼﺸﻢﮔﻴﺮ ﺭﺍ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫ﺗﺮﻓﻨﺪﻫﺎ ﻭ ﺗﻮﺻﻴﻪﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ‬ ‫‪۵.۲‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﺍﺳﺖ ﻭ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﻭﺍﻗﻌﻲ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﮔﺎﻫﻲ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﺎ ﺗﻐﻴﻴﺮ ﻧﺤﻮﻩ ﺍﻧﺤﺎﻡ ﻣﺤﺎﺳﺒﺎﺕ )ﺗﺮﻓﻨﺪﻫﺎﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ(‪ ،‬ﺧﻄﺎﻱ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﻛﺎﻫﺶ ﺩﺍﺩ ﻭ ﻧﺘﻴﺠﻪ ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺑﻪ ﻧﺘﻴﺠﻪ ﻭﺍﻗﻌﻲ‬
‫ﻧﺰﺩﻳﻚﺗﺮ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﻧﻤﻮﻧﻪﻫﺎﻳﻲ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻣﺜﺎﻝﻫﺎﻱ ﺍﻳﻦ ﺑﺨﺶ ﺭﺍ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺑﺎ ﺩﻗﺖ ﻣﻀﺎﻋﻒ‬
‫ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ ﺗﺎ ﭼﻨﻴﻦ ﺑﺮﺩﺍﺷﺖ ﻧﺸﻮﺩ ﻛﻪ ﺍﻳﻦ ﻧﻤﻮﻧﻪﻫﺎ ﻓﻘﻂ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺁﺯﻣﺎﻳﺸﻲ ﻗﺎﺑﻞ ﺍﻧﺠﺎﻡ ﻫﺴﺘﻨﺪ‪.‬‬

‫■ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﭼﻨﺪ ﻋﺪﺩ ﺍﺯ ﻛﻮﭼﻚ ﺑﻪ ﺑﺰﺭﮒ ﺑﺎﺷﺪ‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺧﺎﺻﻴﺖ ﺷﺮﻛﺖﭘﺬﻳﺮﻱ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﻛﺮﺩﻥ ﭼﻨﺪ ﻋﺪﺩ ﺩﺭ ﺣﺎﺻﻞﺟﻤﻊ‬
‫ﻣﺎﺷﻴﻨﻲ ﺗﺎﺛﻴﺮ ﺩﺍﺭﺩ‪ .‬ﺑﺎ ﺁﻧﺎﻟﻴﺰ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﻣﻲﺗﻮﺍﻥ ﺩﺭﻳﺎﻓﺖ ﻛﻪ ﺍﮔﺮ ﺍﻋﺪﺍﺩ ﺑﻪ ﺗﺮﺗﻴﺐ ﺻﻌﻮﺩﻱ ﻣﺮﺗﺐ ﺷﻮﻧﺪ ﻭ ﺳﭙﺲ ﺑﻪ ﺗﺮﺗﻴﺐ‬
‫ﻛﻮﭼﻚ ﺑﻪ ﺑﺰﺭﮒ ﺟﻤﻊ ﺷﻮﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﻛﻮﭼﻚﺗﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ‪ MATLAB‬ﺩﺭ ﻧﻈﺮ‬
‫ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۱−۲‬ﮐﺎﻫﺶ ﺧﻄﺎﯼ ﺭﻭﻧﺪﺁﻑ ﺩﺭ ﺟﻤ ﺍﻋﺪﺍﺩ ﺑﺎ ﻣﺮﺗﺐ ﮐﺮﺩﻥ ﺁﻥﻫﺎ )‪((۱‬‬

‫ﺩﺭ ‪ MATLAB‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻋﺪﺍﺩ ‪ x1 := 1‬ﻭ ‪ x2 = x2 = x4 = x5 = x6 = x7 = x8 = 1e − 16‬ﺭﺍ ﺑﺎ ﻫﻢ ﺟﻤﻊ‬
‫ﻧﻤﺎﻳﻴﻢ‪ .‬ﺩﺭ ﺯﻳﺮ‪ ،‬ﺣﺎﺻﻞﺟﻤﻊ ﺑﻪ ﺍﺯﺍﻱ ﺩﻭ ﺗﺮﺗﻴﺐ ﻣﺘﻔﺎﻭﺕ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﻭﻗﺘﻲ ﺍﻋﺪﺍﺩ ﺍﺯ ﻛﻮﭼﻚ ﺑﻪ ﺑﺰﺭﮒ ﺟﻤﻊ ﻣﻲﺷﻮﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺣﺎﺻﻞ ﺟﻤﻊ ﺑﻪ ﺻﻮﺭﺕ ﺩﻗﻴﻖ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ‬
‫ﻏﻴﺮ ﺍﻳﻦ ﺻﻮﺭﺕ‪،‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ ‪ 1e − 15‬ﺭﺥ ﻣﻲﺩﻫﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۵۶‬‬

‫ﻣﺜﺎﻝ ‪)۱۲−۲‬ﮐﺎﻫﺶ ﺧﻄﺎﯼ ﺭﻭﻧﺪﺁﻑ ﺩﺭ ﺟﻤ ﺍﻋﺪﺍﺩ ﺑﺎ ﻣﺮﺗﺐ ﮐﺮﺩﻥ ﺁﻥﻫﺎ )‪((۲‬‬

‫ﺩﻭ ﻣﺠﻤﻮﻉ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪X‬‬
‫‪1 e8‬‬
‫‪1‬‬ ‫‪X‬‬
‫‪1‬‬
‫‪1‬‬
‫=‪s1 :‬‬ ‫‪,‬‬ ‫=‪s2 :‬‬
‫‪n=1‬‬
‫‪n2‬‬ ‫‪n=1 e8‬‬
‫‪n2‬‬
‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﻫﺮ ﺩﻭ ﻣﺠﻤﻮﻉ ﻳﻜﺴﺎﻥ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻓﻘﻂ ﺗﺮﺗﻴﺐ ﺟﻤﻊ ﺍﻋﺪﺍﺩ ﺩﺭ ﺁﻥﻫﺎ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ .‬ﺩﺭ ‪ s1‬ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﻧﺰﻭﻟﻲ ﺟﻤﻊ ﻣﻲﺷﻮﻧﺪ ﻭﻟﻲ ﺩﺭ ‪ s2‬ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﺻﻌﻮﺩﻱ ﺟﻤﻊ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﺎ ﻛﻤﻚ ‪ MAPLE‬ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﻣﺠﻤﻮﻉ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪:‬‬
‫· · · ‪s1 = s2 = 1.644934056848226486472414999979358522552286567873‬‬
‫ﺩﺭ ‪ MATLAB‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﻭ ﻣﺠﻤﻮﻉ ﻓﻮﻕ ﺭﺍ ﺑﺎ ﻫﻤﺎﻥ ﺗﺮﺗﻴﺐ ﻣﺸﺨﺺ ﺷﺪﻩ ﺩﺭ ﺁﻥﻫﺎ ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﺑﻌﺪ ﺍﺯ ﻣﺤﺎﺳﺒﻪ ‪s1‬‬
‫ﻳﻌﻨﻲ ﺣﺎﻟﺘﻲ ﻛﻪ ﺗﺮﺗﻴﺐ ﺑﻪ ﺻﻮﺭﺕ ﻧﺰﻭﻟﻲ ﺍﺳﺖ ﺑﻪ ﺟﻮﺍﺏ ‪ 1.644934057834575‬ﻣﻲﺭﺳﻴﻢ ﻛﻪ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ‪9.86 e − 10‬‬
‫ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻣﺎ ﻣﺤﺎﺳﺒﻪ ‪ s2‬ﻳﻌﻨﻲ ﺟﻤﻊ ﺑﺎ ﺗﺮﺗﻴﺐ ﺻﻌﻮﺩﻱ‪ ،‬ﺑﻪ ﺟﻮﺍﺏ ‪ 1.644934056848226‬ﻣﻲﺭﺳﻴﻢ ﻛﻪ ﺑﻪ ﺍﻧﺪﺍﺯﻩ‬
‫‪ 4.865 e − 16‬ﺧﻄﺎﻱ ﻣﻄﻠﻖ )ﺗﺎ ‪ ۱۶‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ( ﺩﺍﺭﻳﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﻣﺮﺗﺐ ﻛﺮﺩﻥ ﺍﻋﺪﺍﺩ ﺑﻪ ﺻﻮﺭﺕ ﺻﻌﻮﺩﻱ ﻭ ﺳﭙﺲ ﺟﻤﻊ‬
‫ﺁﻥﻫﺎ‪ ،‬ﺣﺪﺍﻗﻞ ﺻﺪ ﻫﺰﺍﺭ ﺑﺮﺍﺑﺮ ﺩﻗﺖ ﺑﻬﺘﺮﻱ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪.‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺮﺗﺐ ﻛﺮﺩﻥ ﻭ ﺳﭙﺲ ﺟﻤﻊ ﺍﻋﺪﺍﺩ ﻫﻤﻮﺍﺭﻩ ﻣﻨﺠﺮ ﺑﻪ ﺧﻄﺎﻱ ﻛﻤﺘﺮﻱ ﻧﻤﻲﺷﻮﺩ‪ ،‬ﺑﻠﻜﻪ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎ ﻛﺎﻫﺶ‬
‫ﻣﻲﻳﺎﺑﺪ‪ .‬ﻟﺬﺍ ﺩﺭ ﻏﺎﻟﺐ ﻣﻮﺍﺭﺩ‪ ،‬ﻣﺮﺗﺐ ﻛﺮﺩﻥ ﻗﺒﻞ ﺍﺯ ﺟﻤﻊ ﻛﺮﺩﻥ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﻘﺪﺍﺭ ﺧﻄﺎ ﻛﺎﻫﺶ ﻳﺎﺑﺪ‪ .‬ﺍﻣﺎ ﻣﺜﺎﻝﻫﺎﻱ ﻧﻘﻀﻲ‬
‫ﻫﻢ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻣﺮﺗﺐ ﻛﺮﺩﻥ ﺍﻋﺪﺍﺩ ﻗﺒﻞ ﺍﺯ ﺟﻤﻊ ﺑﻪ ﺧﻄﺎﻱ ﺑﺰﺭﮒﺗﺮﻱ ﻣﻨﺠﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺗﻤﺮﻳﻦ ؟؟ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫■ ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺳﺮﺭﻳﺰﻱ )ﻭ ﻳﺎ ﺯﻳﺮﺭﻳﺰﻱ( ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ‬

‫ﻭﻗﺘﻲ ﻛﻪ ﻧﺘﻴﺠﻪ ﻳﻚ ﻣﺤﺎﺳﺒﻪ ﺑﻪ ﻟﺤﺎﻅ ﻗﺪﺭﻣﻄﻠﻖ ﺍﺯ ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺑﺰﺭﮒﺗﺮ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ ﻭ‬
‫ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺑﺎ ‪ ±Inf‬ﺟﺎﻳﮕﺰﻳﻦ ﻣﻲﺷﻮﺩ‪ .‬ﺳﺮﺭﻳﺰﻱ ﻣﻲﺗﻮﺍﻧﺪ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﻳﻚ ﻋﺒﺎﺭﺕ ﻳﺎ ﻳﻚ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‬
‫ﻭ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﺁﻥ ﻋﺒﺎﺭﺕ ﻳﺎ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺑﺮﺍﺑﺮ ‪ ±Inf‬ﻭ ﻳﺎ ‪ NaN‬ﺷﻮﺩ‪ .‬ﺳﺮﺭﻳﺰﻱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺍﺟﺘﻨﺎﺏ ﻧﺎﭘﺬﻳﺮ ﺍﺳﺖ‪ .‬ﺍﻣﺎ‬
‫ﻣﻮﺍﺭﺩﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻧﺘﻴﺠﻪ ﻳﻚ ﻋﺒﺎﺭﺕ ﻳﻚ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ﺍﺳﺖ ﻭ ﻣﻨﺠﺮ ﺑﻪ ﺳﺮﺭﻳﺰﻱ ﻧﻤﻲﺷﻮﺩ‪ ،‬ﺍﻣﺎ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺁﻥ‬
‫ﻋﺒﺎﺭﺕ ﺳﺮﺭﻳﺰﻱ ﺭﺥ ﻣﻲﺩﻫﺪ ﻭ ﺗﺒﻌﺎ ﻧﺘﻴﺠﻪ ﻣﺎﺷﻴﻨﻲ ﻋﺒﺎﺭﺕ ﺑﺮﺍﺑﺮ ‪ ±Inf‬ﻭ ﻳﺎ ‪ NaN‬ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﻣﻲﺗﻮﺍﻥ ﺗﺮﺗﻴﺐ ﻭ‬
‫ﻧﺤﻮﻩ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﻃﻮﺭﻱ ﺗﻐﻴﻴﺮ ﺩﺍﺩ ﻛﻪ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻧﻴﺎﻓﺘﺪ ﻭ ﻧﺘﻴﺠﻪ ﻣﻨﺎﺳﺒﻲ ﺑﺮﺍﻱ ﻋﺒﺎﺭﺕ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﻣﺜﺎﻝﻫﺎﻳﻲ ﺩﺭ ﺍﻳﻦ ﻣﻮﺭﺩ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۳−۲‬ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺳﺮﺭﯾﺰﯼ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺮﮐﯿﺐ(‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭ‬
‫ ‬
‫‪n‬‬ ‫!‪n‬‬
‫=‪:‬‬
‫‪r‬‬ ‫!‪(n − r)!r‬‬
‫ﺭﺍ ﺑﻪ ﺍﺯﺍﻱ ‪ n = 171‬ﻭ ‪ r = 4‬ﺩﺭ ‪ MATLAB‬ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺍﺯ ﺩﺳﺘﻮﺭ ﺯﻳﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫))‪factorial(171)/(factorial(4)*factorial(171-4‬‬
‫ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ‪ MATLAB‬ﺟﻮﺍﺏ ‪ 34389810‬ﺭﺍ ﻧﺘﻴﺠﻪ ﺩﻫﺪ‪ ،‬ﺍﻣﺎ ﺣﺎﺻﻞ ‪ Inf‬ﺭﺍ ﺑﺮﻣﻲﮔﺮﺩﺍﻧﺪ‪ .‬ﻋﻠﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ‬
‫‪ 171! ≃ 1.24 e309‬ﻛﻪ ﺍﺯ ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﻣﺎﺷﻴﻨﻲ ‪ 1.798e308‬ﺑﺰﺭﮒﺗﺮ ﺍﺳﺖ ﻭ ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ‪ MATLAB‬ﻣﻘﺪﺍﺭ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵۷‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫)‪ factorial(171‬ﺑﺮﺍﺑﺮ ‪ Inf‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ ﻭ ﻟﺬﺍ ﻣﻘﺪﺍﺭ ﻛﻞ ﻋﺒﺎﺭﺕ‬

‫!‪171‬‬
‫!‪(171 − 4)! 4‬‬
‫ﺑﺮﺍﺑﺮ ‪ Inf‬ﺧﻮﺍﻫﺪ ﺷﺪ‪ .‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﻧﺠﺎﻡ ﺩﺍﺩ ﻛﻪ ﺍﻳﻦ ﻣﺸﻜﻞ ﭘﻴﺶ ﻧﻴﺎﻳﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺍﺭﻳﻢ‪:‬‬
‫!‪171‬‬ ‫‪171 × 170 × 169 × 168‬‬
‫=‬
‫!‪(171 − 4)! 4‬‬ ‫!‪4‬‬
‫ﻟﺬﺍ ﺍﮔﺮ ﺍﺯ ﺳﻤﺖ ﺭﺍﺳﺖ ﻣﻌﺎﺩﻟﻪ ﻓﻮﻕ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺟﻮﺍﺏ ‪ 34389810‬ﺭﺍ ﺧﻮﺍﻫﻴﻢ ﮔﺮﻓﺖ‪ ،‬ﻛﻪ ﻫﻤﺎﻥ ﺟﻮﺍﺏ ﺩﻗﻴﻖ ﺍﺳﺖ‪.‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﺜﺎﻝ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﭼﻨﻴﻦ ﮔﻔﺖ ﻛﻪ ﺑﻬﺘﺮ ﺍﺳﺖ ﻛﺴﺮﻫﺎﻳﻲ ﺑﺎ ﺻﻮﺭﺕ ﻭ ﻣﺨﺮﺝ ﺑﺰﺭﮒ ﺭﺍ ﺑﻪ ﺣﺎﺻﻞﺿﺮﺏ ﭼﻨﺪ‬
‫ﻛﺴﺮ ﺑﺎ ﺻﻮﺭﺕ ﻭ ﻣﺨﺮﺝ ﻛﻮﭼﻚ )ﻭ ﻳﺎ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﻢ( ﻧﻮﺷﺖ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻣﺜﺎﻝ ﺩﻳﮕﺮﻱ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺑﺮﺩﺍﺭ(‬ ‫ﻣﺜﺎﻝ ‪)۱۴−۲‬ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺳﺮﺭﯾﺰﯼ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭ ﯾ‬

‫‪p‬‬
‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﻧﺮﻡ ﺩﻭ ﺑﺮﺩﺍﺭ ⊤] ‪ v := [v1 , . . . , vn‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ . v12 + · · · + vn2‬ﻣﻤﻜﻦ ﺍﺳﺖ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻳﻚ ﺑﺮﺩﺍﺭ‬
‫ﺍﻋﺪﺍﻱ ﻣﺎﺷﻴﻨﻲ ﺑﺎﺷﻨﺪ ﻭﻟﻲ ﻧﺮﻡ ﺩﻭﻱ ﺁﻥ ﻋﺪﺩﻱ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ‪ xmax‬ﺑﺎﺷﺪ ﻭ ﺗﺒﻌﺎ ﻫﻨﮕﺎﻡ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭﻱ ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﺳﺮﺭﻳﺰﻱ‬
‫ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻧﻴﺰ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻣﻘﺪﺍﺭ ﻧﺮﻡ ﺩﻭﻱ ﻳﻚ ﺑﺮﺩﺍﺭ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ﻣﺎﺷﻴﻦ ﻭﺍﻗﻊ ﺑﺎﺷﺪ ﻭﻟﻲ ﻫﻨﮕﺎﻡ‬
‫ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭﻱ ﺁﻥ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺑﺮﺩﺍﺭ ⊤]‪ v := [3 e200, 4 e200‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﻧﺮﻡ ﺩﻭﻱ‬
‫ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫√‬ ‫√‬
‫= ∥‪∥v‬‬ ‫= ‪9 e400 + 16 e400‬‬ ‫‪25e400 = 5 e200.‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺮﻡ ﺩﻭﻱ ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﻧﺮﻣﺎﻝ ‪ MATLAB‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭﻱ ﺍﻳﻦ‬
‫ﺑﺮﺩﺍﺭ ﺑﻪ ‪ 5e200‬ﻣﻨﺠﺮ ﺷﻮﺩ‪ .‬ﺍﻣﺎ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺑﺮﺩﺍﺭ ﻣﻄﺎﺑﻖ ﺑﺎ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺩﺭ ‪ MATLAB‬ﺑﻪ ‪ Inf‬ﻣﻨﺠﺮ ﻣﻲﺷﻮﺩ‪ .‬ﭼﺮﺍ ﻛﻪ‬
‫ﻣﺤﺎﺳﺒﻪ ‪ (3 e200)2‬ﻭ ‪ (4 e200)2‬ﺩﺭ ‪ MATLAB‬ﺑﻪ ‪ Inf‬ﻣﻨﺠﺮ ﻣﻲﺷﻮﺩ ﻭ ﺭﺍﺩﻳﻜﺎﻝ ﻣﺠﻤﻮﻉ ﺁﻥﻫﺎ ﻧﻴﺰ ‪ Inf‬ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺑﺮﻃﺮﻑ ﻛﺮﺩﻥ ﻣﺸﻜﻞ ﻓﻮﻕ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺑﺘﺪﺍ ﻣﺘﻐﻴﺮ ‪ vmax‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻛﺮﺩ‪:‬‬
‫}| ‪vmax := max {|v1 | , . . . , |vn‬‬
‫ﺣﺎﻝ ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪:‬‬
‫‪s‬‬ ‫‪2‬‬ ‫‬ ‫‪2‬‬
‫‪q‬‬
‫‪v1‬‬ ‫‪vn‬‬
‫‪v12 + · · · + vn2 = vmax‬‬ ‫‪+ ··· +‬‬
‫‪vmax‬‬ ‫‪vmax‬‬
‫ﺗﺒﻌﺎ ﺍﮔﺮ ﻧﺮﻡ ﺩﻭﻱ ﺑﺮﺩﺍﺭ ﻛﻮﭼﻚﺗﺮ ﺍﺯ ‪ xmax‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭﻱ ﺑﺮﺩﺍﺭ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺳﻤﺖ ﺭﺍﺳﺖ ﺗﺴﺎﻭﻱ ﻓﻮﻕ ﺑﻪ‬
‫ﺳﺮﺭﻳﺰﻱ ﻣﻨﺠﺮ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭ ﺑﺮﺩﺍﺭ ⊤]‪ v := [3 e200, 4e200‬ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪:‬‬
‫‪s‬‬ ‫‪2‬‬ ‫‬ ‫‪2‬‬ ‫‪r‬‬
‫‪3e200‬‬ ‫‪4 e200‬‬ ‫‪9‬‬
‫)‪∥v∥ = (4 e200‬‬ ‫‪+‬‬ ‫)‪= (4 e200‬‬ ‫‪+ 1 = 5 e200‬‬
‫‪4e200‬‬ ‫‪4 e200‬‬ ‫‪16‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺮﺧﻼﻑ ﻗﺒﻞ‪ ،‬ﺩﺭ ﻓﺮﻣﻮﻝ ﻓﻮﻕ‪ ،‬ﺑﺪﻭﻥ ﺳﺮﺭﻳﺰﻱ ﻣﻲﺗﻮﺍﻥ ﻧﺮﻡ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﻪ ﺧﻮﺑﻲ ﻣﺤﺎﺳﺒﻪ ﻛﺮﺩ‪.‬‬

‫ﺳﺮﺭﻳﺰﻱ ﺑﺎﻋﺚ ﻣﺨﺪﻭﺵ ﺷﺪﻥ ﻧﺘﻴﺠﻪ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻣﺤﺎﺳﺒﺎﺕ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺗﻐﻴﻴﺮ‬
‫ﺩﺍﺩ ﻛﻪ ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺯ ﺳﺮﺭﻳﺰﻱ ﺍﺟﺘﻨﺎﺏ ﮔﺮﺩﺩ )ﻧﻤﻮﻧﻪﻫﺎﻳﻲ ﺩﺭ ﺑﺎﻻ ﺁﻭﺭﺩﻩ ﺷﺪ(‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪،‬‬
‫ﺯﻳﺮﺭﻳﺰﻱ ﺑﺎﻋﺚ ﻛﺎﻫﺶ ﺩﻗﺖ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﺎﻧﻨﺪ ﺳﺮﺭﻳﺰﻱ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﻧﺘﻴﺠﻪ ﻳﻚ ﻋﺒﺎﺭﺕ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻗﺮﺍﺭ‬
‫ﻧﮕﻴﺮﺩ ﻭﻟﻲ ﻧﺘﻴﺠﻪ ﺑﺮﺧﻲ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺣﺎﻻﺕ‪ ،‬ﺩﻗﺖ ﺁﻥ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﻛﻪ ﺩﺭ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۵۸‬‬

‫ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻭﺍﻗﻊ ﻣﻲﺷﻮﻧﺪ ﭘﺎﻳﻴﻦ ﻣﻲﺁﻳﺪ ﻭ ﻟﺬﺍ ﺩﻗﺖ ﻛﻞ ﺭﺍ ﺗﺤﺖ ﺗﺎﺛﻴﺮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﻛﻪ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﻻﺯﻡ ﺑﺮﺍﻱ ﻳﻚ ﻋﺒﺎﺭﺕ ﺭﺍ ﻃﻮﺭﻱ ﺗﻐﻴﻴﺮ ﺩﺍﺩ ﻛﻪ ﻫﻴﭻ ﻛﺪﺍﻡ ﺍﺯ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺩﺭ ﻣﺤﺪﻭﺩﻩ ﺯﻳﺮﺭﻳﺰﻱ ﻭﺍﻗﻊ ﻧﺸﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪،‬‬
‫ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ ﻧﻴﺎﻓﺘﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻧﺮﻡ ﺩﻭ ﻳﻚ ﺑﺮﺩﺍﺭ ﻛﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺁﻥ ﺍﻋﺪﺍﺩ ﻛﻮﭼﻚ‬
‫ﻫﺴﺘﻨﺪ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺧﻄﺎﻱ ﺯﻳﺮﺭﻳﺰﻱ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﻮﺭﺩ ﻧﻴﺰ ﻫﻤﺎﻧﻨﺪ ﻣﺜﺎﻝ ‪ ۱۴-۲‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺧﻄﺎﻱ‬
‫ﺯﻳﺮﺭﻳﺰﻱ ﺟﻠﻮﮔﻴﺮﻱ ﻛﺮﺩ‪.‬‬

‫■ ﺍﺯ ﺍﻋﺪﺍﺩ ﻣﻤﻴﺰ ﺷﻨﺎﻭﺭ ﺑﻪ ﻋﻨﻮﺍﻥ ﺷﻤﺎﺭﻧﺪﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻧﺸﻮﺩ‬

‫ﺗﻤﺎﻣﻲ ﺯﺑﺎﻥﻫﺎﻱ ﺑﺮﻧﺎﻣﻪﻧﻮﻳﺴﻲ ﺩﺍﺭﺍﻱ ﺍﺑﺰﺍﺭ ﺣﻠﻘﻪ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﻛﺎﺭﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺣﻠﻘﻪﻫﺎ ﻣﻌﻤﻮﻻ ﻳﻚ ﻣﺘﻐﻴﺮ ﺷﻤﺎﺭﻧﺪﻩ‬
‫ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺍﺻﻮﻻ ﺍﻳﻦ ﻣﺘﻐﻴﺮ ﺷﻤﺎﺭﻧﺪﻩ ﺍﺯ ﻧﻮﻉ ﺻﺤﻴﺢ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺷﻤﺎﺭﻧﺪﻩﻫﺎﻱ ﺍﻋﺸﺎﺭﻱ ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪.‬‬
‫ﺗﻮﺻﻴﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﺯ ﺷﻤﺎﺭﻧﺪﻩﻫﺎﻱ ﺍﻋﺸﺎﺭﻱ ﺑﺮﺍﻱ ﺣﻠﻘﻪﻫﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻧﺸﻮﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﮔﺎﻫﻲ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﺩﺭ ﺍﻋﺪﺍﺩ ﺍﻋﺸﺎﺭﻱ‬
‫ﻣﻨﺠﺮ ﺑﻪ ﺁﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﺣﻠﻘﻪﻫﺎ ﺑﺎ ﺗﻜﺮﺍﺭﻫﺎﻱ ﺍﺷﺘﺒﺎﻩ ﺍﻧﺠﺎﻡ ﺩﻫﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻛﺪ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﺯﺑﺎﻥ ‪ C‬ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬

‫;‪float i,x‬‬
‫)‪for(i=0.0 ; i<1.0 ; i+=0.1‬‬
‫;‪x=i‬‬
‫;)‪printf("%f",x‬‬
‫ﻳﺎ ﺑﺮﻧﺎﻣﻪ ﻣﻌﺎﺩﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ‪ MATLAB‬ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬

‫;‪x=0‬‬
‫‪while x<1‬‬
‫;‪x=x+0.1‬‬
‫‪end‬‬
‫)‪disp(x‬‬
‫ﺍﻧﺘﻈﺎﺭ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﻫﺮ ﺩﻭ ﺑﺮﻧﺎﻣﻪ ﻣﻘﺪﺍﺭ ﺧﺮﻭﺟﻲ ﺑﺮﺍﺑﺮ ‪ 1.0‬ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺟﺎﻟﺐ ﺍﺳﺖ ﻛﻪ ﺑﺪﺍﻧﻴﺪ ﻛﻪ ﺧﺮﻭﺟﻲ ﻫﺮ ﺩﻭ ﺑﺮﻧﺎﻣﻪ‬
‫‪ 1.1‬ﺍﺳﺖ‪ .‬ﻋﻠﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺷﺮﻁ ﺧﺎﺗﻤﻪ ﺣﻠﻘﻪ ﺑﻪ ﺩﺭﺳﺘﻲ ﻋﻤﻞ ﻧﻜﻨﺪ‪.‬‬

‫■ ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ‬

‫ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻫﻨﮕﺎﻣﻲ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ ﻛﻪ ﺑﺨﻮﺍﻫﻴﻢ ﺩﻭ ﻋﺪﺩ ﻧﺰﺩﻳﻚ ﺑﻪ ﻫﻢ ﺭﺍ ﺗﻔﺮﻳﻖ ﻛﻨﻴﻢ‪ .‬ﺍﮔﺮ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﻳﻚ‬
‫ﺍﻟﮕﻮﺭﻳﺘﻢ ﻳﺎ ﻳﻚ ﻋﺒﺎﺭﺕ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺩﺭ ﻧﺘﻴﺠﻪ ﺁﻥ ﺍﻟﮕﻮﺭﻳﺘﻢ ﻭ ﻳﺎ ﻋﺒﺎﺭﺕ ﺧﻄﺎﻱ‬
‫ﺑﺰﺭﮔﻲ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ .‬ﺑﻪ ﺍﺻﻄﻼﺡ‪ ،‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭ ﻧﺘﻴﺠﻪ ﺣﺎﺻﻞ ﺷﺪﻩ ﻛﺎﻫﺶ ﺯﻳﺎﺩﻱ ﺑﻴﺎﺑﺪ‪ .‬ﺑﺮﺍﻱ ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺧﻄﺎﻱ‬
‫ﻛﻨﺴﻠﻴﺸﻦ‪ ،‬ﮔﺎﻫﺎ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺩﻭﺑﺎﺭﻩ ﻓﺮﻣﻮﻟﻪ ﻛﺮﺩﻥ ﺑﺎ ﻛﺴﺮﻫﺎ‪ ،‬ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺑﺴﻂ ﺗﻮﺍﺑﻊ ﻭ ﻳﺎ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺍﺗﺤﺎﺩﻫﺎ ﻭ ﻋﺒﺎﺭﺕﻫﺎﻱ‬
‫ﻣﻌﺎﺩﻝ ﺍﺯ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﺍﺟﺘﻨﺎﺏ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺯﻳﺮ ﻣﺜﺎﻝﻫﺎﻳﻲ ﺩﺭ ﺍﻳﻦ ﺯﻣﻴﻨﻪ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۱۵−۲‬ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺧﻄﺎﯼ ﮐﻨﺴﻠﯿﺸﻦ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻣﻘﺪﺍﺭ ﯾ‬

‫ﺗﺎﺑﻊ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪p‬‬
‫‪f (x) := x2 + 1 − 1‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۵۹‬‬ ‫ﺧﻄﺎ ﻭ ﺁﻧﺎﻟﯿﺰ ﺧﻄﺎ ﺩﺭ ﻣﺎﺷﯿﻦﻫﺎ‬ ‫ﻓﺼﻞ ﺩﻭﻡ‪:‬‬

‫ﺷﻜﻞ ‪۱.۲‬‬

‫ﺍﮔﺮ ‪ ،x ≃ 0‬ﺁﻥﮔﺎﻩ ﺩﺭ ﻣﺤﺎﺳﺒﻪ )‪ f (x‬ﺩﭼﺎﺭ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻣﻲﺷﻮﻳﻢ‪ .‬ﺑﺮﺍﻱ ﺑﺮﻃﺮﻑ ﻛﺮﺩﻥ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﺑﻪ ﺍﺯﺍﻱ ‪x‬ﻫﺎﻱ‬
‫ﻧﺰﺩﻳﻚ ﺑﻪ ﺻﻔﺮ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻧﻤﻮﺩ‪:‬‬
‫‪p‬‬ ‫‪p‬‬ ‫‪ √x 2 + 1 + 1‬‬ ‫‪x2‬‬
‫=‪f (x) :‬‬ ‫‪x2‬‬ ‫=‪+1−1‬‬ ‫‪x2‬‬ ‫√ ‪+1−1‬‬ ‫√=‬
‫‪x2 + 1 + 1‬‬ ‫‪x2 + 1 + 1‬‬
‫ﺩﺭ ﺿﺎﺑﻄﻪ ﻓﻮﻕ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻧﺨﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ ﻭ ﻟﺬﺍ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﻛﻤﺘﺮﻱ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۶−۲‬ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﺧﻄﺎﯼ ﮐﻨﺴﻠﯿﺸﻦ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺭﯾﺸەﻫﺎﯼ ﻣﻌﺎﺩﻟﻪ ﺩﺭﺟﻪ ﺩﻭ(‬

‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺭﻳﺸﻪﻫﺎﻱ ﻣﻌﺎﺩﻟﻪ ‪ ax2 + bx + c = 0‬ﻋﺒﺎﺭﺗﻨﺪ ﺍﺯ‪:‬‬
‫√‬ ‫√‬
‫‪−b −‬‬‫‪b2 − 4ac‬‬ ‫‪−b + b2 − 4ac‬‬
‫=‪x1 :‬‬ ‫‪,‬‬ ‫=‪x2 :‬‬
‫√‬ ‫‪2a‬‬ ‫‪2a‬‬
‫ﻫﻨﮕﺎﻣﻲ ﻛﻪ ‪ a‬ﻳﺎ ‪ c‬ﻭ ﻳﺎ ﻫﺮ ﺩﻭ ﻧﺴﺒﺖ ﺑﻪ ‪ b‬ﺑﺴﻴﺎﺭ ﻛﻮﭼﻚ ﺑﺎﺷﻨﺪ ﻭ ﻫﻤﻴﻦﻃﻮﺭ ‪ ،b > 0‬ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ‪ b2 − 4ac‬ﺑﻪ ‪ b‬ﻧﺰﺩﻳﻚ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺻﻮﺭﺕ ﻛﺴﺮ ﻣﺮﺑﻮﻁ ﺑﻪ ‪ x2‬ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﺭﺥ ﻣﻲﺩﻫﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ‪ x1‬ﺩﭼﺎﺭ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ‬
‫ﻧﻤﻲﺷﻮﻳﻢ‪.‬‬
‫ﺑﺮﺍﻱ ﺁﻥﻛﻪ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ‪ x2‬ﺩﭼﺎﺭ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻧﺸﻮﻳﻢ‪ ،‬ﻣﻲﺗﻮﺍﻥ ‪ x2‬ﺭﺍ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺭﺍﺑﻄﻪ ‪ x1 x2 = a‬ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ‪ .‬ﻳﻌﻨﻲ‬
‫‪c‬‬

‫ﺑﻌﺪ ﺍﺯ ﻣﺤﺎﺳﺒﻪ ‪ ،x1‬ﺍﺯ ﺭﺍﻳﻄﻪ ﺯﻳﺮ ﻣﻘﺪﺍﺭ ‪ x2‬ﺭﺍ ﺑﺪﻭﻥ ﺧﻄﺎﻱ ﻛﻨﺴﻠﻴﺸﻦ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪c‬‬
‫= ‪x2‬‬ ‫‪.‬‬
‫‪ax1‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶۰‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۳‬‬
‫ﻣﻘﺪﻣﺎﺗ ﺍﺯ ﺟﺒﺮﺧﻄ‬
‫ﻓﺼﻞ ﺳﻮﻡ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۶۱‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶۲‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۴‬‬
‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄ‬
‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۶۵‬‬ ‫ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴‬‬
‫‪۶۵‬‬ ‫ﺑﺮﺩﺍﺭ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۴‬‬
‫‪۶۸‬‬ ‫ﻣﺎﺗﺮﻳﺲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۴‬‬
‫‪۷۵‬‬ ‫ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺧﺎﺹ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۱.۴‬‬
‫‪۷۹‬‬ ‫ﻣﻌﺮﻓﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۴‬‬
‫‪۸۲‬‬ ‫ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۴‬‬
‫‪۸۴‬‬ ‫ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۴‬‬
‫‪۸۴‬‬ ‫ﺭﻭﺵﻫﺎﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺮﺑﻌﻲ ‪. . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۲.۴‬‬
‫‪۸۵‬‬ ‫ﺭﻭﺵ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﺭﻭﺵ ﻛﺮﺍﻣﺮ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۲.۴‬‬
‫‪۸۶‬‬ ‫ﺭﻭﺵ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۴‬‬
‫‪۸۶‬‬ ‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻭ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۴‬‬
‫‪۹۰‬‬ ‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪. . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۳.۴‬‬
‫‪۹۷‬‬ ‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۳.۴‬‬
‫‪۱۰۲‬‬ ‫∗ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۴‬‬
‫‪۱۰۵‬‬ ‫ﺧﻄﺎ ﻭ ﻛﻨﺘﺮﻝ ﺧﻄﺎ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۴‬‬
‫‪۱۰۷‬‬ ‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۴‬‬
‫‪۱۱۰‬‬ ‫ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﻳﮕﺮ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۶.۴‬‬
‫‪۱۱۰‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . .‬‬ ‫‪۱.۶.۴‬‬
‫‪۱۱۱‬‬ ‫ﻣﺤﺎﺳﺒﻪ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . . . . . . . . . .‬‬ ‫‪۲.۶.۴‬‬
‫‪۱۱۳‬‬ ‫ﺗﺸﺨﻴﺺ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ‪. . . . . . .‬‬ ‫‪۳.۶.۴‬‬

‫‪۶۳‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶۴‬‬

‫‪۱۱۵‬‬ ‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۷.۴‬‬

‫‪۱۱۵‬‬ ‫ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۷.۴‬‬
‫‪۱۲۰‬‬ ‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۷.۴‬‬
‫‪۱۲۲‬‬ ‫ﻫﻤﮕﺮﺍﻳﻲ ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۷.۴‬‬
‫‪۱۲۴‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ‪. . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۷.۴‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۶۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﺍﺯ ﭘﺮﻛﺎﺭﺑﺮﺩﺗﺮﻳﻦ ﻣﺴﺎﻳﻞ ﺩﺭ ﻣﻬﻨﺪﺳﻲ ﻭ ﻋﻠﻮﻡ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻛﺎﺭﺑﺮﺩﻫﺎ ﻇﺎﻫﺮ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ ﺍﺑﺘﺪﺍ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮﺧﻄﻲ ﻳﺎﺩﺁﻭﺭﻱ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺑﻪ ﻣﻌﺮﻓﻲ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻭ ﺧﻮﺍﺹ ﺁﻥﻫﺎ‬
‫ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﺳﭙﺲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﻭ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻭ ﻣﺴﺎﻳﻞ ﻣﺮﺑﻮﻁ ﺑﻪ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮﺧﻄﻲ‬ ‫‪۱.۴‬‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺑﻪ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺯ ﺟﺒﺮ ﺧﻄﻲ ﺩﺭ ﺧﺼﻮﺹ ﺑﺮﺩﺍﺭﻫﺎ ﻭ ﻣﺎﺗﺮﻳﺲﻫﺎ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺑﺮﺩﺍﺭ‬ ‫�‬
‫ﻳﻚ ﺑﺮﺩﺍﺭ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻳﻚ ﻟﻴﺴﺖ ﺍﺯ ﺍﻋﺪﺍﺩ‪ ،‬ﻛﻪ ﺗﺮﺗﻴﺐ ﺍﻳﻦ ﺍﻋﺪﺍﺩ ﺩﺭ ﻟﻴﺴﺖ ﻣﻬﻢ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺟﺒﺮ ﺧﻄﻲ‪ ،‬ﻣﻌﻤﻮﻻ ﺑﺮﺩﺍﺭﻫﺎ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﻳﻚ ﺳﺘﻮﻥ ﻋﻤﻮﺩﻱ ﺍﺯ ﺍﻋﺪﺍﺩ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺑﺎ ﭘﺮﺍﻧﺘﺰ ﻳﺎ ﺑﺮﺍﻛﺖ ﺍﺣﺎﻃﻪ ﺷﺪﻩ ﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺩﺭ ﺯﻳﺮ ﻳﻚ ﺑﺮﺩﺍﺭ ﺁﻭﺭﺩﻩ‬
‫ﻣﻲﺷﻮﺩ‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪12‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪−3.7‬‬ ‫‪−3.7‬‬
‫‪‬‬ ‫‪‬‬ ‫ﻳﺎ‬ ‫‪‬‬ ‫‪‬‬
‫‪ 0 ‬‬ ‫‪ 0 .‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪34.8‬‬ ‫‪34.8‬‬
‫ﺑﺮﺩﺍﺭ ﻓﻮﻕ ﺳﺘﻮﻧﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﺮﺩﺍﺭﻫﺎ ﺳﻄﺮﻱ ﻧﻴﺰ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﻮﻧﺪ‪ .‬ﻓﺮﻡﻫﺎﻱ ﺳﻄﺮﻱ ﺑﺮﺩﺍﺭ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺍﺳﺖ‪:‬‬
‫‪h‬‬ ‫‪i‬‬ ‫‬ ‫‬
‫‪12 −3.7 0‬‬ ‫‪34.8‬‬ ‫ﻳﺎ‬ ‫‪12‬‬ ‫‪−3.7‬‬ ‫‪0‬‬ ‫‪34.8 .‬‬
‫ﺍﻋﺪﺍﺩ ﺩﺍﺧﻞ ﺑﺮﺩﺍﺭ ﺭﺍ ﻣﻮﻟﻔﻪﻫﺎﻱ ‪ ۱‬ﺑﺮﺩﺍﺭ ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺍﺯ ﻭﺍﮊﻩﻫﺎﻱ ﻋﻨﺼﺮ ﻭ ﺩﺭﺍﻳﻪ ﻧﻴﺰ ﺑﻪ ﺟﺎﻱ ﻣﻮﻟﻔﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻳﻚ‬
‫ﺑﺮﺩﺍﺭ ﺑﺎ ‪ n‬ﻣﻮﻟﻔﻪ ﺭﺍ ﻳﻚ ﺑﺮﺩﺍﺭ ‪-n‬ﺗﺎﻳﻲ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬
‫ﺍﺯ ﺑﺮﺩﺍﺭﻫﺎ ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻭﻳﮋﮔﻲﻫﺎ ﻳﺎ ﻣﻘﺎﺩﻳﺮ ﻳﻚ ﭼﻴﺰ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻫﻨﺪﺳﻪ‬
‫ﺩﻛﺎﺭﺗﻲ ﺩﻳﺪﻳﻢ‪ ،‬ﺍﺯ ﻳﻚ ﺑﺮﺩﺍﺭ ﺩﻭﺗﺎﻳﻲ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﻳﻚ ﻧﻘﻄﻪ ﺩﺭ ﺻﻔﺤﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺎ ﻳﻚ ﺑﺮﺩﺍﺭ‬
‫ﺳﻪﺗﺎﻳﻲ ﻣﻲﺗﻮﺍﻥ ﺁﺩﺭﺱ ﻳﺎ ﻣﻮﻗﻌﻴﺖ ﻳﻚ ﻧﻘﻄﻪ ﺩﺭ ﻓﻀﺎ ﺭﺍ ﻣﺸﺨﺺ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﻳﮕﺮ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻧﻤﺎﻳﺶ ﻳﻚ ﺭﻧﮓ‬
‫ﺑﺎ ﻳﻚ ﺑﺮﺩﺍﺭ ﺳﻪﺗﺎﻳﻲ ﺍﺷﺎﺭﻩ ﻛﺮﺩ ﻛﻪ ﺳﻪ ﻣﻮﻟﻔﻪ ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﻣﻴﺰﺍﻥ ﺷﺪﺕ ﺭﻧﮓ ﻗﺮﻣﺰ‪ ،‬ﺳﺒﺰ ﻭ ﺁﺑﻲ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪.‬‬

‫■ ﻧﻤﺎﺩﮔﺬﺍﺭﻱ ﺩﺭ ﺑﺮﺩﺍﺭﻫﺎ‬
‫ﺩﺭ ﺍﻳﻦ ﻛﺘﺎﺏ‪ ،‬ﻗﻮﺍﻋﺪ ﺯﻳﺮ ﺩﺭ ﻧﻤﺎﻳﺶ ﻭ ﻧﺎﻡﮔﺬﺍﺭﻱ ﺑﺮﺩﺍﺭﻫﺎ ﺭﻋﺎﻳﺖ ﻣﻲﺷﻮﺩ‪.‬‬

‫• ﺑﺮﺩﺍﺭﻫﺎ ﺭﺍ ﺑﻪ ﻓﺮﻡ ﺳﺘﻮﻧﻲ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺩﺭ ﻧﻤﺎﻳﺶ ﺁﻥﻫﺎ ﺍﺯ ﺑﺮﺍﻛﺖ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻳﻚ ﺑﺮﺩﺍﺭ‬
‫ﺭﺍ ﺑﺎ ﺳﺘﻮﻧﻲ ﺍﺯ ﺍﻋﺪﺍﺩ ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﺑﺎ ﺑﺮﺍﻛﺖ ﺍﺣﺎﻃﻪ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫• ﻣﺠﻤﻮﻋﻪ ﺗﻤﺎﻣﻲ ﺑﺮﺩﺍﺭﻫﺎﻱ ‪-n‬ﺗﺎﻳﻲ ﺭﺍ ﺑﺎ ‪ Rn‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫• ﺑﺮﺍﻱ ﻧﺎﻣﮕﺬﺍﺭﻱ ﻳﻚ ﺑﺮﺩﺍﺭ ﺍﺯ ﺣﺮﻭﻑ ﺍﻧﮕﻠﻴﺴﻲ ﻛﻮﭼﻚ ﻭ ﺑﺮﺟﺴﺘﻪ ‪ ۲‬ﻣﺎﻧﻨﺪ ‪ a‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺑﺮﺩﺍﺭ‪ ،‬ﺑﺎ ﺣﺮﻭﻑ‬
‫ﻛﻮﭼﻚ ﻭ ﻏﻴﺮﺑﻮﻟﺪ ﻣﺘﻨﺎﻇﺮ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﻳﻚ ﺑﺮﺩﺍﺭ ‪-n‬ﺗﺎﻳﻲ ﺑﺎ ﻧﺎﻡ ‪ a‬ﺭﺍ ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪1 Elements‬‬ ‫)‪(Components) (Entries) (Coefficients‬‬ ‫‪2 Bold‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶۶‬‬

‫ﺯﻳﺮ ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ‪:‬‬

‫‪‬‬ ‫‪‬‬
‫‪a1‬‬
‫‪ ‬‬
‫‪ a2 ‬‬
‫‪ ‬‬
‫‪a= . ‬‬
‫‪ .. ‬‬
‫‪ ‬‬
‫‪an‬‬
‫ﺍﻟﺒﺘﻪ ﺑﺮﺍﻱ ﺟﻤﻊﺗﺮ ﻧﻮﺷﺘﻦ ﺑﺮﺩﺍﺭ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻧﻤﺎﺩ ⊤ ﻧﻴﺰ ﻛﻤﻚ ﮔﺮﻓﺖ‪ ،‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﻧﻤﺎﻳﺶ ﺑﺮﺩﺍﺭ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‬
‫‪h‬‬ ‫⊤‪i‬‬
‫‪a = a1‬‬ ‫‪a2‬‬ ‫‪...‬‬ ‫‪an‬‬ ‫‪.‬‬
‫ﻧﻤﺎﺩ ⊤ ﺑﻪ ﺗﺮﺍﻧﻬﺎﺩﻩ ‪ ۳‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ ﻭ ﺑﺮﺩﺍﺭ ﺳﻄﺮﻱ ﺭﺍ ﺑﻪ ﺑﺮﺩﺍﺭ ﻣﺘﻨﺎﻇﺮ ﺳﺘﻮﻧﻲ )ﻭ ﺑﺮﻋﻜﺲ( ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪.‬‬

‫■ ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺩﺭ ﺑﺮﺩﺍﺭﻫﺎ‬

‫ﺩﻭ ﺑﺮﺩﺍﺭ‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬
‫‪a1‬‬ ‫‪b1‬‬
‫‪ ‬‬ ‫‪ ‬‬
‫‪ a2 ‬‬ ‫‪ b2 ‬‬
‫‪ ‬‬ ‫‪ ‬‬
‫‪a =  . ,‬‬ ‫‪b= . ‬‬ ‫)‪(۱.۴‬‬
‫‪ .. ‬‬ ‫‪ .. ‬‬
‫‪ ‬‬ ‫‪ ‬‬
‫‪an‬‬ ‫‪bn‬‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﺍﻳﻦ ﺩﻭ ﺑﺮﺩﺍﺭ ﺑﻪ ﺻﻮﺭﺕ ﻣﻮﻟﻔﻪﺍﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a 1 + b1‬‬ ‫‪a 1 − b1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a 2 + b2 ‬‬ ‫‪ a 2 − b2 ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a+b=‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫‪a−b=‬‬ ‫‪.. ‬‬
‫‪‬‬ ‫‪. ‬‬ ‫‪‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a n + bn‬‬ ‫‪a n − bn‬‬
‫ﺩﺭ ﻣﻮﺭﺩ ﺿﺮﺏ ﺩﻭ ﺑﺮﺩﺍﺭ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺍﺷﺎﺭﻩ ﺷﻮﺩ ﻛﻪ ﺿﺮﺏﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺩﺭ ﺑﺮﺩﺍﺭﻫﺎ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ ﻛﻪ ﻫﺮﻛﺪﺍﻡ ﺑﻨﺎ ﺑﺮ ﻛﺎﺭﺑﺮﺩﻱ‬
‫ﺧﺎﺹ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺿﺮﺏ ﺩﺍﺧﻠﻲ ‪) ۴‬ﻳﺎ ﺿﺮﺏ ﻧﻘﻄﻪﺍﻱ ‪ ،(۵‬ﺿﺮﺏ ﺧﺎﺭﺟﻲ ‪ ۶‬ﻭ ﺿﺮﺏ ﻣﻮﻟﻔﻪﺍﻱ ‪) ۷‬ﻳﺎ ﺿﺮﺏ‬
‫ﻫﺎﺩﺍﻣﺎﺭﺩﻱ ‪ (۸‬ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺩﻭ ﺑﺮﺩﺍﺭ ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺩﺭ )‪ ،(۱.۴‬ﺿﺮﺏ ﻧﻘﻄﻪﺍﻱ ﺑﺎ ﻧﻤﺎﺩ ‪ a · b‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫‪X‬‬
‫‪n‬‬
‫=‪a · b :‬‬ ‫‪a i bi .‬‬ ‫)‪(۲.۴‬‬
‫‪i=1‬‬
‫ﺿﺮﺏ ﻣﻮﻟﻔﻪﺍﻱ ﺩﻭ ﺑﺮﺩﺍﺭ ‪ a‬ﻭ ‪ b‬ﺑﺎ ﻧﻤﺎﺩ ‪ a ⊙ b‬ﻭ ﻳﺎ ﻧﻤﺎﺩ ‪ a. ∗ b‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫‪‬‬ ‫‪‬‬
‫‪a 1 b1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ a 2 b2 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪a ⊙ b :=  .  .‬‬ ‫)‪(۳.۴‬‬
‫‪ .. ‬‬
‫‪‬‬ ‫‪‬‬
‫‪a n bn‬‬

‫‪3 Transpose‬‬ ‫‪6 Cross‬‬ ‫‪product‬‬

‫‪4 Inner‬‬‫‪product‬‬ ‫‪7 Element-wise‬‬ ‫‪product‬‬
‫‪5 Dot product‬‬ ‫‪8 Hadamard‬‬ ‫‪product‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۶۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺿﺮﺏ ﺩﺍﺧﻠﻲ ﺩﻭ ﺑﺮﺩﺍﺭ ﻳﻚ ﻋﺪﺩ)ﺍﺳﻜﺎﻟﺮ( ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺿﺮﺏ ﻣﻮﻟﻔﻪﺍﻱ ﺩﻭ ﺑﺮﺩﺍﺭ ﻳﻚ ﺑﺮﺩﺍﺭ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ )ﺍﺳﻜﺎﻟﺮﻫﺎ(‪ ،‬ﻫﺮ ﺩﻭ ﻋﺪﺩ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﻳﻜﺪﻳﮕﺮ ﻣﻘﺎﻳﺴﻪ ﻛﺮﺩ‪ .‬ﺍﺻﻮﻻ ﺩﻭ ﺍﺳﻜﺎﻟﺮ ﻳﺎ ﺑﺎ ﻫﻢ ﻣﺴﺎﻭﻳﻨﺪ ﻳﺎ ﻳﻜﻲ ﺍﺯ‬
‫ﺩﻳﮕﺮﻱ ﺑﺰﺭﮒﺗﺮ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻝ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺩﻭ ﺑﺮﺩﺍﺭ ﺩﻟﺨﻮﺍﻩ ﺭﺍ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﻘﺎﻳﺴﻪ ﻧﻤﻮﺩ؟ ﭘﺎﺳﺦ ﻣﻨﻔﻲ ﺍﺳﺖ‪.‬‬
‫ﺍﺻﻮﻻ ﻫﺮ ﻣﻌﻴﺎﺭ ﻭ ﻗﺎﻋﺪﻩﺍﻱ ﻛﻪ ﺑﺮﺍﻱ ﻣﻘﺎﻳﺴﻪ ﺩﻭ ﺑﺮﺩﺍﺭ ﺍﺭﺍﻳﻪ ﻧﻤﺎﻳﻴﻢ ﺩﺭ ﻧﻬﺎﻳﺖ ﺑﻪ ﺗﻨﺎﻗﺾ ﻣﻨﺠﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻘﺎﻳﺴﻪ ﺩﻭ‬
‫ﺑﺮﺩﺍﺭ ﺩﻟﺨﻮﺍﻩ ﺍﻣﻜﺎﻥﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺍﺻﻄﻼﺣﺎ ﮔﻔﺘﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺮﺩﺍﺭﻫﺎ ﻣﻘﺎﻳﺴﻪﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﻨﺪ ﻭ ﻳﺎ ﻣﺠﻤﻮﻋﻪ ﺑﺮﺩﺍﺭﻫﺎﻱ ‪-n‬ﺗﺎﻳﻲ‬
‫)ﻛﻪ ‪ (n > 1‬ﺗﺮﺗﻴﺐﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺍﻧﺪﺍﺯﻩ ﺩﻭ ﺑﺮﺩﺍﺭ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﻘﺎﻳﺴﻪ ﻛﺮﺩ‪ .‬ﺩﺭ ﻫﻨﺪﺳﻪ ﺍﻧﺪﺍﺯﻩ ﺑﺮﺩﺍﺭ ﻋﺒﺎﺭﺕ‬
‫ﺍﺳﺖ ﺍﺯ ﻃﻮﻝ ﺍﻗﻠﻴﺪﺳﻲ ﺑﺮﺩﺍﺭ‪ .‬ﺩﺭ ﺟﺒﺮﺧﻄﻲ ﻣﻔﻬﻤﻮﻡ ﻛﻠﻲﺗﺮﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﻧﺮﻡ ‪ ۹‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺁﻥ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫■ ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ‬
‫ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ ﺗﺎﺑﻌﻲ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﺑﺮﺩﺍﺭ ﺭﺍ ﻣﻲﮔﻴﺮﺩ ﻭ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻧﺎﻣﻨﻔﻲ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺍﻧﺪﺍﺯﻩ ﺑﺮﺩﺍﺭ‪ ،‬ﺍﺭﺯﺵ ﺑﺮﺩﺍﺭ ﻳﺎ ‪ ...‬ﭘﺲ‬
‫ﻣﻲﺩﻫﺪ ﻭ ﺩﺍﺭﺍﻱ ﺧﻮﺍﺹ ﺯﻳﺮ ﺍﺳﺖ‬
‫• ﻧﺮﻡ ﺑﺮﺩﺍﺭ ﺻﻔﺮ )ﺑﺮﺩﺍﺭﻱ ﻛﻪ ﺗﻤﺎﻡ ﻣﻮﻟﻔﻪﻫﺎﻳﺶ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ( ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﻧﺮﻡ ﻫﺮ ﺑﺮﺩﺍﺭ ﻏﻴﺮ ﺻﻔﺮ ﻋﺪﺩﻱ ﻣﺜﺒﺖ‬
‫ﺍﺳﺖ‪.‬‬
‫• ﺍﮔﺮ ﺑﺮﺩﺍﺭﻱ ﺭﺍ ﺩﺭ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ λ‬ﺿﺮﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺮﻡ ﺑﺮﺩﺍﺭ ﺣﺎﺻﻞ |‪ |λ‬ﺑﺮﺍﺑﺮ ﻣﻲﺷﻮﺩ‪.‬‬
‫• ﻧﺮﻡ ﺣﺎﺻﻞﺟﻤﻊ ﺩﻭ ﺑﺮﺩﺍﺭ ﻛﻮﭼﻚﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﻣﺠﻤﻮﻉ ﻧﺮﻡﻫﺎﻱ ﺩﻭ ﺑﺮﺩﺍﺭ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻣﻌﻤﻮﻻ ﺗﻮﺍﺑﻊ ﺑﺎ ﻳﻚ ﺣﺮﻑ )ﻣﺎﻧﻨﺪ ‪ (f‬ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﻧﺪ‪ ،‬ﺍﻣﺎ ﺗﺎﺑﻊ ﻧﺮﻡ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ﻣﺘﻔﺎﻭﺕ ∥ ∥ ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﮔﺮ ‪ v‬ﻳﻚ‬
‫ﺑﺮﺩﺍﺭ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺮﻡ ﺍﻳﻦ ﺑﺮﺩﺍﺭ ﺭﺍ ﺑﺎ ∥‪ ∥v‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ ﺭﺍ ﺩﺭ ﻗﺎﻟﺐ ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺟﻤﻊﺑﻨﺪﻱ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﻧﺮﻡ ﺑﺮﺩﺍﺭﯼ‬ ‫ﺗﻌﺮﯾﻒ ‪۱−۴‬‬

‫ﻳﻚ ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ ﺗﺎﺑﻌﻲ ﺑﻪ ﺻﻮﺭﺕ ‪ ∥ ∥ : Rn → R‬ﺍﺳﺖ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺩﺍﺭﺍﻱ ﺧﻮﺍﺹ ﺯﻳﺮ ﺑﺎﺷﺪ‬

‫‪• ∀v ∈ Rn : ∥v∥ ≥ 0,‬‬

‫‪• ∥v∥ = 0 ⇐⇒ v = 0,‬‬

‫‪• ∀v ∈ Rn , ∀λ ∈ R: ∥λv∥ = |λ|∥v∥,‬‬

‫‪• ∀v, w ∈ Rn : ∥v + w∥ ≤ ∥v∥ + ∥w∥.‬‬

‫ﺗﻮﺍﺑﻊ ﺑﺴﻴﺎﺭﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﺗﻌﺮﻳﻒ ﻓﻮﻕ ﺻﺪﻕ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﻲﻧﻬﺎﻳﺖ ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ‬
‫ﺫﻛﺮ ﺷﺪ‪ ،‬ﻧﺮﻡ ﺑﺮﺩﺍﺭﻱ ﻳﻚ ﻋﺪﺩ ﺑﻪ ﺁﻥ ﺑﺮﺩﺍﺭ ﻣﺘﻨﺎﻇﺮ ﻣﻲﻛﻨﺪ ﻛﻪ ﻣﺒﻴﻦ ﺍﻧﺪﺍﺯﻩ‪ ،‬ﺍﺭﺯﺵ ﻭ ‪ ...‬ﺁﻥ ﺑﺮﺩﺍﺭ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﻧﺮﻡﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ‬
‫ﺑﻪ ﻃﺮﻕ ﻣﺘﻔﺎﻭﺗﻲ ﺍﻧﺪﺍﺯﻩ‪ ،‬ﺍﺭﺯﺵ ﻭ ‪ ...‬ﻳﻚ ﺑﺮﺩﺍﺭ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﻧﺮﻡﻫﺎﻱ ﭘﺮﻛﺎﺭﺑﺮﺩ ﺍﺷﺎﺭﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻧﺮﻡ ﺩﻭ ﺍﮔﺮ ⊤] ‪ v := [v1 , . . . , vn‬ﺁﻥﮔﺎﻩ ﻧﺮﻡ ﺩﻭ ‪ ۱۰‬ﻳﺎ ﻧﺮﻡ ﺍﻗﻠﻴﺪﺳﻲ ‪ ۱۱‬ﺑﺮﺩﺍﺭ ‪ v‬ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ ∥v∥2‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻭ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫‪q‬‬
‫=‪∥v∥2 :‬‬ ‫‪v12 + v22 + · · · + vn2 .‬‬ ‫)‪(۴.۴‬‬
‫‪9 Norm‬‬ ‫‪11 Euclidean‬‬ ‫‪norm‬‬
‫‪10 2-norm‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۶۸‬‬

‫ﻧﺮﻡ ﺑﻲﻧﻬﺎﻳﺖ ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻧﺮﻡ ﺑﻲﻧﻬﺎﻳﺖ ‪ ۱۲‬ﺑﺮﺩﺍﺭ ‪ ،v‬ﻛﻪ ﺑﺎ ﻧﻤﺎﺩ ∞∥‪ ∥v‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫‪∥v∥∞ := max {|v1 |, |v2 |, . . . , |vn |} .‬‬ ‫)‪(۵.۴‬‬
‫ﻭ ﻧﺮﻡ ﻣﺎﻛﺰﻳﻤﻢ ﻧﻴﺰ ﺷﻨﺎﺧﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻧﺮﻡ ‪ ۱‬ﻧﺮﻡ ﻳﻚ ﺑﺮﺩﺍﺭ ‪ ،v‬ﺑﺎ‬ ‫ﻳﻜﻨﻮﺍﺧﺖ ‪۱۴‬‬ ‫ﻧﺮﻡ‬ ‫ﭼﺒﻲﺷﻔﻲ ‪،۱۳‬‬ ‫ﻧﺮﻡ ﺑﻲﻧﻬﺎﻳﺖ ﺑﺎ ﻧﺎﻡﻫﺎﻱ ﻧﺮﻡ‬
‫ﻧﻤﺎﺩ ‪ ∥v∥1‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ‬
‫‪∥v∥1 := |v1 | + |v2 | + · · · + |vn |.‬‬ ‫)‪(۶.۴‬‬
‫ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﻫﺮ ﺳﻪ ﻧﺮﻡ ﻣﻌﺮﻓﻲ ﺷﺪﻩ‪ ،‬ﻛﻠﻴﻪ ﺧﻮﺍﺹ ﻣﻄﺮﺡ ﺷﺪﻩ ﺩﺭ ﺗﻌﺮﻳﻒ ‪ ۱-۴‬ﺭﺍ ﺩﺍﺭﺍ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬

‫ﺑﺮﺩﺍﺭ(‬ ‫‪ ،‬ﺩﻭ ﻭ ﺑﯽﻧﻬﺎﯾﺖ ﯾ‬ ‫ﻣﺜﺎﻝ ‪)۱−۴‬ﻧﺮﻡ ﯾ‬

‫‪h‬‬ ‫⊤‪i‬‬
‫‪ a := −1 3 2 −4‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬ ‫ﺑﺮﺩﺍﺭ‬
‫‪∥a∥1 = |−1| + |3| + |2| + |−4| = 10,‬‬
‫‪p‬‬ ‫√‬
‫‪∥a∥2 = (−1)2 + (3)2 + (2)2 + (−4)2 = 30,‬‬
‫‪∥a∥∞ = max{|−1|, |3|, |2|, |−4|} = 4.‬‬

‫ﺍﮔﺮ ﻧﺮﻡ ﺩﻭ ﺑﺮﺩﺍﺭ ﻣﺴﺎﻭﻱ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻟﺰﻭﻣﺎ ﺩﻭ ﺑﺮﺩﺍﺭ ﻳﻜﺴﺎﻥ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﺻﻮﻻ‪ ،‬ﺑﻲﻧﻬﺎﻳﺖ ﺑﺮﺩﺍﺭ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻧﺮﻡ ﺁﻥﻫﺎ‬
‫ﻳﻜﺴﺎﻥ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺍﮔﺮ ﻧﺮﻡ ﺩﻭ ﺑﺮﺩﺍﺭ ﺑﻪ ﻫﻢ ﻧﺰﺩﻳﻚ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻟﺰﻭﻣﺎ ﺧﻮﺩ ﺑﺮﺩﺍﺭﻫﺎ ﺑﻪ ﻫﻢ ﻧﺰﺩﻳﻚ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬
‫ﺑﺎ ﻣﻘﺎﻳﺴﻪ ﻧﺮﻡ ﺩﻭ ﺑﺮﺩﺍﺭ ﻧﻤﻲﺗﻮﺍﻥ ﻧﺰﺩﻳﻜﻲ ﺩﻭ ﺑﺮﺩﺍﺭ ﺑﻪ ﻳﻚﺩﻳﮕﺮ ﺭﺍ ﺗﺸﺨﻴﺺ ﺩﺍﺩ‪ .‬ﺍﻣﺎ ﻧﺮﻡ ﺗﻔﺎﺿﻞ ﺩﻭ ﺑﺮﺩﺍﺭ ﻣﻲﺗﻮﺍﻧﺪ ﻧﺰﺩﻳﻜﻲ ﺩﻭ‬
‫ﺑﺮﺩﺍﺭ ﺑﻪ ﻳﻚﺩﻳﮕﺮ ﺭﺍ ﻣﺸﺨﺺ ﻧﻤﺎﻳﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺍﮔﺮ ﺩﻭ ﺑﺮﺩﺍﺭ ﺑﻪ ﻫﻢ ﻧﺰﺩﻳﻚ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺮﻡ ﺗﻘﺎﺿﻞ ﺩﻭ ﺑﺮﺩﺍﺭ ﻋﺪﺩﻱ ﻛﻮﭼﻚ‬
‫ﺍﺳﺖ‪.‬‬

‫ﻣﺎﺗﺮﻳﺲ‬ ‫�‬

‫ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺟﺪﻭﻟﻲ ﺍﺯ ﺍﻋﺪﺍﺩ ﺍﺳﺖ‪ .‬ﻣﻌﻤﻮﻻ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺑﻪ ﺻﻮﺭﺕ ﻳﻚ ﺁﺭﺍﻳﻪ ﺍﺯ ﺍﻋﺪﺍﺩ ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﻧﺪ ﻛﻪ ﺩﺍﺧﻞ ﺑﺮﺍﻛﺖ ﻳﺎ‬
‫ﭘﺮﺍﻧﺘﺰ ﻗﺮﺍﺭ ﮔﺮﻓﺘﻪﺍﻧﺪ‪ .‬ﺩﺭ ﺯﻳﺮ ﻳﻚ ﻧﻤﻮﻧﻪ ﻣﺎﺗﺮﻳﺲ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪32‬‬ ‫‪−45‬‬ ‫‪23‬‬ ‫‪12‬‬ ‫‪32‬‬ ‫‪−45‬‬ ‫‪23‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪−7‬‬ ‫‪0‬‬ ‫‪45‬‬ ‫‪−42‬‬ ‫ﻳﺎ‬ ‫‪−7‬‬ ‫‪0‬‬ ‫‪45‬‬ ‫‪−42‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪13‬‬ ‫‪−4‬‬ ‫‪−9‬‬ ‫‪11‬‬ ‫‪13‬‬ ‫‪−4‬‬ ‫‪−9‬‬ ‫‪11‬‬
‫ﻳﺎ ﻋﻨﺎﺻﺮ ﻭ ﻳﺎ ﺩﺭﺍﻳﻪﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ‬ ‫ﻣﻮﻟﻔﻪﻫﺎﻱ ‪۱۵‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻓﻮﻕ ﺩﺍﺭﺍﻱ ﺳﻪ ﺳﻄﺮ ﻭ ﭼﻬﺎﺭ ﺳﺘﻮﻥ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻋﺪﺍﺩ ﺩﺍﺧﻞ ﻣﺎﺗﺮﻳﺲ ﺭﺍ‬
‫ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﻣﺎﺗﺮﻳﺲ ﻓﻮﻕ ﺩﺍﺭﺍﻱ ﺩﻭﺍﺯﺩﻩ ﻣﻮﻟﻔﻪ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺁﻥ ﻣﻮﻟﻔﻪﻫﺎﻳﻲ ﻛﻪ ﺷﻤﺎﺭﻩ ﺳﻄﺮ ﻭ ﺳﺘﻮﻥ ﺁﻥﻫﺎ ﻳﻜﺴﺎﻥ ﺍﺳﺖ ﺑﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ‬
‫ﻗﻄﺮﻱ ‪ ۱۶‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻣﻮﻗﻌﻴﺖ ﻛﻠﻴﻪ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ﺑﻪ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ‪ ۱۷‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﻣﺎﺗﺮﻳﺴﻲ ﻛﻪ ﺗﻌﺪﺍﺩ ﺳﻄﺮﻫﺎ ﻭ ﺳﺘﻮﻥﻫﺎﻱ ﺁﻥ ﻳﻜﺴﺎﻥ ﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ‪ ۱۸‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻣﺎﺗﺮﻳﺲﻫﺎﻳﻲ ﻛﻪ ﻣﺮﺑﻌﻲ‬
‫ﻧﻴﺴﺘﻨﺪ ﺭﺍ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻏﻴﺮﻣﺮﺑﻌﻲ ﻳﺎ ﻣﺴﺘﻄﻴﻠﻲ ﻣﻲﮔﻮﻳﻴﻢ‪.‬‬

‫‪12 Infinity‬‬
‫‪norm‬‬ ‫‪16 Diagonal‬‬ ‫‪elements‬‬
‫‪13 Chebyshev‬‬ ‫‪norm‬‬ ‫‪17 Matrix‬‬ ‫‪diagonal‬‬
‫‪14 Uniform norm‬‬ ‫‪18 Square matrix‬‬
‫)‪15 Elements (Components) (Entries) (Coefficients‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۶۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫■ ﻧﻤﺎﺩﮔﺬﺍﺭﻱ ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﻧﺤﻮﻩ ﻧﻤﺎﺩﮔﺬﺍﺭﻱ ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻭ ﻧﻤﺎﺩﻫﺎﻱ ﻣﻮﺭﺩ ﺍﺳﺘﻔﺎﺩﻩ ﺑﺮﺍﻱ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺧﺎﺹ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫• ﺑﺮﺍﻱ ﻧﺎﻡﮔﺬﺍﺭﻱ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺍﺯ ﺣﺮﻭﻑ ﺑﺰﺭﮒ ﻭ ﺑﻮﻟﺪ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ ﺣﺮﻑ ﻛﻮﭼﻚ ﻣﺘﻨﺎﻇﺮ ﺑﻪ‬
‫ﻫﻤﺮﺍﻩ ﺩﻭ ﺍﻧﺪﻳﺲ ﺳﻄﺮ ﻭ ﺳﺘﻮﻥ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺩﺭ ﺯﻳﺮ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﭼﻬﺎﺭ ﺳﻄﺮﻱ ﻭ ﭼﻬﺎﺭ ﺳﺘﻮﻧﻲ‬
‫ﺑﻪ ﻫﻤﺮﺍﻩ ﻣﻮﻟﻔﻪﻫﺎﻳﺶ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‬
‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫‪a13‬‬ ‫‪a14‬‬
‫‪‬‬ ‫‪‬‬
‫‪a21‬‬ ‫‪a22‬‬ ‫‪a23‬‬ ‫‪‬‬
‫‪a24 ‬‬
‫‪A=‬‬
‫‪a‬‬ ‫‪‬‬
‫‪ 31‬‬ ‫‪a32‬‬ ‫‪a33‬‬ ‫‪a34 ‬‬
‫‪a41‬‬ ‫‪a42‬‬ ‫‪a43‬‬ ‫‪a44‬‬

‫• ﻳﻚ ﻣﺎﺗﺮﻳﺲ ‪ n‬ﺳﻄﺮﻱ ﻭ ‪ m‬ﺳﺘﻮﻧﻲ‪ ،‬ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﺩﺍﺭﺍﻱ ﻓﺮﻡ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫‪...‬‬ ‫‪a1m‬‬
‫‪‬‬ ‫‪‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2m ‬‬
‫‪‬‬ ‫‪‬‬
‫‪A= .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫)‪(۷.۴‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫‪...‬‬ ‫‪anm‬‬
‫ﺑﺮﺍﻱ ﺳﺎﺩﮔﻲ ﻭ ﺧﻼﺻﻪ ﻧﻮﻳﺴﻲ ﺩﺭ ﺗﻌﺮﻳﻒ ﻓﻮﻕ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻣﻌﺮﻓﻲ ﻛﺮﺩ‬
‫‪h i‬‬
‫‪A = aij i=1,...,n,‬‬
‫‪j=1,...,m.‬‬

‫• ﻣﺠﻤﻮﻋﻪ ﺗﻤﺎﻣﻲ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ‪ n‬ﺳﻄﺮﻱ ﻭ ‪ m‬ﺳﺘﻮﻧﻲ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ Rn×m‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫• ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺳﻄﺮ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ a′i‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻭ ﺳﺘﻮﻥ ‪j‬ﺍﻡ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ‬
‫‪ aj‬ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫■ ﺗﺮﺍﻧﻬﺎﺩﻩ ﻣﺎﺗﺮﻳﺲ‬
‫ﻋﻤﻞ ﺗﺮﺍﻧﻬﺎﺩﻩ ‪ ۱۹‬ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺣﻮﻝ ﻗﻄﺮ ﺍﺻﻠﻲﺍﺵ ﺑﭽﺮﺧﺪ‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪3 1 7 0‬‬ ‫‪ 3 2‬‬ ‫‪1 ‬‬
‫‪‬‬ ‫‪‬‬ ‫ﺗﺮﺍﻧﻬﺎﺩﻩ‬ ‫‪ 1 4‬‬ ‫‪1 ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 2 4 6 5‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ 7 6‬‬ ‫‪3 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪1 1 3 2‬‬
‫‪0 5‬‬ ‫‪2‬‬

‫ﺑﺎ ﺗﺮﺍﻧﻬﺎﺩﻩ ﻛﺮﺩﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﺳﻄﺮ ‪i‬ﺍﻡ ﺩﺭ ﺳﺘﻮﻥ ‪i‬ﺍﻡ ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪ .‬ﺗﺮﺍﻧﻬﺎﺩﻩ ﻣﺎﺗﺮﻳﺲ ‪ A ∈ Rn×m‬ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ⊤‪ A‬ﻧﺸﺎﻥ‬
‫ﻣﻲﺩﻫﻴﻢ ﻭ ‪ .A⊤ ∈ Rm×n‬ﺍﻟﺒﺘﻪ ﺍﺯ ﻧﻤﺎﺩﻫﺎﻱ ‪ At ،A′‬ﻭ ‪ Atr‬ﻧﻴﺰ ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ ﺑﻪ ﺗﺮﺍﻧﻬﺎﺩﻩ ﻣﺎﺗﺮﻳﺲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻃﻮﺭ‬
‫‪19 Transpose‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۷۰‬‬

‫ﺧﻼﺻﻪ‪ ،‬ﺍﮔﺮ ﻣﺎﺗﺮﻳﺲ ‪A‬ﺑﻪ ﺻﻮﺭﺕ )‪ (۷.۴‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬

‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a21‬‬ ‫‪...‬‬ ‫‪am1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ a12‬‬ ‫‪a22‬‬ ‫···‬ ‫‪am2 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪A⊤ =  .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬
‫‪a1n‬‬ ‫‪a2n‬‬ ‫‪...‬‬ ‫‪amn‬‬

‫■ ﻋﻤﻠﻴﺎﺕ ﺣﺴﺎﺑﻲ ﺭﻭﻱ ﻣﺎﺗﺮﻳﺲﻫﺎ‬

‫ﻭﻗﺘﻲ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻫﻢﺑﻌﺪ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﺻﻮﺭﺕ ﻣﻮﻟﻔﻪﺍﻱ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺍﮔﺮ‬
‫‪h i‬‬ ‫‪h‬‬ ‫‪i‬‬
‫‪A = aij i=1,...,n‬‬ ‫‪ B = bij‬ﻭ‬ ‫‪i=1,...,n‬‬
‫‪j=1,...,m‬‬ ‫‪j=1,...,m‬‬
‫ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﺩﺭ ‪ Rn×m‬ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪h‬‬ ‫‪i‬‬ ‫‪h‬‬ ‫‪i‬‬
‫‪A + B = aij + bij‬‬ ‫‪i=1,...,n‬‬ ‫‪A − B = aij − bij‬‬ ‫‪i=1,...,n‬‬
‫‪j=1,...,m‬‬ ‫‪j=1,...,m‬‬
‫ﺑﺮﺍﻱ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻫﻢﺑﻌﺪ ﻓﻮﻕ‪ ،‬ﺿﺮﺏ ﻣﻮﻟﻔﻪﺍﻱ ﻳﺎ ﺿﺮﺏ ﻫﺎﺩﺍﻣﺎﺭﺩﻱ ﻧﻴﺰ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻗﺎﺑﻞ ﺗﻌﺮﻳﻒ ﺍﺳﺖ‬
‫‪h‬‬ ‫‪i‬‬
‫‪A ⊙ B = aij bij i=1,...,n‬‬
‫‪j=1,...,m‬‬

‫ﻧﻴﺰ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﻻﺯﻣﻪ ﺍﻧﺠﺎﻡ ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ﺁﻥ ﺍﺳﺖ ﻛﻪ‬ ‫ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ‪۲۰‬‬
‫ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ‪ ،‬ﻋﻼﻭﻩ ﺑﺮ ﺿﺮﺏ ﻣﻮﻟﻔﻪﺍﻱ‪،‬‬
‫‪B∈R‬‬ ‫‪k×m‬‬
‫‪A∈R‬ﻭ‬ ‫‪n×k‬‬
‫ﺗﻌﺪﺍﺩ ﺳﺘﻮﻥﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﺍﻭﻝ ﺑﺮﺍﺑﺮ ﺗﻌﺪﺍﺩ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺩﻭﻡ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺍﮔﺮ‬
‫‪R‬‬ ‫‪n×m‬‬
‫ﺁﻥﮔﺎﻩ ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺩﺭ ﻣﺎﺗﺮﻳﺲ ‪ B‬ﻗﺎﺑﻞ ﺗﻌﺮﻳﻒ ﺍﺳﺖ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺣﺎﺻﻞ ﺍﻳﻦ ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺭ‬
‫ﺍﺳﺖ ﻛﻪ ﻣﻮﻟﻔﻪﻱ )‪(i, j‬ﺍﻡ ﺁﻥ ﺍﺯ ﺿﺮﺏ ﺩﺍﺧﻠﻲ ﺳﻄﺮ ‪i‬ﺍﻡ ‪ A‬ﺩﺭ ﺳﺘﻮﻥ ‪j‬ﺍﻡ ‪ B‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ ﻣﻲﺗﻮﺍﻥ‬
‫ﻧﻮﺷﺖ‬
‫‪h‬‬ ‫‪i‬‬ ‫‪hP‬‬ ‫‪i‬‬
‫= ‪AB = a′i · bj i=1,...,n‬‬ ‫‪k‬‬
‫‪s=1 ais bsj i=1,...,n‬‬
‫‪j=1,...,m‬‬ ‫‪j=1,...,m‬‬
‫ﺩﺭ ﺭﺍﺑﻄﻪ ﻓﻮﻕ‪ ،‬ﻣﻨﻈﻮﺭ ﺍﺯ ‪ a′i‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺳﻄﺮ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻭ ﻭ ‪ bj‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺳﺘﻮﻥ ‪j‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪.B‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ﻳﻚ ﻋﻤﻞ ﺟﺎﺑﺠﺎﻳﻲﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ‪.AB ̸= BA‬‬
‫ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ‪ ،‬ﻋﻀﻮ ﺧﻨﺜﻲ ﻋﻤﻞ ﺿﺮﺏ ﻣﺎﺗﺮﻳﺴﻲ ﺍﺳﺖ ﻛﻪ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱﺍﺵ ﺑﺮﺍﺑﺮ ‪ 1‬ﻭ ﺑﻘﻴﻪ ﻋﻨﺎﺻﺮﺵ ﺑﺮﺍﺑﺮ ‪ 0‬ﺍﺳﺖ‪ .‬ﺍﻳﻦ‬
‫ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ‪ ۲۱‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ ﻭ ﺑﺎ ﻧﻤﺎﺩ ‪ In‬ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ n‬ﺑﻌﺪ ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﺣﺎﺻﻞ‬
‫ﺿﺮﺏ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ )ﺍﺯ ﭼﭗ ﻳﺎ ﺭﺍﺳﺖ( ﺩﺭ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺑﺮﺍﺑﺮ ﺧﻮﺩ ‪ A‬ﺍﺳﺖ )ﺍﻟﺒﺘﻪ ﻣﺸﺮﻭﻁ ﺑﺮ ﺍﻳﻦ ﻛﻪ ﺍﻳﻦ ﺿﺮﺏ ﻣﻤﻜﻦ‬
‫ﺑﺎﺷﺪ(‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ﻣﺎﺗﺮﻳﺲﻫﺎ ﻫﻤﺎﻥ ﻧﻘﺶ ﻋﺪﺩ ﻳﻚ ﺩﺭ ﻣﺠﻤﻮﻋﻪ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺭﺍ ﺑﺎﺯﻱ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ‬
‫ﺯﻳﺮ ﻣﺎﺗﺮﻳﺲ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻫﻤﺎﻧﻲ ﺩﻭ‪ ،‬ﺳﻪ ﻭ ‪-n‬ﺗﺎﻳﻲ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪1 0‬‬ ‫‪...‬‬ ‫‪0‬‬
‫"‬ ‫‪#‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪0 1‬‬ ‫‪...‬‬ ‫‪0‬‬
‫‪1‬‬ ‫‪0‬‬ ‫‪‬‬ ‫‪‬‬
‫= ‪I2‬‬ ‫‪,‬‬ ‫‪I3 = ‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬
‫‪,‬‬ ‫‪In =  . .‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪ .. ..‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫‪...‬‬ ‫‪1‬‬

‫‪20 Matrix‬‬ ‫‪product‬‬ ‫‪21 Identity‬‬ ‫‪matrix‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫■ ﺗﻘﺴﻴﻢ ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ ﻭ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ‬

‫ﺩﺭ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ‪ ،‬ﺗﻘﺴﻴﻢ ‪ a/b‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺿﺮﺏ ‪ a‬ﺩﺭ ﻣﻌﻜﻮﺱ ‪ ،b‬ﻳﻌﻨﻲ ‪ a/b = a ∗ b−1‬ﻛﻪ ‪ b−1‬ﻭﺍﺭﻭﻥ ‪ b‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﭼﻨﻴﻦ ﺗﻌﺒﻴﺮﻱ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﻣﺎﺗﺮﻳﺲﻫﺎ ﻧﻴﺰ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻭﺍﺭﻭﻥ ‪ ۲۲‬ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺗﻌﺮﻳﻒ ﻧﻤﻮﺩ‪.‬‬

‫ﻣﺎﺗﺮﯾﺲ‬ ‫ﻭﺍﺭﻭﻥ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۴‬‬

‫ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ﺩﺭ ‪ Rn×n‬ﺑﺎﺷﺪ ﻭ ﺍﮔﺮ ﻣﺎﺗﺮﻳﺲ ‪ B ∈ Rn×n‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ ﻛﻪ‬
‫‪AB = BA = In‬‬
‫ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ‪ B‬ﺭﺍ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ )ﻳﺎ ﻣﻌﻜﻮﺱ( ‪ A‬ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ A−1‬ﻭ ﻳﺎ )‪ inv(A‬ﻧﺸﺎﻥ‬
‫ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﻌﻜﻮﺱ ﻫﺮ ﻣﺎﺗﺮﻳﺴﻲ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﻣﻌﻜﻮﺱ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻌﻜﻮﺱ ﻳﻜﺘﺎ‬
‫ﻭ ﻳﺎ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻣﺎﺗﺮﻳﺴﻲ ﻛﻪ ﻣﻌﻜﻮﺱ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥﻧﺎﭘﺬﻳﺮ ‪ ۲۳‬ﻭ ﻳﺎ ﻣﻨﻔﺮﺩ ‪ ۲۴‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ‬
‫ﺑﺮﺧﻲ ﻣﻨﺎﺑﻊ ﺍﺯ ﻭﺍﮊﻩ ﺗﻜﻴﻦ ﺑﻪ ﺟﺎﻱ ﻣﻨﻔﺮﺩ ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﻘﺎﺑﻞ‪ ،‬ﻣﺎﺗﺮﻳﺲﻫﺎﻳﻲ ﻛﻪ ﻣﻌﻜﻮﺱ ﺁﻥﻫﺎ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ ﺑﻪ‬
‫ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﻳﺎ ﻧﺎﻣﻨﻔﺮﺩ ﻳﺎ ﻧﺎﺗﻜﻴﻦ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬
‫ﺣﺎﻝ ﺑﻪ ﺗﻘﺴﻴﻢ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺑﺮﻣﻲﮔﺮﺩﻳﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺗﻘﺴﻴﻢ ﻣﺎﺗﺮﻳﺴﻲ ﭼﻨﺪﺍﻥ ﺭﺍﻳﺞ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺑﻴﺸﺘﺮ ﺍﺯ ﻭﺍﮊﻩ ”ﺿﺮﺏ ﺩﺭ ﻣﻌﻜﻮﺱ“‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ ﺗﻘﺴﻴﻢ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ a‬ﺑﺮ ﻋﺪﺩ ‪ b‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ‪ ab−1‬ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﻪ‬
‫ﻧﻈﺮ ﻣﻲﺁﻳﺪ ﻛﻪ ﺗﻘﺴﻴﻢ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺑﺮ ﻣﺎﺗﺮﻳﺲ ‪ B‬ﺭﺍ ﻧﻴﺰ ﺑﻪ ﺻﻮﺭﺕ ‪ AB−1‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺍﻣﺎ ﺗﻮﺟﻪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﺪ ﻛﻪ ﺩﺭ‬
‫ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ ab−1‬ﻭ ‪ b−1 a‬ﺑﺎ ﻳﻜﺪﻳﮕﺮ ﻣﻌﺎﺩﻝﺍﻧﺪ ﻭ ﻫﺮﻛﺪﺍﻡ ﺍﺯ ﺍﻳﻦ ﻋﺒﺎﺭﺍﺕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ‪ a/b‬ﺩﺭﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺍﻣﺎ‬
‫ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ‪ ،‬ﻟﺰﻭﻣﺎ ‪ AB−1‬ﻭ ‪ B−1 A‬ﻣﻌﺎﺩﻝ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﺩﺭ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺩﻭ ﻧﻮﻉ ﺗﻘﺴﻴﻢ ﻣﻮﺳﻮﻡ ﺑﻪ ﺗﻘﺴﻴﻢ ﺍﺯ‬
‫ﺭﺍﺳﺖ ‪ ۲۵‬ﻭ ﺗﻘﺴﻴﻢ ﺍﺯ ﭼﭗ ‪ ۲۶‬ﺗﻌﺮﻳﻒ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﺮﺍﻱ ﺗﻘﺴﻴﻢ ﺍﺯ ﺭﺍﺳﺖ ﺍﺯ ﻧﻤﺎﺩ ”‪ “/‬ﻭ ﺑﺮﺍﻱ ﺗﻘﺴﻴﻢ ﺍﺯ ﭼﭗ ﺍﺯ ﻧﻤﺎﺩ ”\“ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﻧﺪ‬

‫‪A/B := AB−1‬‬
‫‪A\B := A−1 B‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ‪ B‬ﻣﺎﺗﺮﻳﺴﻲ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻄﺎﺑﻖ ﺑﺎ ﺗﻌﺮﻳﻒ ﻓﻮﻕ‪ ،‬ﺗﻘﺴﻴﻢ ‪ A/B‬ﺍﻣﻜﺎﻥﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫■ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ‬
‫ﺩﺭ ﺑﺨﺶ ﻗﺒﻞ‪ ،‬ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺗﻌﺮﻳﻒ ﺷﺪ ﻭ ﺍﺷﺎﺭﻩ ﺷﺪ ﻛﻪ ﺑﺮﺍﻱ ﻫﻤﻪ ﻣﺎﺗﺮﻳﺲﻫﺎ ﻭﺍﺭﻭﻥ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ‬
‫ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﭼﮕﻮﻧﻪ ﺗﺸﺨﻴﺺ ﺩﻫﻴﻢ ﻛﻪ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﺩﺍﺭﺩ ﻳﺎ ﺧﻴﺮ؟ ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻣﻔﻬﻮﻡ ﺩﺗﺮﻣﻴﻨﺎﻥ ‪ ۲۷‬ﻣﻲﺗﻮﺍﻧﺪ ﻛﻤﻚ‬
‫ﻛﻨﻨﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﺑﺮﺍﻱ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﺮﺑﻌﻲ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ ﻭ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺗﺮﻣﻴﻨﺎﻥ‬
‫ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ )‪ det(A‬ﻭ ﻳﺎ ﻧﻤﺎﺩ |‪ |A‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺩﺭ ﺯﻳﺮ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬

‫‪22 Inverse‬‬ ‫‪25 Right‬‬‫‪division‬‬

‫‪23 Non-invertible‬‬ ‫‪matrix‬‬ ‫‪26 Left‬‬‫‪division‬‬
‫‪24 Singular matrix‬‬ ‫‪27 Determinant‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۷۲‬‬

‫ﺩﺗﺮﻣﯿﻨﺎﻥ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۴‬‬

‫ﺩﺗﺮﻣﻴﻨﺎﻥ ﺗﺎﺑﻌﻲ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ﺭﺍ ﻣﻲﮔﻴﺮﺩ ﻭ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺭﺍ ﭘﺲ ﻣﻲﺩﻫﺪ‪ .‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﮔﺮﺩﺩ‬

‫• ﺍﮔﺮ ] ‪ A = [a11‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺩﺭ ‪ R1×1‬ﺑﺎﺷﺪ )ﻳﻌﻨﻲ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻳﻚ ﺩﺭ ﻳﻚ ﺑﺎﺷﺪ(‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺗﺮﻣﻴﻨﺎﻥ ‪ A‬ﺑﺮﺍﺑﺮ‬
‫ﺗﻨﻬﺎ ﻣﻮﻟﻔﻪ ﺁﻥ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ‪.det(A) = a11‬‬
‫"‬ ‫‪#‬‬
‫‪a11‬‬ ‫‪a12‬‬
‫= ‪ A‬ﺁﻥﮔﺎﻩ ‪.det(A) = a11 a22 − a21 a12‬‬ ‫• ﺍﮔﺮ‬
‫‪a21‬‬ ‫‪a22‬‬

‫• ﺍﮔﺮ ‪ A ∈ Rn×n‬ﺑﺎﺷﺪ )‪ ،(n > 1‬ﺁﻥﮔﺎﻩ‬

‫) ‪det(A) = (−1)k+1 ak1 det(Bk1 )+(−1)k+2 ak2 det(Bk2 )+· · ·+(−1)k+n akn det(Bkn‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ k‬ﻋﺪﺩﻱ ﺩﻟﺨﻮﺍﻩ ﺩﺭ ﻣﺠﻤﻮﻋﻪ }‪ {1, 2, . . . , n‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺳﻄﺮ ‪ k‬ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ ﻭ ‪Bkj‬‬
‫ﻣﺎﺗﺮﻳﺴﻲ ﺩﺭ )‪ R(n−1)×(n−1‬ﺍﺳﺖ ﻛﻪ ﺍﺯ ﺣﺬﻑ ﺳﻄﺮ ‪k‬ﺍﻡ ﻭ ﺳﺘﻮﻥ ‪j‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭ ﺗﻌﺮﻳﻒ ﻓﻮﻕ ﻳﻚ ﺭﺍﺑﻄﻪ ﺑﺎﺯﮔﺸﺘﻲ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ‬
‫ﺻﻮﺭﺕ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ‪ n × n‬ﺑﺮﺣﺴﺐ ﺩﺗﺮﻣﻴﻨﺎﻥ ‪ n‬ﻣﺎﺗﺮﻳﺲ ﺩﻳﮕﺮ ﺑﺎ ﺍﺑﻌﺎﺩ )‪ (n − 1) × (n − 1‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺎﺗﺮﯾﺲ ﺑﺎ ﺑﻌﺪ ‪(۴‬‬ ‫ﻣﺜﺎﻝ ‪)۲−۴‬ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﯿﻨﺎﻥ ﯾ‬

‫ﻣﺎﺗﺮﻳﺲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪3‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬
‫‪4‬‬
‫‪A := ‬‬
‫‪−1‬‬
‫‪‬‬
‫‪‬‬ ‫‪0 1 0‬‬‫‪‬‬
‫‪2‬‬ ‫‪0 4 1‬‬
‫ﻃﺒﻖ ﺗﻌﺮﻳﻒ ﻭ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺳﻄﺮ ﺍﻭﻝ‪ ،‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫‪0‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪det(A) = 1 det ‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪ − 4 det −1‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪ + 2 det −1 0‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪ − 3 det −1 0‬‬ ‫‪1‬‬
‫‪‬‬
‫‪0‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪0‬‬ ‫‪4‬‬
‫ﺣﺎﻝ ﺑﺎﻳﺪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﭼﻬﺎﺭ ﻣﺎﺗﺮﻳﺲ ﺳﻪ ﺩﺭ ﺳﻪ ﻓﻮﻕ ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﺑﺎ ﻛﻤﻚ ﺗﻌﺮﻳﻒ ﻣﻲﺗﻮﺍﻥ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺍﻳﻦ ﭼﻬﺎﺭ ﻣﺎﺗﺮﻳﺲ ﺭﺍ‬
‫ﻧﻴﺰ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻛﺮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬
‫‪‬‬ ‫‪‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪0 0‬‬ ‫‪0 1‬‬
‫‪det ‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬‫‪ = 1 det 4‬‬ ‫‪− 4 det‬‬ ‫‪+ 1 det‬‬
‫‪1‬‬ ‫‪0 1‬‬ ‫‪0 4‬‬
‫‪0‬‬ ‫‪4‬‬ ‫‪1‬‬

‫‪= 1 × 1 − 4 × 0 + 4 × 0 = 1,‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬
‫‪‬‬ ‫‪‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪−1 0‬‬ ‫‪−1 1‬‬
‫‪det ‬‬ ‫‪‬‬
‫‪−1 1 0 = 0 det 4‬‬ ‫‪− 4 det‬‬ ‫‪+ 4 det‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪4‬‬
‫‪2 4 1‬‬

‫‪= 0 × 1 − 4 × (−1) + 4 × (−6) = −20,‬‬

‫‪‬‬ ‫‪‬‬
‫‪0 1 4‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬
‫‪‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−1 0‬‬ ‫‪−1 0‬‬
‫‪det ‬‬ ‫‪‬‬
‫‪−1 0 0 = 0 det 0‬‬ ‫‪− 1 det‬‬ ‫‪+ 4 det‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪0‬‬
‫‪2 0 1‬‬

‫‪= 0 × 0 − 1 × (−1) + 4 × (0) = 1,‬‬

‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬
‫‪‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪0‬‬
‫‪det ‬‬ ‫‪‬‬
‫‪−1 0 1 = 0 det 0 4 − 1 det 2 4 + 4 det 2 0‬‬
‫‪2 0 4‬‬

‫‪= 0 × 0 − 1 × (−6) + 4 × (0) = 6.‬‬

‫ﺣﺎﻝ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻫﺮ ﭼﻬﺎﺭ ﻣﺎﺗﺮﻳﺲ ‪ 3 × 3‬ﻣﺤﺎﺳﺒﻪ ﺷﺪ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫‪det(A) = 1 × 1 − 4 × (−20) + 2 × 1 − 3 × 6‬‬
‫‪= 65‬‬
‫ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺷﻤﺎﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪11 12 13‬‬ ‫‪14‬‬
‫‪21 22 23‬‬ ‫‪24‬‬
‫‪31 32 33‬‬ ‫‪34‬‬
‫‪41 42 43‬‬ ‫‪44‬‬

‫‪22 23 24‬‬ ‫‪21 23 24‬‬ ‫‪21 22 24‬‬ ‫‪21 22 23‬‬

‫‪32 33 34‬‬ ‫‪31 33 34‬‬ ‫‪31 32 34‬‬ ‫‪31 32 33‬‬
‫‪42 43 44‬‬ ‫‪41 43 44‬‬ ‫‪41 42 44‬‬ ‫‪41 42 43‬‬

‫‪33 34 32 34 32 33 33 34 31 34 31 33 32 34 31 34 31 32 32 33 31 33 31 32‬‬
‫‪43 44 42 44 42 43 43 44 41 44 41 43 42 44 41 44 41 42 42 43 41 43 41 42‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ‪ 4 × 4‬ﻧﻴﺎﺯ ﺑﻪ ﻣﺤﺎﺳﺒﻪ ‪ 4 × 3 = 12‬ﺗﻌﺪﺍﺩ ﺩﺗﺮﻣﻴﻨﺎﻥ‬
‫ﻣﺎﺗﺮﻳﺲ ‪ 2 × 2‬ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪.‬‬

‫ﻣﻄﺎﺑﻖ ﺑﺎ ﺗﻌﺮﻳﻒ ‪ ۳-۴‬ﻭ ﻫﻤﺎﻧﻨﺪ ﻣﺜﺎﻝ ‪ ،۲-۴‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ‪ ،n×n‬ﻻﺯﻡ ﺍﺳﺖ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ‬
‫ﺑﺎ ﺍﺑﻌﺎﺩ ﻛﻮﭼﻚﺗﺮ ﺑﻪ ﺻﻮﺭﺕ ﺳﻠﺴﻠﻪ ﻣﺮﺍﺗﺒﻲ ﻭ ﺩﺭﺧﺘﻲ ﻣﺤﺎﺳﺒﻪ ﺷﻮﻧﺪ ﻛﻪ ﺩﺭ ﭘﺎﻳﻴﻦﺗﺮﻳﻦ ﺳﻄﺢ ﺍﻳﻦ ﺩﺭﺧﺖ ‪ n!2‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ‬
‫‪ 2 × 2‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ‪ n‬ﺑﺰﺭﮒ ﺑﺎﺷﺪ‪ ،‬ﺍﻧﺪﺍﺯﻩ ﺍﻳﻦ ﺩﺭﺧﺖ ﺑﺴﻴﺎﺭ ﺑﺰﺭﮒ ﺍﺳﺖ ﻭ ﺗﻌﺪﺍﺩ ﺑﺴﻴﺎﺭ ﺯﻳﺎﺩﻱ ﻣﺎﺗﺮﻳﺲ ‪ 2 × 2‬ﺩﺭ‬
‫ﭘﺎﻳﻴﻦﺗﺮﻳﻦ ﺳﻄﺢ ﺩﺭﺧﺖ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۷۴‬‬

‫!‪ 20‬ﺩﺗﺮﻣﻴﻨﺎﻥ ‪ 2 × 2‬ﺩﺭ ﭘﺎﻳﻴﻦﺗﺮﻳﻦ‬ ‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺍﮔﺮ ‪ ،n = 20‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺗﻌﺪﺍﺩ ‪2 = 1216451004088320000‬‬
‫ﺳﻄﺢ ﺩﺭﺧﺖ ﺑﺎﻳﺪ ﻣﺤﺎﺳﺒﻪ ﮔﺮﺩﺩ ﻛﻪ ﻣﺴﺘﻠﺰﻡ ﺣﺎﻓﻈﻪ ﻭ ﺯﻣﺎﻥ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺑﺴﻴﺎﺭ ﺯﻳﺎﺩﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ ﺯﻣﺎﻥ ﻻﺯﻡ‪،‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺯﻣﺎﻥ ﻻﺯﻡ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺩﻭ ﺩﺭ ﺩﻭ ﺑﺮﺍﺑﺮ ‪ 10−12‬ﺛﺎﻧﻴﻪ )ﻳﻚ ﺗﺮﻳﻠﻴﺎﺭﺩﻡ ﺛﺎﻧﻴﻪ( ﻃﻮﻝ ﺑﻜﺸﺪ‪.‬‬
‫!‪ 20‬ﺛﺎﻧﻴﻪ ﺯﻣﺎﻥ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻣﺎ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪2 × 10‬‬
‫‪−12‬‬
‫ﺣﺎﻝ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺑﻌﺪ ‪ 20‬ﺑﻪ ﺑﻴﺶ ﺍﺯ‬
‫!‪20‬‬
‫ﺛﺎﻧﻴﻪ ‪× 10−12 = 1, 216, 451.004‬‬
‫‪2‬‬
‫ﺳﺎﻋﺖ ‪= 337.903056691‬‬
‫ﺭﻭﺯ ‪= 14.0792940288‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ‪ 20 × 20‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻌﺮﻳﻒ ﺍﻭﻟﻴﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺭﻭﻱ ﻳﻚ ﻛﺎﻣﭙﻴﻮﺗﺮ ﺳﺮﻳﻊ ﺑﻪ ﺑﻴﺶ‬
‫ﺍﺯ ‪ ۱۴‬ﺭﻭﺯ ﺯﻣﺎﻥ ﻧﻴﺎﺯ ﺩﺍﺭﺩ‪.‬‬
‫ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ‪ ،‬ﺗﻌﺮﻳﻒ ﻓﻮﻕ ﻳﻚ ﺗﻌﺮﻳﻒ ﻛﻼﺳﻴﻚ ﻭ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺍﺳﺖ ﻛﻪ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﮔﻔﺘﻪ ﺷﺪ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺁﻥ ﺩﺭ ﻋﻤﻞ ﺑﻪ ﻫﻴﭻ ﻭﺟﻪ ﻣﻨﺎﺳﺐ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺭﻭﺵﻫﺎﻱ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﺯ ﺟﻤﻠﻪ‬
‫ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪.‬‬
‫ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻣﻮﺭﺩ ﺧﻮﺍﺹ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺩﺍﺭﻳﻢ‪.‬‬

‫ﺑﺮﺧ ﺧﻮﺍﺹ ﺩﺗﺮﻣﯿﻨﺎﻥ‬ ‫ﻗﻀﯿﻪ ‪۴−۴‬‬

‫• ﺟﺎﺑﺠﺎﻳﻲ ﺩﻭ ﺳﻄﺮ )ﺳﺘﻮﻥ( ﻣﺎﺗﺮﻳﺲ ﺑﺎﻋﺚ ﻗﺮﻳﻨﻪ ﺷﺪﻥ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺍﮔﺮ ﺩﻭ ﺳﻄﺮ )ﻳﺎ ﺩﻭ‬
‫ﺳﺘﻮﻥ( ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﻋﻮﺽ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺣﺎﺻﻞ ﺑﺮﺍﺑﺮ ﻗﺮﻳﻨﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺍﻭﻟﻴﻪ ﺍﺳﺖ‪.‬‬

‫• ﺟﻤﻊ ﻣﻀﺮﺑﻲ ﺍﺯ ﻳﻚ ﺳﻄﺮ ﺑﺎ ﺳﻄﺮ ﺩﻳﮕﺮ ﺑﺎﻋﺖ ﺗﻐﻴﻴﺮ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ﻣﻀﺮﺑﻲ ﺍﺯ ﻳﻚ ﺳﻄﺮ‬
‫ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ ﺳﻄﺮ ﺩﻳﮕﺮﻱ ﺍﺯ ﺁﻥ ﻣﺎﺗﺮﻳﺲ ﺟﻤﻊ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺁﻥ ﺑﺮﺍﺑﺮ ﺩﺗﺮﻣﻴﻨﺎﻥ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﻭﻟﻴﻪ ﺍﺳﺖ‪.‬‬

‫• ﺍﮔﺮ ﻳﻚ ﺳﻄﺮ )ﺳﺘﻮﻥ( ﻣﺎﺗﺮﻳﺴﻲ ﺭﺍ ﺩﺭ ‪ λ‬ﺿﺮﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺁﻥ ‪ λ‬ﺑﺮﺍﺑﺮ ﻣﻲﺷﻮﺩ‪.‬‬

‫• ﺍﮔﺮ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺩﺭ ﻋﺪﺩ ‪ λ‬ﺿﺮﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺁﻥ ‪ λn‬ﺑﺮﺍﺑﺮ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ n‬ﺑﻌﺪ ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ‪.‬‬

‫• ﺗﺮﺍﻧﻬﺎﺩﻩ ﻛﺮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻋﺚ ﺗﻐﻴﻴﺮ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪det(A⊤ ) = det(A).‬‬

‫• ﺍﮔﺮ ﻣﺎﺗﺮﻳﺴﻲ ﺭﺍ ﻭﺍﺭﻭﻥ ﻛﻨﻴﻢ‪ ،‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﺁﻥ ﻣﻌﻜﻮﺱ ﻣﻲﺷﻮﺩ‪ .‬ﻳﻌﻨﻲ‬

‫
‬ ‫‪1‬‬
‫= ‪det A−1‬‬ ‫‪.‬‬
‫)‪det(A‬‬
‫• ﺩﺗﺮﻣﻴﻨﺎﻥ ﺣﺎﺻﻞﺿﺮﺏ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﺑﺮﺍﺑﺮ ﺣﺎﺻﻞﺿﺮﺏ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ‪.‬‬
‫‪det(AB) = det(A) det(B).‬‬

‫ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﻘﺪﺍﺭ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﺸﺨﺺ ﻛﺮﺩ ﻛﻪ ﺁﻳﺎ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻜﻮﺱﭘﺬﻳﺮ ﺍﺳﺖ ﻳﺎ ﺧﻴﺮ‪ .‬ﻗﻀﻴﻪ ﺯﻳﺮ‬
‫ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺩﺗﺮﻣﯿﻨﺎﻥ ﻭ ﻭﺍﺭﻭﻥﭘﺬﯾﺮﯼ‬ ‫ﻗﻀﯿﻪ ‪۵−۴‬‬

‫ﻣﺎﺗﺮﻳﺲ ‪ ،A ∈ Rn×n‬ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﺍﺳﺖ‪ ،‬ﺍﮔﺮ ﻭ ﺗﻨﻬﺎ ﺍﮔﺮ ‪.det(A) ̸= 0‬‬

‫ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺧﺎﺹ‬ ‫�‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺑﻪ ﻣﻌﺮﻓﻲ ﺑﺮﺧﻲ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺑﺎ ﺳﺎﺧﺘﺎﺭ ﺧﺎﺹ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ‬
‫ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬ ‫ﻣﺘﻘﺎﺭﻥ ‪۲۸‬‬ ‫ﻣﺎﺗﺮﻳﺴﻲ ﻛﻪ ﺑﺎ ﺗﺮﺍﻧﻬﺎﺩﻩﺍﺵ ﺑﺮﺍﺑﺮ ﺑﺎﺷﺪ ﺑﻪ ﻣﺎﺗﺮﻳﺲ‬
‫ﺧﻮﺍﺹ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﺘﻘﺎﺭﻥ‬

‫‪ .۱‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﺘﻘﺎﺭﻥ ﻧﺴﺒﺖ ﺑﻪ ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﺑﺴﺘﻪﺍﻧﺪ‪ .‬ﻳﻌﻨﻲ ﺟﻤﻊ ﻭ ﻳﺎ ﺗﻔﺮﻳﻖ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ ﺧﻮﺩ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ‬
‫ﺩﻳﮕﺮﻱ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺿﺮﺏ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ ﻟﺰﻭﻣﺎ ﻣﺘﻘﺎﺭﻥ ﻧﻴﺴﺖ‪.‬‬

‫‪ .۲‬ﻭﺍﺭﻭﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ )ﺩﺭ ﺻﻮﺭﺕ ﻭﺟﻮﺩ( ﻣﺘﻘﺎﺭﻥ ﺍﺳﺖ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻗﻄﺮﻱ‬
‫ﻣﺎﺗﺮﻳﺴﻲ ﻛﻪ ﺗﻤﺎﻣﻲ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻏﻴﺮﻗﻄﺮﻱﺍﺵ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺑﺎﺷﺪ ﻭ ﻋﻨﺎﺻﺮ ﺭﻭﻱ ﻗﻄﺮ ﻣﻲﺗﻮﺍﻧﻨﺪ ﻏﻴﺮﺻﻔﺮ ﻭ ﻳﺎ ﺻﻔﺮ ﺑﺎﺷﻨﺪ‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪d11‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪0‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪d22‬‬ ‫‪...‬‬ ‫‪0 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪D=‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪dnn‬‬
‫ﺧﻮﺍﺹ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻗﻄﺮﻱ‬

‫‪ .۱‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻗﻄﺮﻱ ﻧﺴﺒﺖ ﺑﻪ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ ﻭ ﺿﺮﺏ ﺑﺴﺘﻪﺍﻧﺪ‪.‬‬

‫ﺳﻄﺮ‬ ‫ﭼﭗ‬
‫‪ .۲‬ﺍﮔﺮ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻗﻄﺮﻱ ‪ D‬ﺭﺍ ﺍﺯ ﺳﻤﺖ ﺭﺍﺳﺖ ﺩﺭ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺳﺘﻮﻥ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ﺩﺭ ‪dii‬‬
‫ﺿﺮﺏ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪ .۳‬ﻭﺍﺭﻭﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻗﻄﺮﻱ )ﺩﺭ ﺻﻮﺭﺕ ﻭﺟﻮﺩ(‪ ،‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻗﻄﺮﻱ ﺍﺳﺖ ﻛﻪ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ﺁﻥ ﻣﻌﻜﻮﺱ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﺻﻠﻲ ﺍﺳﺖ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ‬
‫ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬ ‫ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ‪۲۹‬‬‫ﺍﮔﺮ ﺩﺭ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻛﻠﻴﻪ ﻣﻮﻟﻔﻪﻫﺎﻳﻲ ﻛﻪ ﺩﺭ ﺑﺎﻻﻱ ﻗﻄﺮ ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ﺭﺍ‬
‫ﺩﺭ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ‪ ،‬ﻣﻮﻟﻔﻪﻫﺎﻱ ﺭﻭﻱ ﻗﻄﺮ ﺍﺻﻠﻲ ﻭ ﺯﻳﺮ ﻗﻄﺮ ﺍﺻﻠﻲ ﻣﻲﺗﻮﺍﻧﻨﺪ ﺻﻔﺮ ﻭ ﻳﺎ ﻏﻴﺮﺻﻔﺮ ﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ‪،‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻛﻪ ﻛﻠﻴﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﻗﻄﺮ ﺍﺻﻠﻲ ﺁﻥ ﺻﻔﺮ ﺑﺎﺷﺪ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ‪ ۳۰‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺩﺭ ﺯﻳﺮ ﻓﺮﻡ ﻛﻠﻲ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ‬
‫ﺑﺎﻻ ﻭ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪28 Symetric‬‬ ‫‪30 Upper‬‬ ‫‪triangular‬‬
‫‪29 Lower‬‬ ‫‪triangular matrix‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۷۶‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪u11 u12 u13 . . .‬‬ ‫‪u1n‬‬ ‫‪l11‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪0‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪u22 u23 . . .‬‬ ‫‪u2n‬‬ ‫‪‬‬ ‫‪ l21 l22 0 . . . 0‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪u33 . . .‬‬ ‫‪u3n‬‬ ‫‪‬‬ ‫‪ l31 l32 l33 . . . 0‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪‬‬ ‫‪ .‬‬ ‫‪..‬‬ ‫‪.. . .‬‬ ‫‪‬‬
‫‪‬‬ ‫‪. un−1,n‬‬ ‫‪‬‬ ‫‪ .‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪‬‬ ‫‪ .‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. 0‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪unn‬‬ ‫‪ln1 ln2 ln3 . . . lnn‬‬

‫ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻ ﻣﺜﻠﺜﻲ‬ ‫ﻣﺎﺗﺮﻳﺲ ﭘﺎﻳﻴﻦ ﻣﺜﻠﺜﻲ‬

‫ﺍﮔﺮ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻫﻢ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﺑﺎﺷﺪ ﻭ ﻫﻢ ﺑﺎﻻ ﻣﺜﻠﺜﻲ‪ ،‬ﺁﻥﮔﺎﻩ ﺁﻥ ﻣﺎﺗﺮﻳﺲ ﻗﻄﺮﻱ ﺍﺳﺖ‪.‬‬
‫ﺧﻮﺍﺹ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﺜﻠﺜﻲ‬
‫‪ .۱‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﺜﻠﺜﻲ ﻧﺴﺒﺖ ﺑﻪ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ ﻭ ﺿﺮﺏ ﺑﺴﺘﻪﺍﻧﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺟﻤﻊ‪،‬ﺗﻔﺮﻳﻖ ﻭ ﺿﺮﺏ ﺩﻭ ﻣﺎﺗﺮﻳﺲ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ‬ ‫ﺑﺎﻻﻣﺜﻠﺜﻲ‬
‫ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﻣﺎﺗﺮﻳﺴﻲ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﺍﺳﺖ‪.‬‬
‫‪ .۲‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺑﺮﺍﺑﺮ ﺣﺎﺻﻞﺿﺮﺏ ﻋﻨﺎﺻﺮ ﺭﻭﻱ ﻗﻄﺮﺵ ﺍﺳﺖ‪.‬‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ‬ ‫ﺑﺎﻻﻣﺜﻠﺜﻲ‬
‫‪ .۳‬ﻭﺍﺭﻭﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ )ﺩﺭ ﺻﻮﺭﺕ ﻭﺟﻮﺩ(‪ ،‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﺍﺳﺖ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ‬

‫ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ‪ A ∈ Rn×n‬ﺭﺍ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﻄﺮﻱ ‪ ۳۱‬ﮔﻮﻳﻴﻢ ﻫﺮﮔﺎﻩ ﺩﺭ ﻫﺮ ﺳﻄﺮ‪ ،‬ﻗﺪﺭﻣﻄﻠﻖ ﻋﻨﺼﺮ ﻣﻮﺟﻮﺩ ﺩﺭ‬
‫ﺭﻭﻱ ﻗﻄﺮ ﺍﺯ ﻣﺠﻤﻮﻉ ﻗﺪﺭﻣﻄﻠﻖ ﻋﻨﺎﺻﺮ ﺩﻳﮕﺮ ﺩﺭ ﻫﻤﺎﻥ ﺳﻄﺮ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﻗﺪﺭﻣﻄﻠﻖ ﻋﻨﺼﺮ ﺭﻭﻱ ﻗﻄﺮ ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ‬
‫ﻣﺠﻤﻮﻉ ﻗﺪﺭﻣﻄﻠﻖ ﻋﻨﺎﺻﺮ ﺩﻳﮕﺮ ﺩﺭ ﻫﻤﺎﻥ ﺳﻄﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ‪ ۳۲‬ﻣﻲﻧﺎﻣﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭ‬
‫ﻣﺸﺎﺑﻪ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﻭ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﺘﻮﻧﻲ ﻧﻴﺰ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺑﻪ ﺯﺑﺎﻥ ﺭﻳﺎﺿﻲ‬
‫‪h i‬‬ ‫ﺑﻴﺎﻥ ﻣﻲﮔﺮﺩﻧﺪ‪.‬‬
‫ﺍﮔﺮ ‪ ،A := aij i=1,...,n, ∈ Rn×n‬ﺁﻥﮔﺎﻩ‪:‬‬
‫‪j=1,...,m.‬‬
‫‪X‬‬
‫‪n‬‬
‫‪:‬ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ‬ ‫≥ | ‪|aii‬‬ ‫‪|aij | ,‬‬ ‫‪i = 1, . . . , n,‬‬
‫‪j=1‬‬
‫‪j̸=i‬‬
‫‪X‬‬ ‫‪n‬‬
‫‪:‬ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﺍﻛﻴﺪ‬ ‫> | ‪|aii‬‬ ‫‪|aij | ,‬‬ ‫‪i = 1, . . . , n,‬‬
‫‪j=1‬‬
‫‪j̸=i‬‬
‫‪X‬‬ ‫‪n‬‬
‫‪:‬ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﺘﻮﻧﻲ‬ ‫≥ | ‪|ajj‬‬ ‫‪|aij | ,‬‬ ‫‪j = 1, . . . , n,‬‬
‫‪i=1‬‬
‫‪j̸=i‬‬
‫‪X‬‬ ‫‪n‬‬
‫‪:‬ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﺘﻮﻧﻲ ﺍﻛﻴﺪ‬ ‫> | ‪|ajj‬‬ ‫‪|aij | ,‬‬ ‫‪j = 1, . . . , n.‬‬
‫‪i=1‬‬
‫‪j̸=i‬‬

‫‪31 Row‬‬ ‫‪diagonally dominant matrix‬‬ ‫‪32 Row‬‬ ‫‪diagonally dominant matrix‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‬

‫‪‬‬ ‫‪‬‬
‫‪−3‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪‬‬ ‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﺍﺳﺖ‪.‬‬
‫‪ 7 12‬‬ ‫‪−1‬‬ ‫‪−3 ‬‬
‫‪‬‬ ‫‪‬‬ ‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﻄﺮﻱ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫‪2‬‬ ‫‪−3 ‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪21‬‬ ‫‪‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ )ﺍﻛﻴﺪ( ﺳﺘﻮﻧﻲ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫‪−2 3‬‬ ‫‪7‬‬ ‫‪−14‬‬
‫‪‬‬ ‫‪‬‬
‫‪−13‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 7‬‬ ‫‪12‬‬ ‫‪−1‬‬ ‫‪−3 ‬‬ ‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﻄﺮﻱ ﻭ ﺳﺘﻮﻧﻲ ﺍﺳﺖ‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪−3 ‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪21‬‬ ‫‪‬‬
‫‪−2 3‬‬ ‫‪7‬‬ ‫‪−14‬‬
‫‪‬‬ ‫‪‬‬
‫‪−13 0‬‬ ‫‪3‬‬ ‫‪11‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 7‬‬
‫‪‬‬ ‫‪12‬‬ ‫‪−1‬‬ ‫‪−3 ‬‬
‫‪‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ )ﺍﻛﻴﺪ( ﺳﻄﺮﻱ ﻧﻤﻲﺑﺎﺷﺪ‬
‫‪ 2‬‬ ‫‪−3 ‬‬ ‫ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ )ﺍﻛﻴﺪ( ﺳﺘﻮﻧﻲ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪21‬‬ ‫‪‬‬
‫‪−2 3‬‬ ‫‪7‬‬ ‫‪−14‬‬
‫ﺧﻮﺍﺹ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻏﺎﻟﺐ ﻗﻄﺮﻱ)ﺍﻛﻴﺪ(‬

‫‪ .۱‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﻄﺮﻱ ﻭ ﺳﺘﻮﻧﻲ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬

‫‪ .۲‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﻏﻴﺮﺍﻛﻴﺪ ﻟﺰﻭﻣﺎ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻭ ﻣﻨﻔﻲ‬

‫ﻧﺎﻣﻴﺪﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﻫﺮﮔﺎﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺑﺮﺩﺍﺭ ﻏﻴﺮﺻﻔﺮ‬ ‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ ‪۳۳‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ﻭ ﻣﺘﻘﺎﺭﻥ ‪ A ∈ Rn×n‬ﺭﺍ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‬
‫‪ ،x ∈ Rn‬ﻣﻘﺪﺍﺭ ‪ x⊤ Ax‬ﻋﺪﺩﻱ ﻣﺜﺒﺖ )ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﺻﻔﺮ( ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪∀x ̸= 0 : x⊤ Ax > 0‬‬
‫ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ﻧﺎﻣﺴﺎﻭﻱ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﻏﻴﺮﺍﻛﻴﺪ ﺑﺮﻗﺮﺍﺭ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ‪ ۳۴‬ﻣﻲﻧﺎﻣﻨﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﭼﻨﺪ ﻣﺜﺎﻝ‬
‫ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻣﺎﺗﺮﻳﺲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫"‬ ‫‪#‬‬
‫‪1 2‬‬
‫=‪A :‬‬
‫‪2‬‬ ‫‪9‬‬
‫‪h‬‬ ‫⊤‪i‬‬
‫‪ x := x1 x2‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺩﺍﺭﻳﻢ‬ ‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺪﺍﻧﻴﻢ ﻛﻪ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ ﻳﺎ ﺧﻴﺮ‪ .‬ﺑﺮﺩﺍﺭ ﺩﻟﺨﻮﺍﻩ‬
‫"‬ ‫‪#" #‬‬
‫‪h‬‬ ‫‪i 1 2 x‬‬
‫⊤‬ ‫‪1‬‬
‫‪x Ax = x1‬‬ ‫‪x2‬‬
‫‪2 9‬‬ ‫‪x2‬‬
‫‪=x21‬‬ ‫‪+ 4x1 x2 +‬‬ ‫‪9x22 .‬‬

‫‪33 Positive‬‬ ‫‪Definite‬‬ ‫‪34 Positive‬‬ ‫‪semi-definite‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۷۸‬‬

‫ﺗﺎﺑﻊ ﻓﻮﻕ ﺑﻪ ﺍﺯﺍﻱ ‪ x1 = x2 = 0‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﻫﻨﮕﺎﻣﻲ ﻛﻪ ‪ x1‬ﻭ ‪ x2‬ﻫﺮ ﺩﻭ ﻣﻨﻔﻲ ﻭ ﻳﺎ ﻫﺮ ﺩﻭ ﻣﺜﺒﺖ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫ﺗﺎﺑﻊ ﻓﻮﻕ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ‪ (x1 + 3x2 )2 − 2x1 x2‬ﻧﻴﺰ ﻧﻮﺷﺖ‪ ،‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ‬
‫ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﺍﮔﺮ ‪ x1‬ﻣﻨﻔﻲ ﻭ ‪ x2‬ﻣﺜﺒﺖ ﺑﺎﺷﺪ )ﻭ ﻳﺎ ﺑﺮﻋﻜﺲ(‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ‪x ̸= 0‬‬
‫ﻣﻘﺪﺍﺭ ‪ x⊤ Ax‬ﻣﺜﺒﺖ ﻣﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪.‬‬
‫ﺣﺎﻝ ﻣﺎﺗﺮﻳﺲ‬
‫"‬ ‫‪#‬‬
‫‪1‬‬ ‫‪−1‬‬
‫=‪B :‬‬
‫‪−1‬‬ ‫‪1‬‬
‫‪h‬‬ ‫⊤‪i‬‬
‫‪ x := x1 x2‬ﺩﺍﺭﻳﻢ‪:‬‬ ‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﻪ ﺍﺯﺍﻱ ﺑﺮﺩﺍﺭ ﺩﻟﺨﻮﺍﻩ‬
‫‪x⊤ Bx = x21 − 2x1 x2 + x22 = (x1 − x2 )2 .‬‬
‫ﻟﺬﺍ ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ﻫﻤﻮﺍﺭﻩ ‪ .x⊤ Bx ≥ 0‬ﺍﻣﺎ ﺑﺰﺭﮒﺗﺮ ﺍﻛﻴﺪ ﺑﻮﺩﻥ ‪ .x⊤ Bx‬ﺑﻪ ﺍﺯﺍﻱ ‪ x ̸= 0‬ﻧﺘﻴﺠﻪ ﻧﻤﻲﺷﻮﺩ‪ .‬ﭼﺮﺍ‬
‫ﻛﻪ ﺩﺭ ﺣﺎﻻﺗﻲ ﻛﻪ ‪ x1 = x2 ̸= 0‬ﺁﻥﮔﺎﻩ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ ﻛﻪ ‪ x ̸= 0‬ﺍﻣﺎ ‪ .x⊤ Bx = 0‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ‬
‫‪ B‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻟﻲ ﺩﻳﮕﺮ‪ ،‬ﻣﺎﺗﺮﻳﺲ‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪‬‬
‫‪B := ‬‬
‫‪3‬‬ ‫‪2‬‬
‫‪‬‬‫‪−7‬‬
‫‪5 2 3‬‬
‫‪h‬‬ ‫⊤‪i‬‬
‫‪ .x := 0 1‬ﺣﺎﻝ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬ ‫‪0‬‬ ‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‬
‫‪x⊤ Bx = −7.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ‪ B‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻭ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺗﻮﺍﻥ‬
‫ﻣﺜﺒﺖ‬ ‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﻧﺎﻣﻨﻔﻲ ﺑﺎﺷﻨﺪ‪ .‬ﺑﺮﺍﻱ ﺍﺛﺒﺎﺕ ﺍﻳﻦ‬ ‫ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ﺍﮔﺮ ﻣﺎﺗﺮﻳﺴﻲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻟﺰﻭﻣﺎ ﺑﺎﻳﺪ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ﺁﻥ‬
‫ﺍﺩﻋﺎ ﻛﺎﻓﻲ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺗﻌﺮﻳﻒ ﻗﺮﺍﺭ ﺩﻫﻴﻢ‪ .x = ei :‬ﺍﻟﺒﺘﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺭ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ‪ ،‬ﻋﻨﺎﺻﺮ ﻏﻴﺮﻗﻄﺮﻱ ﻣﻨﻔﻲ‬
‫ﺑﺎﺷﻨﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﺗﻤﺎﻣﻲ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﺩﻟﻴﻠﻲ ﺑﺮ ﻣﺜﺒﺖ ﻣﻌﻴﻦ ﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫ﺍﺻﻮﻻ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻌﺮﻳﻒ ﻭ ﺭﻭﺵﻫﺎﻱ ﺟﺒﺮﻱ ﺑﻪ ﺻﻮﺭﺕ ﻓﻮﻕ‪ ،‬ﺑﺮﺍﻱ ﺗﺸﺨﻴﺺ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻭ ﻳﺎ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﺳﺎﺩﻩ‬
‫ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺧﺼﻮﺻﺎ ﺍﮔﺮ ﺑﻌﺪ ﻣﺎﺗﺮﻳﺲ)ﺗﻌﺪﺍﺩ ﺳﻄﺮ ﻳﺎ ﺳﺘﻮﻥﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ( ﺑﺰﺭﮒ ﺑﺎﺷﺪ‪ .‬ﺭﻭﺵﻫﺎﻱ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﺗﺸﺨﻴﺺ ﻣﻌﻴﻦ‬
‫ﻭ ﻳﺎ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﻣﺒﺎﺣﺚ ﺁﺗﻲ ﺑﻪ ﺁﻥ ﺍﺷﺎﺭﻩ ﺧﻮﺍﻫﺪ ﺷﺪ )؟؟ ﺭﺍ ﺑﺒﻴﻨﻴﺪ(‪.‬‬
‫ﻣﺸﺎﺑﻪ ﺑﺎ ﻣﻌﻴﻦﻣﺜﺒﺖ‪ ،‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻴﻦ ﻣﻨﻔﻲ ‪ ۳۵‬ﻭ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﻨﻔﻲ ‪ ۳۶‬ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ ﻭ ﻣﺮﺑﻌﻲ ‪ A‬ﺭﺍ‬
‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬ ‫ﻣﻌﻴﻦ ﻣﻨﻔﻲ‬
‫ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﻨﻔﻲ ﮔﻮﻳﻴﻢ ﻫﺮﮔﺎﻩ ‪ −A‬ﻣﺎﺗﺮﻳﺴﻲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ .‬ﻣﺎﺗﺮﻳﺴﻲ ﻛﻪ ﻧﻪ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻭ ﻧﻪ‬
‫)ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﻨﻔﻲ ﺑﺎﺷﺪ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻧﺎﻣﻌﻴﻦ ‪ ۳۷‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺧﻮﺍﺹ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻴﻦ ﻭ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬ ‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬ ‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫‪ .۱‬ﺟﻤﻊ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻣﺎﺗﺮﻳﺴﻲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﻟﺰﻭﻣﺎ ﺗﻔﺮﻳﻖ ﺩﻭ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻳﻚ‬
‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫‪35 Negative‬‬ ‫‪definit‬‬ ‫‪37 Indifinite‬‬
‫‪36 Negative‬‬ ‫‪semi-definite‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۷۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪ .۲‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﻭ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻫﻤﻮﺍﺭﻩ ﻭﺍﺭﻭﻥﭘﺬﻳﺮﻧﺪ‪.‬‬
‫ﺍﻣﺎ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ )ﻛﻪ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻧﺒﺎﺷﻨﺪ( ﺩﺍﺭﺍﻱ ﺩﺗﺮﻣﻴﻨﺎﻥ ﺻﻔﺮ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻟﺬﺍ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﻧﻴﺴﺘﻨﺪ‪.‬‬

‫‪ .۳‬ﺍﮔﺮ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﻣﺘﻘﺎﺭﻥ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺍﻋﻀﺎﻱ ﻗﻄﺮﻱﺍﺵ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﺍﺳﺖ‪.‬‬

‫■ ﻣﺎﺗﺮﻳﺲ ﺧﻠﻮﺕ)ﺗﻨﻚ(‬
‫ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲﻫﺎ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ‬ ‫ﺧﻠﻮﺕ ‪۳۸‬‬
‫ﻣﺎﺗﺮﻳﺲﻫﺎﻳﻲ ﻛﻪ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺁﻥﻫﺎ ﺻﻔﺮ ﺑﺎﺷﺪ ﺑﻪ ﻣﺎﺗﺮﻳﺲ‬
‫ﻛﺎﺭﺑﺮﺩﻫﺎ ﻇﺎﻫﺮ ﻣﻲﺷﻮﻧﺪ ﻭ ﻣﻲﺗﻮﺍﻥ ﺫﺧﻴﺮﻩ ﺁﻥﻫﺎ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﻧﺠﺎﻡ ﺩﺍﺩ ﻛﻪ ﻓﻀﺎﻱ ﻛﻢﺗﺮﻱ ﺍﺳﺘﻔﺎﺩﻩ ﮔﺮﺩﺩ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪﺍﻱ ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺗﻨﻚ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﺑﺎﺭﺑﻞ ‪ ۳۹‬ﺍﺷﺎﺭﻩ ﻛﺮﺩ ﻛﻪ ﺩﺭ ﻣﺒﺤﺚ ﮔﺮﺍﻑﻫﺎ ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪0‬‬

‫ﺑﻌﺪ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ‪ 2713‬ﺍﺳﺖ ﻭ ﺩﺍﺭﺍﻱ‬

‫‪ 7, 360, 369‬ﻣﻮﻟﻔﻪ ﺍﺳﺖ‪ .‬ﺍﻣﺎ‪ ،‬ﺗﻨﻬﺎ‬
‫‪500‬‬
‫‪ 18, 441‬ﻣﻮﻟﻔﻪ ﺍﺯ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ‬
‫ﻏﻴﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﺑﻘﻴﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺁﻥ‬
‫‪1000‬‬
‫ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ‪ 0.251 %‬ﻣﻮﻟﻔﻪﻫﺎ‬
‫ﻏﻴﺮ ﺻﻔﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺷﻜﻞ ﺭﻭﺑﺮﻭ‪،‬‬
‫‪1500‬‬ ‫ﻣﻮﻟﻔﻪﻫﺎﻱ ﻏﻴﺮﺻﻔﺮ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﻧﻘﻄﻪﻫﺎﻱ‬
‫ﺳﻴﺎﻩ ﺭﻧﮓ ﻣﺸﺨﺺ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺗﺒﻌﺎ‪ ،‬ﺑﺎ‬
‫‪2000‬‬ ‫ﺗﻮﺟﻪ ﺑﻪ ﺗﻨﻚﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﺎ‬
‫ﻓﻀﺎﻱ ﻛﻤﺘﺮﻱ ﺫﺧﻴﺮﻩﺳﺎﺯﻱ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﺍﺩ ﻭ‬
‫‪2500‬‬
‫ﻫﻤﭽﻨﻴﻦ ﻣﺤﺎﺳﺒﺎﺕ ﺭﻭﻱ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ‬
‫ﺻﺮﻓﻪﺟﻮﻳﻲ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪.‬‬
‫‪0‬‬ ‫‪500‬‬ ‫‪1000‬‬ ‫‪1500‬‬ ‫‪2000‬‬ ‫‪2500‬‬
‫)‪Nonzeros = 18441 (0.251%‬‬

‫ﻣﻌﺮﻓﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ‬ ‫‪۲.۴‬‬

‫ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ‪ ۴۰‬ﺍﺯ ‪ n‬ﻣﺘﻐﻴﺮ ‪ x1 , x2 , . . . , xn‬ﺑﻪ ﻓﺮﻡ ﻛﻠﻲ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪a1 x1 + a2 x2 + · · · + an xn = b‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ai‬ﻫﺎ ﻭ ‪ b‬ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻭ ﻣﻌﻠﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ ﺩﺭ ﻓﺮﻡ ﻛﻠﻲ ﻓﻮﻕ ﺩﻳﺪﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ‪،‬‬
‫ﻣﺘﻐﻴﺮﻫﺎ ﺩﺭ ﻳﻚ ﻋﺪﺩ ﺛﺎﺑﺖ ﺿﺮﺏ ﻣﻲﺷﻮﻧﺪ ﻭ ﺑﺎ ﻫﻢ ﺟﻤﻊ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺑﻪ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺍﻣﺎ‬
‫ﺩﺭ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ‪ ،‬ﻣﺘﻐﻴﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺩﺭ ﻫﻢ ﺿﺮﺏ ﻧﻤﻲﺷﻮﻧﺪ ﻳﺎ ﺑﻪ ﺗﻮﺍﻥ ﻧﻤﻲﺭﺳﻨﺪ ﻭ ﻳﺎ ﺩﺍﺧﻞ ﺁﺭﮔﻮﻣﺎﻥ ﺗﺎﺑﻊ ﺩﻳﮕﺮﻱ ﻗﺮﺍﺭ‬
‫ﻧﻤﻲﮔﻴﺮﻧﺪ‪.‬‬
‫ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﻓﻮﻕ‪ ،‬ﺍﮔﺮ ‪ ،n > 1‬ﺁﻥﮔﺎﻩ ﺑﻴﺶ ﺍﺯ ﻳﻚ ﻣﺠﻬﻮﻝ ﺩﺍﺭﻳﻢ ﻭ ﻟﺬﺍ ﻣﻌﺎﺩﻟﻪ ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺣﺎﻝ‪،‬‬
‫ﺍﮔﺮ ﭼﻨﺪﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﺩﺭ ﻛﻨﺎﺭ ﻫﻢ ﻗﺮﺍﺭ ﮔﻴﺮﻧﺪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ‪ ۴۱‬ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻨﺪ‪ .‬ﻓﺮﻡ ﻛﻠﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ‬
‫‪38 Sparse‬‬ ‫‪40 Linear‬‬ ‫‪equation‬‬
‫‪39 Barbell‬‬ ‫‪41 System‬‬ ‫‪of linear equations‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸۰‬‬

‫ﺧﻄﻲ ﺑﺎ ‪ m‬ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﻭ ‪ n‬ﻣﺘﻐﻴﺮ ﻣﺠﻬﻮﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬

‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪a11 x1 + a12 x2 + · · · + a1n xn = b1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪a21 x1 + a22 x2 + · · · + a2n xn = b2‬‬
‫‪..‬‬ ‫‪..‬‬ ‫)‪(۸.۴‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪am1 x1 + am2 x2 + · · · + amn xn = bm‬‬
‫ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ ،‬ﺍﮔﺮ ﻛﻠﻴﻪ ‪bi‬ﻫﺎ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻫﻤﮕﻦ ‪ ۴۲‬ﮔﻮﻳﻴﻢ‪ .‬ﺑﺮ ﺍﻳﻦ ﻣﺒﻨﺎ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺑﻪ ﺩﻭ‬
‫ﺩﺳﺘﻪ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻫﻤﮕﻦ ﻭ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻧﺎﻫﻤﮕﻦ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻓﻮﻕ‪ ،‬ﺗﻌﺪﺍﺩ ‪ n‬ﻣﺠﻬﻮﻝ ﻭ ‪m‬‬
‫ﻣﻌﺎﺩﻟﻪ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﺴﺘﻪ ﺑﻪ ‪ m‬ﻭ ‪ ،n‬ﺩﺳﺘﻪﺑﻨﺪﻱ ﺯﻳﺮ ﺍﺯ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪:‬‬

‫• ﺍﮔﺮ ‪) m < n‬ﺗﻌﺪﺍﺩ ﻣﻌﺎﺩﻻﺕ ﻛﻤﺘﺮ ﺍﺯ ﺗﻌﺪﺍﺩ ﻣﺠﻬﻮﻻﺕ ﺑﺎﺷﺪ(‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ‬
‫ﺯﻳﺮﻣﻌﻴﻦ ‪ ۴۳‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﻣﺮﺑﻌﻲ ‪۴۴‬‬ ‫• ﺍﮔﺮ ‪) m = n‬ﺗﻌﺪﺍﺩ ﻣﻌﺎﺩﻻﺕ ﺑﺎ ﺗﻌﺪﺍﺩ ﻣﺠﻬﻮﻻﺕ ﺑﺮﺍﺑﺮ ﺑﺎﺷﺪ(‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ‬
‫ﮔﻮﻳﻴﻢ‪.‬‬
‫• ﺍﮔﺮ ‪) m > n‬ﺗﻌﺪﺍﺩ ﻣﻌﺎﺩﻻﺕ ﺑﻴﺸﺘﺮ ﺍﺯ ﺗﻌﺪﺍﺩ ﻣﺠﻬﻮﻻﺕ ﺑﺎﺷﺪ(‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻓﻮﻕﻣﻌﻴﻦ ‪ ۴۵‬ﮔﻮﻳﻴﻢ‪.‬‬

‫ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ ﺑﺮ ﺭﻭﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺮﺑﻌﻲ ﻣﺘﻤﺮﻛﺰ ﻣﻲﺷﻮﻳﻢ ﻭ ﺩﺭ ﻣﻮﺭﺩ ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﺟﻮﺍﺏﻫﺎﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻭ‬
‫ﻫﻤﻴﻨﻄﻮﺭ ﺭﻭﺵﻫﺎﻱ ﺣﻞ ﺁﻥ ﺑﺤﺚ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫■ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻭ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻓﺮﻡ ﻛﻠﻲ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬
‫‪a11 x1 + a12 x2 + · · · + a1n xn = b1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪a21 x1 + a22 x2 + · · · + a2n xn = b2‬‬
‫‪..‬‬ ‫‪..‬‬ ‫)‪(۹.۴‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪an1 x1 + an2 x2 + · · · + ann xn = bn‬‬
‫ﻛﻪ ‪ x1 , . . . , xn‬ﻣﺠﻬﻮﻻﺕ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ‪ai,j‬ﻫﺎ ﻭ ‪bj‬ﻫﺎ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﻣﻌﻠﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺯﻳﺮ ﻧﻴﺰ ﻧﺸﺎﻥ ﺩﺍﺩ‪:‬‬
‫‪Ax = b‬‬ ‫)‪(۱۰.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪x1‬‬ ‫‪b1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n ‬‬ ‫‪ x2 ‬‬ ‫‪ b2 ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪A :=  .‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫‪x :=  .  ,‬‬ ‫‪b :=  . ‬‬ ‫)‪(۱۱.۴‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫···‬ ‫‪. ‬‬ ‫‪ .. ‬‬ ‫‪ .. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪xn‬‬ ‫‪bn‬‬

‫‪42 Homogeneous‬‬ ‫‪44 Squred‬‬

‫‪43 Under-determined‬‬ ‫‪45 Over-determined‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۸۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ‪ Ax = b‬ﻳﻚ ﻓﺮﻡ ﻓﺸﺮﺩﻩ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ )‪ (۹.۴‬ﺍﺳﺖ‪ .‬ﻏﺎﻟﺐ ﻣﻨﺎﺑﻊ ﻭ ﻣﺮﺍﺟﻊ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ‬
‫ﺭﺍ ﺑﻪ ﻓﺮﻡ ‪ Ax = b‬ﻣﺸﺨﺺ ﻣﻲﻛﻨﻨﺪ ﻭ ﻟﺬﺍ ﺍﻳﻦ ﻓﺮﻡ ﻳﻚ ﻓﺮﻡ ﻛﺎﻧﻮﻧﻲ ﻭ ﺍﺳﺘﺎﻧﺪﺍﺭﺩ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺍﺳﺖ‪ .‬ﺗﺎﻛﻴﺪ‬
‫ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎ ﺗﻐﻴﻴﺮ ﻧﺎﻡ ﻣﺘﻐﻴﺮﻫﺎ ﻭ ﺩﺳﺘﻜﺎﺭﻱﻫﺎﻱ ﺟﺒﺮﻱ ﻣﻲﺗﻮﺍﻥ ﻫﺮ ﺩﺳﺘﮕﺎﻫﻲ ﺭﺍ ﺑﻪ ﻓﺮﻡ ‪ Ax = b‬ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ(‬ ‫ﻣﺜﺎﻝ ‪)۳−۴‬ﻓﺮﻡ ﺑﺮﺩﺍﺭﯼ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪3θ − 2x = 2r − 1‬‬
‫‪‬‬
‫‪θ =r+8‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2θ + x − 3r = 6‬‬
‫ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺍﺳﺖ ﻛﻪ ‪ x ،θ‬ﻭ ‪ r‬ﻣﺠﻬﻮﻻﺕ ﺁﻥ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻗﺪﻡ ﺍﻭﻝ ﺩﺭ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻡ ﺑﺮﺩﺍﺭﻱ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‬
‫ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻫﻢﻧﺎﻡ ﻛﺮﺩﻥ ﻣﺘﻐﻴﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺗﻐﻴﻴﺮ ﻣﺘﻌﻴﺮﻫﺎﻱ ﺯﻳﺮ ﺭﺍ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪θ → x1 ,‬‬ ‫‪x → x2 ,‬‬ ‫‪r → x3‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪3x − 2x2 = 2x3 − 1‬‬
‫‪ 1‬‬
‫‪x1 = x3 + 8‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x1 + x2 − 3x3 = 6‬‬
‫ﻫﻤﭽﻨﻴﻦ ﺑﺎ ﺩﺳﺘﻜﺎﺭﻱ ﺟﺒﺮﻱ ﻣﻌﺎﺩﻻﺕ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪3x − 2x2 − 2x3 = −1‬‬
‫‪ 1‬‬
‫‪x1 − x3 = 8‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x1 + x2 − 3x3 = 6‬‬
‫ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻣﻲ ﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ‪ Ax = b‬ﻧﻮﺷﺖ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬‫‪‬‬
‫‪3‬‬ ‫‪−2‬‬ ‫‪−2‬‬ ‫‪x1‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪A=‬‬
‫‪1‬‬ ‫‪0‬‬ ‫‪−1‬‬
‫‪,‬‬ ‫‪x=‬‬ ‫‪‬‬
‫‪x2  ,‬‬ ‫‪b=‬‬
‫‪ 8‬‬
‫‪‬‬
‫‪2‬‬ ‫‪1‬‬ ‫‪−3‬‬ ‫‪x3‬‬ ‫‪6‬‬

‫■ ﺩﺳﺘﻪﺑﻨﺪﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺮﺑﻌﻲ‬

‫ﻣﻌﻤﻮﻻ ﺩﺳﺘﮕﺎﻩ ﻣﺮﺑﻌﻲ ‪ Ax = b‬ﺑﺴﺘﻪ ﺑﻪ ﻧﻮﻉ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻧﺎﻣﮕﺬﺍﺭﻱ ﻣﻲﺷﻮﺩ‪ .‬ﻣﺜﻼ ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻ )ﻳﺎ ﭘﺎﻳﻴﻦ(‬
‫ﻣﺜﻠﺜﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻ )ﻳﺎ ﭘﺎﻳﻴﻦ( ﻣﺜﻠﺜﻲ ﮔﻮﻳﻴﻢ‪ .‬ﺑﻪ ﻫﻤﻴﻦ ﺭﻭﺍﻝ‪ ،‬ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻏﺎﻟﺐ ﻗﻄﺮﻱ‬
‫)ﺍﻛﻴﺪ( ﺳﻄﺮﻱ ﻭ ﺳﺘﻮﻧﻲ‪ ،‬ﺩﺳﺘﮕﺎﻩﻫﺎﻱ )ﻧﻴﻤﻪ( ﻣﻌﻴﻦ ﻣﺜﺒﺖ‪ ،‬ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺗﻨﻚ ﻭ ﻏﻴﺮﻩ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺩﺭ ﺩﺳﺘﮕﺎﻩ ‪Ax = b‬‬
‫ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ ‪ Ax = b‬ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﻭﻟﻲ ﺍﮔﺮ ‪ A‬ﻭﺍﺭﻭﻥ ﭘﺬﻳﺮ )ﻧﺎﻣﻨﻔﺮﺩ( ﺑﺎﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ‪ Ax = b‬ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﻳﺎﺩﺁﻭﺭﻱ ﻣﻲﺷﻮﺩ ﻛﻪ‪ ،‬ﺍﮔﺮ ﺑﺮﺩﺍﺭ ‪ b‬ﺑﺮﺍﺑﺮ ﺑﺮﺩﺍﺭ ﺻﻔﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ‬
‫ﻫﻤﮕﻦ ﻧﺎﻣﻴﺪﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸۲‬‬

‫■ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺩﺳﺘﮕﺎﻩ‬

‫ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﻭ ﻧﮕﻪﺩﺍﺭﻱ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺩﺭ ﻛﺎﻣﭙﻴﻮﺗﺮ‪ ،‬ﻛﺎﻓﻲ ﺍﺳﺖ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻭ ﺑﺮﺩﺍﺭ ‪ b‬ﺭﺍ ﺫﺧﻴﺮﻩ ﻛﻨﻴﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺮﺍﻱ‬
‫ﻳﻚﭘﺎﺭﭼﮕﻲ ﺩﺭ ﺫﺧﻴﺮﻩ‪ ،‬ﻣﻲﺗﻮﺍﻧﻴﻢ ﺑﺮﺩﺍﺭ ‪ b‬ﺭﺍ ﺩﺭ ﺍﻧﺘﻬﺎﻱ ﺳﺘﻮﻥﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻗﺮﺍﺭ ﺩﻫﻴﻢ ﻭ ﻣﺎﺗﺮﻳﺴﻲ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ ﻛﻪ ﺑﻪ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ‪ ۴۶‬ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺩﺳﺘﮕﺎﻩ )‪ (۹.۴‬ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪b1‬‬
‫‪‬‬ ‫‪‬‬
‫‪h‬‬ ‫‪i‬‬ ‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪b2 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪A‬‬ ‫‪b = .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫)‪(۱۲.۴‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪bn‬‬

‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻓﻮﻕ ﻳﻚ ﻣﺎﺗﺮﻳﺲ )‪ n × (n + 1‬ﺍﺳﺖ‪ .‬ﮔﺎﻫﻲ ﺑﺮﺍﻱ ﺑﻬﺘﺮ ﻣﺸﺨﺺ ﺷﺪﻥ ﺳﺎﺧﺘﺎﺭ ﻣﺎﺗﺮﻳﺲ‬
‫ﺍﻓﺰﻭﺩﻩ‪ ،‬ﺳﺘﻮﻥ ‪n + 1‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﻳﻚ ﺧﻂ ﺍﺯ ﺑﻘﻴﻪ ﺳﺘﻮﻥﻫﺎ ﻣﺠﺰﺍ ﻣﻲﺷﻮﺩ ﻭ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﺷﻜﻞ ﺯﻳﺮ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬

‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪b1‬‬
‫‪‬‬ ‫‪‬‬
‫‪h‬‬ ‫‪i‬‬ ‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪b2 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪A‬‬ ‫‪b = .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫)‪(۱۳.۴‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪bn‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻛﻠﻴﻪ ﺍﻃﻼﻋﺎﺕ ﺁﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺩﺭ ﺧﻮﺩ ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﭼﻮﻥ ﺫﺧﻴﺮﻩ‬
‫ﻣﺎﺗﺮﻳﺲﻫﺎ ﺩﺭ ﻛﺎﻣﭙﻴﻮﺗﺮ ﺑﻪ ﺳﺎﺩﮔﻲ ﺍﻧﺠﺎﻡ ﻣﻲﮔﻴﺮﺩ‪ ،‬ﻟﺬﺍ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﺮﺍﻱ ﺫﺧﻴﺮﻩ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﻛﺎﻣﭙﻴﻮﺗﺮﻫﺎ ﮔﺰﻳﻨﻪ‬
‫ﻣﻨﺎﺳﺒﻲ ﺍﺳﺖ‪.‬‬

‫ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ‬ ‫�‬

‫ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﺮﺑﻌﻲ‪ ،‬ﺍﻧﺘﻈﺎﺭ ﺩﺍﺭﻳﻢ ﻛﻪ ﻳﻚ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺣﺎﻟﺖﻫﺎﻱ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ‬
‫ﺧﻄﻲ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ ﻛﻪ ﺑﻪ ﺑﺮﺭﺳﻲ ﺁﻥﻫﺎ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬
‫ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ )ﻣﺮﺑﻌﻲ( ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﻭ ﻣﻌﺎﺩﻟﻪ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻧﺎﺳﺎﺯﮔﺎﺭ ‪ ۴۷‬ﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺩﻭ ﻣﻌﺎﺩﻟﻪ ﺑﺎ ﻳﻚﺩﻳﮕﺮ‬
‫ﺩﺭ ﺗﻨﺎﻗﺾ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪2x + x2 − x3 = 1‬‬
‫‪ 1‬‬
‫‪x1 − 2x2 + 3x3 = 4‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪4x1 + 2x2 − 2x3 = 1‬‬
‫ﻣﻌﺎﺩﻻﺕ ﺍﻭﻝ ﻭ ﺳﻮﻡ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﺩﺭ ﺗﻨﺎﻗﺾ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﺑﻪ ﺍﺻﻄﻼﺡ ﻧﺎﺳﺎﺯﮔﺎﺭﻧﺪ‪ .‬ﺗﺒﻌﺎ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪ .‬ﺩﺳﺘﮕﺎﻫﻲ ﻛﻪ‬
‫ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻧﺎﺳﺎﺯﮔﺎﺭ ‪ ۴۸‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻟﺒﺘﻪ ﮔﺎﻫﻲ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺩﻭ ﻣﻌﺎﺩﻟﻪ ﺑﺎ ﻫﻢ ﻧﺎﺳﺎﺯﮔﺎﺭ‬

‫‪46 Augmented‬‬ ‫‪matrix‬‬ ‫‪48 Inconsistent‬‬ ‫‪linear system‬‬

‫‪47 Inconsistent‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۸۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻧﺒﺎﺷﻨﺪ‪ ،‬ﺑﻠﻜﻪ ﭼﻨﺪﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺑﺎ ﻫﻢ ﺳﺎﺯﮔﺎﺭ ﻧﺒﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﺭ ﺩﺳﺘﮕﺎﺭ ﺯﻳﺮ‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x1 + x2 − x3 − x4 = 2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪x1 − 2x2 + x3 − x4 = 4‬‬
‫‪‬‬
‫‪‬‬ ‫‪− x1 − x2 + 2x3 + x4 = 1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x − x − 2x = 4‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪4‬‬

‫ﻫﻴﭻ ﺩﻭ ﻣﻌﺎﺩﻟﻪﺍﻱ ﺑﺎ ﻫﻢ ﻧﺎﺳﺎﺯﮔﺎﺭ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻣﺎ ﻣﻌﺎﺩﻻﺕ ﺍﻭﻝ‪ ،‬ﺩﻭﻡ ﻭ ﭼﻬﺎﺭﻡ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻧﺎﺳﺎﺯﮔﺎﺭﻧﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺳﻤﺖ ﭼﭗ ﻣﻌﺎﺩﻟﻪ‬
‫ﭼﻬﺎﺭﻡ ﺑﺮﺍﺑﺮ ﻣﺠﻤﻮﻉ ﺳﻤﺖ ﭼﭗ ﻣﻌﺎﺩﻟﻪﻫﺎﻱ ﺍﻭﻝ ﻭ ﺩﻭﻡ ﺍﺳﺖ ﻭﻟﻲ ﺍﻳﻦ ﻧﻜﺘﻪ ﺩﺭ ﻣﻮﺭﺩ ﺳﻤﺖ ﺭﺍﺳﺖ ﻣﻌﺎﺩﻻﺕ ﺑﺮﻗﺮﺍﺭ ﻧﻴﺴﺖ‪.‬‬
‫ﻫﻤﭽﻨﻴﻦ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ )ﻣﺮﺑﻌﻲ( ﻣﻌﺎﺩﻻﺕ ﺑﺎ ﻳﻜﺪﻳﮕﺮ ﻧﺎﺳﺎﺯﮔﺎﺭ ﻧﺒﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ﻣﻌﺎﺩﻟﻪ ﻳﺎ ﻣﻌﺎﺩﻻﺕ ﺯﺍﻳﺪ‬
‫ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x1 + x2 − x3 − x4 = 2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪x1 − 2x2 + x3 − x4 = 4‬‬
‫‪‬‬
‫‪‬‬ ‫‪− x1 − x2 + 2x3 + x4 = 1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x − x − 2x = 6‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪4‬‬

‫ﻣﻌﺎﺩﻟﻪ ﭼﻬﺎﺭﻡ ﺍﺯ ﺟﻤﻊ ﻣﻌﺎﺩﻻﺕ ﺍﻭﻝ ﻭ ﺩﻭﻡ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺯﺍﻳﺪ ﺩﺍﺭﻳﻢ‪ .‬ﻭﻗﺘﻲ ﺩﺳﺘﮕﺎﻩ‬
‫ﺳﺎﺯﮔﺎﺭ ﺑﺎﺷﺪ ﻭ ﻣﻌﺎﺩﻟﻪ ﺯﺍﻳﺪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺧﺎﻃﺮ ﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﺣﺎﻟﺖ ﺩﺍﺷﺘﻦ ﺩﻭ ﻳﺎ ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﺟﻮﺍﺏ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﮔﺮ‬
‫ﺩﺳﺘﮕﺎﻩ ﺩﻭ ﺟﻮﺍﺏ ﻳﺎ ﭼﻨﺪ ﺟﻮﺍﺏ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﻫﻢ ﺧﻮﺍﻫﺪ ﺩﺍﺷﺖ‪ .‬ﺑﺮﺍﻱ ﺍﺛﺒﺎﺕ ﺍﻳﻦ ﺍﺩﻋﺎ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪v‬‬
‫ﻭ ‪ w‬ﺩﻭ ﺟﻮﺍﺏ ﻣﺘﻤﺎﻳﺰ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺩﺍﺭﻳﻢ‪ Av = b :‬ﻭ ‪ .Aw = b‬ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﻢ ‪ λ‬ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ‬
‫ﺩﻟﺨﻮﺍﻩ ﺑﺎﺷﺪ ﻭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ ‪ .y := λv + (1 − λ)w‬ﺣﺎﻝ ﺩﺍﺭﻳﻢ‪:‬‬
‫)‪Ay =A (λv + (1 − λ)w‬‬
‫‪=λAv + (1 − λ)Aw‬‬
‫‪=λb + (1 − λ)b = b‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ‪ ،λ‬ﺑﺮﺩﺍﺭ ‪ y‬ﻳﻚ ﺟﻮﺍﺏ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﺍﮔﺮ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺩﻭ ﺟﻮﺍﺏ ‪ v‬ﻭ ‪w‬‬
‫ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﻳﮕﺮ ﺑﻪ ﻓﺮﻡ ‪ λv + (1 − λ)w‬ﻧﻴﺰ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬
‫ﺑﺎ ﺟﻤﻊﺑﻨﺪﻱ ﻣﻄﺎﻟﺐ ﻓﻮﻕ‪ ،‬ﺣﺎﻟﺖﻫﺎﻱ ﺯﻳﺮ‪ ،‬ﺩﺭ ﻣﻮﺭﺩ ﺗﻌﺪﺍﺩ ﺟﻮﺍﺏﻫﺎﻱ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻣﺮﺑﻌﻲ‪ ،‬ﻣﻲﺗﻮﺍﻧﺪ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪:‬‬

‫‪ .۱‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪،‬‬

‫‪ .۲‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪،‬‬

‫‪ .۳‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺳﺘﮕﺎﻩ ﻫﻴﭻ ﺟﻮﺍﺑﻲ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ‪.‬‬

‫ﺍﻣﺎ ﺳﻮﺍﻟﻲ ﻛﻪ ﭘﻴﺶ ﻣﻲﺁﻳﺪ‪ ،‬ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺩﻩ ﺷﺪﻩ‪ ،‬ﭼﮕﻮﻧﻪ ﺗﺸﺨﻴﺺ ﺩﻫﻴﻢ ﻛﻪ ﻛﺪﺍﻡﻳﻚ ﺍﺯ ﺳﻪ ﺣﺎﻟﺖ‬
‫ﻓﻮﻕ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ؟ ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﻗﻀﻴﻪ ﺯﻳﺮ ﻣﻲﺗﻮﺍﻧﺪ ﻣﻔﻴﺪ ﻭﺍﻗﻊ ﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸۴‬‬

‫ﻗﻀﯿﻪ ‪ ۶−۴‬ﺩﺗﺮﻣﯿﻨﺎﻥ ﻣﺎﺗﺮﯾﺲ ﺑﺮﺍﯼ ﺗﺸﺨﯿﺺ ﻭﺟﻮﺩ ﻭ ﯾ ﺘﺎﯾﯽ ﺟﻮﺍﺏ‬

‫ﺩﺭ ﺩﺳﺘﮕﺎﻩ ‪:Ax = b‬‬

‫• ﺍﮔﺮ ‪ det(A) ̸= 0‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ‪،‬‬

‫• ﺍﮔﺮ ‪ det(A) = 0‬ﺁﻥﮔﺎﻩ ﻳﺎ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺩﺳﺘﮕﺎﻩ ﺍﺻﻼ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪.‬‬

‫ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﻦ ‪ ،Ax = 0‬ﺑﻪ ﺳﺎﺩﮔﻲ ﻧﺘﻴﺠﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ x = 0‬ﻳﻚ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺟﻮﺍﺏ ‪ 0‬ﺑﺮﺍﻱ‬
‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻫﻤﮕﻦ ﺑﻪ ﺟﻮﺍﺏ ﺑﺪﻳﻬﻲ ‪ ۴۹‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻫﻤﮕﻦ ﺣﺎﻟﺖ ﺑﺪﻭﻥ ﺟﻮﺍﺏ )ﻧﺎﺳﺎﺯﮔﺎﺭ( ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪.‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﻗﻀﻴﻪ ﻓﻮﻕ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻫﻤﮕﻦ ﺑﻪ ﺻﻮﺭﺕ ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺗﻄﺒﻴﻖ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻭﺟﻮﺩ ﻭ ﯾ ﺘﺎﯾﯽ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎەﻫﺎﯼ ﻫﻤ ﻦ‬ ‫ﻧﺘﯿﺠﻪ ‪۷−۴‬‬
‫ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﻦ ‪ Ax = 0‬ﻫﻤﻮﺍﺭﻩ ‪ x = 0‬ﻳﻚ ﺟﻮﺍﺏ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ‬

‫• ﺍﮔﺮ ‪ ،det(A) ̸= 0‬ﺁﻥﮔﺎﻩ ‪ x = 0‬ﺗﻨﻬﺎ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻣﻲﺑﺎﺷﺪ )ﺩﺳﺘﮕﺎﻩ ﻓﻘﻂ ﺟﻮﺍﺏ ﺑﺪﻳﻬﻲ ﺩﺍﺭﺩ(‪.‬‬

‫• ﺍﮔﺮ ‪ ،det(A) = 0‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‬ ‫�‬

‫ﺩﺭ ﺟﺒﺮ ﺧﻄﻲ‪ ،‬ﻣﻨﻈﻮﺭ ﺍﺯ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ ۵۰‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺁﻥ ﺩﺳﺘﻪ ﻋﻤﻞﻫﺎﻳﻲ ﻛﻪ ﺍﻋﻤﺎﻝ ﺁﻥﻫﺎ ﺑﺮ ﺭﻭﻱ ﺩﺳﺘﮕﺎﻩ ﻣﻨﺠﺮ‬
‫ﺑﻪ ﺗﻐﻴﻴﺮ ﺟﻮﺍﺏ ﻳﺎ ﺟﻮﺍﺏﻫﺎﻱ ﺩﺳﺘﮕﺎﻩ ﻧﺸﻮﺩ ﻭ ﻫﻤﭽﻨﻴﻦ ﺑﺎﻋﺚ ﺍﺿﺎﻓﻪ ﻭ ﻳﺎ ﻛﺎﻫﺶ ﺟﻮﺍﺏﻫﺎﻱ ﺩﺳﺘﮕﺎﻩ ﻧﺸﻮﺩ‪ .‬ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ‬
‫ﻣﻘﺪﻣﺎﺗﻲ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‪:‬‬
‫‪ : E1‬ﺟﻤﻊ ﻣﻀﺮﺑﻲ ﺍﺯ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺑﺎ ﻣﻌﺎﺩﻟﻪ ﺩﻳﮕﺮ‪.‬‬
‫‪ : E2‬ﺟﺎﺑﺠﺎﻳﻲ ﺩﻭ ﻣﻌﺎﺩﻟﻪ‪،‬‬
‫‪ : E3‬ﺿﺮﺏ ﻃﺮﻓﻴﻦ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺩﺭ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻏﻴﺮﺻﻔﺮ‪،‬‬
‫ﺍﻟﺒﺘﻪ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺮ ﺭﻭﻱ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻧﻴﺰ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪ .‬ﺍﻳﻦ ﺍﻋﻤﺎﻝ ﺑﺮﺍﻱ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ‬
‫ﻣﻲﺑﺎﺷﻨﺪ‪:‬‬
‫‪ : E1‬ﺟﻤﻊ ﻣﻀﺮﺑﻲ ﺍﺯ ﻳﻚ ﺳﻄﺮ ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﺎ ﺳﻄﺮ ﺩﻳﮕﺮ‪.‬‬
‫‪ : E2‬ﺟﺎﺑﺠﺎﻳﻲ ﺩﻭ ﺳﻄﺮ ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ‪،‬‬
‫‪ : E3‬ﺿﺮﺏ ﻳﻚ ﺳﻄﺮ ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺩﺭ ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻏﻴﺮﺻﻔﺮ‪،‬‬

‫ﺭﻭﺵﻫﺎﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺮﺑﻌﻲ‬ ‫�‬

‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻛﺎﺭﺑﺮﺩﻫﺎ ﻇﺎﻫﺮ ﻣﻲﺷﻮﻧﺪ ﻭ ﻧﻴﺎﺯ ﺑﻪ ﺣﻞ ﺳﺮﻳﻊ ﻭ ﺩﻗﻴﻖ ﺁﻥﻫﺎ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﻪ ﻫﻤﻴﻦ ﺩﻟﻴﻞ ﺭﻭﺵﻫﺎﻱ‬
‫ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻫﺮﻛﺪﺍﻡ ﻧﻘﺎﻁ ﺿﻌﻒ ﻭ ﻗﻮﺕ ﺧﺎﺹ ﺧﻮﺩ ﺭﺍ ﺩﺍﺭﻧﺪ‪ .‬ﻳﻚ ﺩﺳﺘﻪﺑﻨﺪﻱ ﻛﻠﻲ ﺍﺯ‬
‫ﺭﻭﺵﻫﺎﻱ ﺭﺍﻳﺞ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪49 Trivial‬‬ ‫‪solution‬‬ ‫‪50 Elementary‬‬ ‫‪row operations‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۸۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫• ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‬

‫ﺭﻭﺵ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ‬ ‫‪-‬‬

‫ﺭﻭﺵ ﻛﺮﺍﻣﺮ‬ ‫‪-‬‬
‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫‪-‬‬
‫ﺭﻭﺵﻫﺎﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺗﺠﺰﻳﻪ‬ ‫‪-‬‬

‫• ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ‬

‫‪ -‬ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‬
‫‪ -‬ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﺎﻳﺪﻝ‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺼﻞ ﺑﻪ ﻣﻄﺎﻟﻌﻪ ﻭ ﺑﺮﺭﺳﻲ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺭﻭﺵ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﺭﻭﺵ ﻛﺮﺍﻣﺮ‬ ‫�‬

‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥ ﮔﺎﻩ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪x = A−1 b.‬‬
‫ﻳﻌﻨﻲ ﺑﺮﺩﺍﺭ ﺟﻮﺍﺏ ﺍﺯ ﺿﺮﺏ ﻣﻌﻜﻮﺱ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺭ ﺑﺮﺩﺍﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﺣﺎﺻﻞ ﻣﻲ ﺷﻮﺩ‪ .‬ﺍﺯ ﺍﻳﻦ ﻧﻜﺘﻪ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﺣﻞ‬
‫ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﺑﺘﺪﺍ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﺭﺍ ﺩﺭ‬
‫ﺑﺮﺩﺍﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﺿﺮﺏ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫‪۵۱‬‬
‫ﺭﻭﺵ ﻛﺮﺍﻣﺮ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺑﺮﻣﺒﻨﺎﻱ ﻗﺎﻋﺪﻩ ﻛﺮﺍﻣﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫ﻗﺎﻋﺪﻩ ﮐﺮﺍﻣﺮ‬ ‫ﻗﻀﯿﻪ ‪۸−۴‬‬

‫ﺍﮔﺮ ‪ A ∈ Rn×n‬ﻣﺎﺗﺮﻳﺴﻲ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺑﻪ ﺻﻮﺭﺕ ﺻﺮﻳﺢ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫) ‪det(Bi‬‬
‫= ‪xi‬‬ ‫‪,‬‬ ‫‪i = 1, . . . , n,‬‬
‫)‪det(A‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ Bi‬ﻣﺎﺗﺮﻳﺴﻲ ﺍﺳﺖ ﻛﻪ ﺍﺯ ﺟﺎﺑﺠﺎﻳﻲ ﺳﺘﻮﻥ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺑﺎ ﺑﺮﺩﺍﺭ ‪ b‬ﺣﺎﺻﻞ ﻣﻲ ﺷﻮﺩ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺭﻭﺵ ﻛﺮﺍﻣﺮ‪ ،‬ﺑﺎ ﻣﺤﺎﺳﺒﻪ ‪ n + 1‬ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﺳﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ‬
‫ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺧﻄ ﺑﺎ ﺭﻭﺵ ﮐﺮﺍﻣﺮ(‬ ‫ﻣﺜﺎﻝ ‪)۴−۴‬ﺣﻞ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪3x − 2x2 + x3 = 9‬‬
‫‪ 1‬‬
‫‪x1 + 2x2 − 2x3 = −5‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪x1 + x2 − 4x3 = −2‬‬

‫‪51 Cramer’s‬‬ ‫‪rule‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸۶‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻗﺎﻋﺪﻩ ﻛﺮﺍﻣﺮ‪ ،‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻗﺎﺑﻞ ﻣﺤﺎﺳﺒﻪ ﺍﺳﺖ‪:‬‬
‫‪9 −2 1‬‬ ‫‪3 9 1‬‬ ‫‪3 −2 9‬‬
‫‪−5 2 −2‬‬ ‫‪1 −5 −2‬‬ ‫‪1 2 −5‬‬
‫‪−2 1 −4‬‬ ‫‪−23‬‬ ‫‪1 −2 −4‬‬ ‫‪69‬‬ ‫‪1 1 −2‬‬ ‫‪0‬‬
‫= ‪x1‬‬ ‫‪3 −2 1‬‬ ‫=‬ ‫= ‪= 1, x2‬‬ ‫‪3 −2 1‬‬ ‫=‬ ‫= ‪= −3, x3‬‬ ‫‪3 −2 1‬‬ ‫=‬ ‫‪=0‬‬
‫‪1 2 −2‬‬ ‫‪−23‬‬ ‫‪1 2 −2‬‬ ‫‪−23‬‬ ‫‪1 2 −2‬‬ ‫‪−23‬‬
‫‪1 1 −4‬‬ ‫‪1 1 −4‬‬ ‫‪1 1 −4‬‬

‫■ ﻣﻌﺎﻳﺐ ﺭﻭﺵﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﻛﺮﺍﻣﺮ‬

‫ﺭﻭﺵﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﻛﺮﺍﻣﺮ ﺑﻪ ﻟﺤﺎﻅ ﺗﺌﻮﺭﻱ ﺟﺬﺍﺏ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﻨﺪ ﻭ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺁﻥﻫﺎ ﺑﺮﺍﻱ ﺍﺛﺒﺎﺕ ﻗﻀﺎﻳﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬
‫ﺍﻣﺎ ﺑﻪ ﺩﻻﻳﻞ ﺯﻳﺮ‪ ،‬ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺭ ﻋﻤﻞ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺑﻪ ﻛﺎﺭ ﻧﻤﻲﺭﻭﻧﺪ‪:‬‬

‫• ﺭﻭﺵﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻭ ﻛﺮﺍﻣﺮ ﺩﺍﺭﺍﻱ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺑﺎﻻ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻳﻦ ﺭﻭﺵﻫﺎ‪ ،‬ﺑﻪ ﺗﺮﺗﻴﺐ‪ ،‬ﺑﻪ ﻣﺤﺎﺳﺒﻪ ﻣﺎﺗﺮﻳﺲ‬
‫ﻭﺍﺭﻭﻥ ﻭ ﺩﺗﺮﻣﻴﻨﺎﻥ ‪ n + 1‬ﻣﺎﺗﺮﻳﺲ ﻧﻴﺎﺯ ﺩﺍﺭﻧﺪ ﻛﻪ ﺑﻪ ﻟﺤﺎﻅ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻣﺴﺘﻠﺰﻡ ﺯﻣﺎﻥ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺑﺎﻻ )ﺧﺼﻮﺻﺎ ﺑﺮﺍﻱ‬
‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺰﺭﮒ( ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬

‫ﻣﻨﺎﺳﺐ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﻣﻤﻜﻦ ﺍﺳﺖ ﺣﻞ‬ ‫ﭘﺎﻳﺪﺍﺭﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ‬ ‫• ﺍﻳﻦ ﺭﻭﺵﻫﺎ )ﺧﺼﻮﺻﺎ ﺭﻭﺵ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ( ﺑﻪ ﻟﺤﺎﻅ‬
‫ﺩﺳﺘﮕﺎﻩﻫﺎ ﺑﺎ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺭ ﻣﺎﺷﻴﻦ‪ ،‬ﺑﻪ ﻧﺘﺎﻳﺞ ﻧﺎﺩﻗﻴﻘﻲ ﻣﻨﺠﺮ ﮔﺮﺩﺩ‪.‬‬

‫• ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻓﻘﻂ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﻗﺎﺑﻞ ﺍﻋﻤﺎﻝ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺭ‬
‫ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻗﺎﺩﺭ ﻧﻴﺴﺘﻨﺪ ﻛﻪ ﺗﺸﺨﻴﺺ ﺩﻫﻨﺪ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ ﻭ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ‬
‫ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ‪ ،‬ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻗﺎﺩﺭ ﺑﻪ ﻳﺎﻓﺘﻦ ﺟﻮﺍﺏﻫﺎ ﻧﻤﻲﺑﺎﺷﻨﺪ‪.‬‬

‫ﺭﻭﺵ ﮔﻮﺱ‬ ‫‪۳.۴‬‬

‫ﺍﺳﺎﺱ ﺭﻭﺵ ﮔﻮﺱ‪ ،‬ﺗﺒﺪﻳﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺩﺭ ﺍﺑﺘﺪﺍ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﻄﺮﺡ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻭ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻗﺒﻼ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺩﺭ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺍﮔﺮ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻳﻚ ﺩﺳﺘﮕﺎﻩ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ ﮔﻮﻳﻴﻢ‪ .‬ﻓﺮﻡ ﻛﻠﻲ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪u11 x1 + u12 x2 + · · · + u1n xn‬‬ ‫‪= b1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪u22 x2 + · · · + u2n xn‬‬ ‫‪= b2‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪unn xn‬‬ ‫‪= bn‬‬
‫ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻧﻴﺰ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬
‫‪‬‬
‫‪u11‬‬ ‫‪u12‬‬ ‫‪u13‬‬ ‫‪...‬‬ ‫‪u1n‬‬ ‫‪x1‬‬ ‫‪b1‬‬
‫‪‬‬ ‫‪   ‬‬
‫‪0‬‬ ‫‪u22‬‬ ‫‪u23‬‬ ‫‪...‬‬ ‫‪u2n   x2   b2 ‬‬
‫‪‬‬ ‫‪   ‬‬
‫‪0‬‬ ‫‪u3n ‬‬ ‫‪   ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪u33‬‬ ‫‪...‬‬ ‫‪  x 3  =  b3 ‬‬
‫‪.‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..  ‬‬
‫‪‬‬ ‫‪  ‬‬
‫‪.‬‬ ‫‪..‬‬ ‫‪ .  .‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.   ..   .. ‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪. . . unn‬‬ ‫‪xn‬‬ ‫‪bn‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۸۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺧﺼﻮﺻﻴﺖ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺜﻠﺜﻲ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺳﺎﺩﮔﻲ ﺣﻞ ﻣﻲ ﺷﻮﻧﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ‪:‬‬

‫• ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺁﺧﺮ‪ ،‬ﺁﺧﺮﻳﻦ ﻣﺠﻬﻮﻝ ﻳﻌﻨﻲ ‪ xn‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪:‬‬
‫‪bn‬‬
‫= ‪xn‬‬
‫‪unn‬‬

‫• ﻭﻗﺘﻲ ‪ xn‬ﺑﻪ ﺩﺳﺖ ﺁﻣﺪ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﻳﻜﻲ ﺑﻪ ﺁﺧﺮ‪ xn-1 ،‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪:‬‬
‫‪bn-1 − un-1,n xn‬‬
‫= ‪x n- 1‬‬
‫‪un-1,n-1‬‬

‫• ﺑﻪ ﻃﻮﺭ ﻛﻠﻲ‪ ،‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ‪ xi+1 ،. . . ،xn-1 ،xn‬ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ‪i‬ﺍﻡ ﻣﻘﺪﺍﺭ ‪ xi‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ‬
‫ﺩﺳﺖ ﻣﻲﺁﻭﺭﻳﻢ‪:‬‬
‫‪bi − ui,i+1 xi+1 − ui,i+2 xi+2 − · · · − uin xn‬‬
‫= ‪xi‬‬ ‫‪.‬‬
‫‪uii‬‬

‫ﺑﻪ ﻫﻤﻴﻦ ﺗﺮﺗﻴﺐ ﻛﻠﻴﻪ ﻣﺘﻐﻴﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ‪ ،‬ﺍﺯ ﺁﺧﺮ ﺑﻪ ﺍﻭﻝ‪ ،‬ﻭ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱﻫﺎﻱ ﺳﺎﺩﻩ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ‪ .‬ﺍﻳﻦ ﺭﻭﺍﻝ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﻪ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪ ۵۲‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺩﺭ ﺯﻳﺮ ﺷﻤﺎﻱ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬
‫‪‬‬ ‫‪bn‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪xn‬‬
‫‪‬‬
‫‪‬‬ ‫‪unn‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪bn-1 − un-1,n xn‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪x n- 1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪un-1,n−1‬‬
‫‪‬‬
‫‪‬‬
‫‪u11 u12 . . . u1,n-1‬‬ ‫‪u1n‬‬ ‫‪b1‬‬ ‫‪‬‬
‫‪‬‬ ‫‪..‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪0 u22 . . . u2,n-1‬‬ ‫‪u2n‬‬ ‫‪b2 ‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪ ‬ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪‬‬
‫‪‬‬
‫‪‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪=======⇒ xk = k‬‬
‫‪b − uk,k+1 xk+1 − uk,k+1 xk+2 − · · · − ukn xn‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪ukk‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0 . . . un-1,n-1 un-1,n‬‬ ‫‪b n- 1 ‬‬ ‫‪‬‬
‫‪‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬
‫‪0‬‬ ‫‪0 ...‬‬ ‫‪0‬‬ ‫‪unn‬‬ ‫‪bn‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪b2 − u23 x3 − u24 x4 − · · · − u2n xn‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪x2‬‬
‫‪‬‬
‫‪‬‬ ‫‪u22‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪b − u12 x2 − u13 x3 − · · · − u1n xn‬‬
‫‪ x1 = 1‬‬
‫‪u11‬‬

‫ﻣﺜﺎﻝ ‪)۵−۴‬ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜ ﺑﺎ ﺟﺎﯾ ﺬﺍﺭﯼ ﭘﺲﺭﻭ(‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪2x + 4x2 − 2x3 = 2‬‬
‫‪ 1‬‬
‫‪x2 + x3 = 4‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪4x3 = 8‬‬

‫‪52 Backward‬‬ ‫‪substitution‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۸۸‬‬

‫ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪‬‬ ‫‪   ‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪−2 x1‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪   ‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪   ‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪  x 2  =  4‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪4‬‬ ‫‪x3‬‬ ‫‪8‬‬
‫ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪8‬‬
‫= ‪ : x3‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺳﻮﻡ‬ ‫‪= 2,‬‬
‫‪4‬‬
‫‪4 − x3‬‬ ‫‪4−2‬‬
‫= ‪ : x2‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ‬ ‫=‬ ‫‪= 2,‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪2 − 4x2 + 2x3‬‬ ‫)‪2 − 4(2) + 2(2‬‬
‫= ‪ : x1‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ‬ ‫=‬ ‫‪= −1.‬‬
‫‪2‬‬ ‫‪2‬‬

‫ﺍﻛﻨﻮﻥ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺑﺎﺷﺪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ‪:‬‬

‫» ﺁﻳﺎ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻫﻤﻮﺍﺭﻩ ﻣﻲﺗﻮﺍﻧﺪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺜﻠﺜﻲ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻛﻨﺪ؟ «‬

‫ﺭﻭﺍﻝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ‪ ،‬ﻳﻚ ﺭﻭﺍﻝ ﺑﺎﺯﮔﺸﺘﻲ ﻭ ﺳﺎﺩﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺳﺎﺩﮔﻲ ﻣﻲﺗﻮﺍﻥ ﺩﺭﻳﺎﻓﺖ ﻛﻪ ﺍﮔﺮ ﺗﻤﺎﻣﻲ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ‬
‫ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻏﻴﺮﺻﻔﺮ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺣﺘﻤﺎ ﺧﺎﺗﻤﻪ ﻣﻲﻳﺎﺑﺪ ﻭ ﻳﻚ ﺟﻮﺍﺏ ﺍﺭﺍﻳﻪ ﻣﻲﺩﻫﺪ‪ .‬ﺍﺯ ﺳﻮﻱ ﺩﻳﮕﺮ‪،‬‬
‫ﺍﮔﺮ ﻛﻠﻴﻪ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻏﻴﺮﺻﻔﺮ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻧﺎﻣﻨﻔﺮﺩ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ‬
‫ﻓﺮﺩ ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺍﮔﺮ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺟﻮﺍﺏ ﺁﻥ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﺩ‪.‬‬

‫ﺣﺎﻝ ﺍﮔﺮ ﻳﻜﻲ ﺍﺯ ﻋﻨﺎﺻﺮ ﺭﻭﻱ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺻﻔﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ )ﺑﻪ ﺷﻜﻠﻲ ﻛﻪ ﺩﺭ ﺑﺎﻻ ﺗﻮﺿﻴﺢ‬
‫ﺩﺍﺩﻩ ﺷﺪ( ﺑﻪ ﺷﻜﺴﺖ ﻣﻲﺍﻧﺠﺎﻣﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺩﺭ ﺧﻼﻝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ‪ ،‬ﺍﻋﻀﺎﻱ ﻗﻄﺮﻱ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺭ ﻣﺨﺮﺝ ﻛﺴﺮ ﻇﺎﻫﺮ‬
‫ﻣﻲﺷﻮﻧﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺍﺻﻼ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻣﺎ ﻧﻜﺘﻪ ﻗﺎﺑﻞ‬
‫ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﺎ ﺗﻐﻴﻴﺮ ﻛﻮﭼﻜﻲ ﺩﺭ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﻣﻲﺗﻮﺍﻥ ﺁﻥ ﺭﺍ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺜﻠﺜﻲ ﻣﻨﻔﺮﺩ ﻧﻴﺰ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬
‫ﺍﻳﻦ ﻧﻜﺘﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺭ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺜﻠﺜﻲ ﻣﻨﻔﺮﺩ‬

‫ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﻨﻔﺮﺩ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺣﺪﺍﻗﻞ ﻳﻜﻲ ﺍﺯ ‪uii‬ﻫﺎ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫ﻣﺜﺎﻝ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ ukk‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ ﻧﻤﻲﺗﻮﺍﻥ ‪ xk‬ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ‬
‫ﻓﻘﻂ ﺷﺎﻣﻞ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪ xk+1‬ﺗﺎ ‪ xn‬ﺍﺳﺖ ﻛﻪ ﻗﺒﻼ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩﺍﻧﺪ‪ .‬ﺣﺎﻝ ﺍﮔﺮ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪ xk+1‬ﺗﺎ ‪ xn‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ ﺻﺪﻕ‬
‫ﻛﻨﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺯﺍﻳﺪ ﺍﺳﺖ ﻭ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ‬
‫ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪ xk+1‬ﺗﺎ ‪ xn‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ ﺻﺪﻕ ﻧﻜﺮﺩﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﻌﺎﺩﻟﻪ ‪k‬ﺍﻡ ﻧﺎﺳﺎﺯﮔﺎﺭ‬
‫ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪ .‬ﻣﺜﺎﻝﻫﺎﻱ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۸۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻣﺜﺎﻝ ‪)۶−۴‬ﺟﺎﯾ ﺬﺍﺭﯼ ﭘﺲﺭﻭ ﺭﻭﯼ ﺩﺳﺘﮕﺎﻫ ﺑﺎﻻﻣﺜﻠﺜ ﺑﺪﻭﻥ ﺟﻮﺍﺏ(‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬‫‪ ‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪x1‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪   ‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪x2  2‬‬
‫‪‬‬ ‫‪  =  ‬‬
‫‪0‬‬ ‫‪2‬‬ ‫‪   ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪ x3  5‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪x4‬‬ ‫‪2‬‬
‫ﺑﺎ ﺍﻋﻤﺎﻝ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪2‬‬
‫‪ : x4 = = 1,‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﭼﻬﺎﺭﻡ‬
‫‪2‬‬
‫‪5 − 2x4‬‬ ‫)‪5 − 2(1‬‬
‫= ‪ : x3‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺳﻮﻡ‬ ‫=‬ ‫‪= 3,‬‬
‫‪1‬‬ ‫‪1‬‬

‫ﺍﻣﺎ ﺿﺮﻳﺐ ‪ x2‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﻓﻘﻂ ﺷﺎﻣﻞ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪ x4‬ﻭ ‪ x3‬ﺍﺳﺖ‪،‬ﻛﻪ ﻣﻄﺎﺑﻖ ﻓﻮﻕ ﻣﻘﺪﺍﺭﺷﺎﻥ‬
‫ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﻣﻘﺪﺍﺭ ‪ x4‬ﻭ ‪ x3‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪ : 2x3 + x4 = 2 =⇒ 2(3) + 1 = 2 =⇒ 7 = 2 E‬ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ‬
‫ﺗﻨﺎﻗﺾ ﺑﺎﻻ ﻧﺸﺎﻥ ﻣﻲﺩﻫﺪ ﻛﻪ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺑﺎ ﻣﻌﺎﺩﻟﻪ ﺳﻮﻡ ﻭ ﭼﻬﺎﺭﻡ ﻧﺎﺳﺎﺯﮔﺎﺭ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۷−۴‬ﺟﺎﯾ ﺬﺍﺭﯼ ﭘﺲﺭﻭ ﺭﻭﯼ ﺩﺳﺘﮕﺎﻫ ﺑﺎﻻﻣﺜﻠﺜ ﺑﺎ ﺑﯽﻧﻬﺎﯾﺖ ﺟﻮﺍﺏ(‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪x1‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪   ‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪   ‬‬
‫‪‬‬ ‫‪ x2  = 7‬‬
‫‪0‬‬ ‫‪2 ‬‬
‫‪‬‬ ‫‪  ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪x3  5‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪x4‬‬ ‫‪2‬‬
‫ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﻣﺜﺎﻝ ‪ ۷-۴‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﻣﻮﻟﻔﻪ ﺩﻭﻡ ﺑﺮﺩﺍﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﺍﺯ ‪ 2‬ﺑﻪ ‪ 7‬ﺗﻐﻴﻴﺮ ﻳﺎﻓﺘﻪ‪ .‬ﻫﻤﺎﻧﻨﺪ‬
‫ﻣﺜﺎﻝ ‪ ۷-۴‬ﺍﺯ ﻣﻌﺎﺩﻻﺕ ﭼﻬﺎﺭﻡ ﻭ ﺳﻮﻡ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪ x4 = 1 :‬ﻭ ‪ .x3 = 3‬ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﻣﻘﺪﺍﺭ ‪ x4‬ﻭ ‪ x3‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ‬
‫ﺩﺍﺭﻳﻢ‪:‬‬
‫‪ : 2x3 + x4 = 7 =⇒ 2(3) + 1 = 7 =⇒ 7 = 7.‬ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﻫﻤﻮﺍﺭﻩ ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺯﺍﻳﺪ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺑﻪ ﻣﺎ ﻣﻲﮔﻮﻳﺪ ﻛﻪ‬
‫ﻣﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ‪ x2‬ﺭﺍ ﺩﻟﺨﻮﺍﻩ ﺍﻧﺘﺨﺎﺏ ﻛﺮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪ : 0x2 + 2x3 + x4 = 7 =⇒ 0x2 + 2(3) + 1 = 7 =⇒ 0x2 + 7 = 7.‬ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ‬
‫ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﻣﻘﺪﺍﺭ ‪ x2‬ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﻫﻨﻮﺯ ﻛﺎﺭ ﺑﻪ ﭘﺎﻳﺎﻥ ﻧﺮﺳﻴﺪﻩ ﺍﺳﺖ‪ .‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪1 − 2x2 − 4x3 − x4‬‬ ‫‪1 − 2x2 − 4(3) − 1‬‬
‫= ‪ : x1‬ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ‬ ‫=‬ ‫‪= −x2 − 6.‬‬
‫‪2‬‬ ‫‪2‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۹۰‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ x1‬ﻧﻴﺰ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﺩ ﻭ ﺍﻟﺒﺘﻪ ﺑﺮﺣﺴﺐ ‪ x2‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ‬
‫ﺩﻟﺨﻮﺍﻩ ﺑﻮﺩﻥ ‪ x2‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻭ ﻓﺮﻡ ﻛﻠﻲ ﺟﻮﺍﺏﻫﺎﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪−x2 − 6‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪x2‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ x2‬ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺩﻟﺨﻮﺍﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺯﻳﺮ ﺗﻌﺪﺍﻱ ﺍﺯ ﺟﻮﺍﺏﻫﺎﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪        ‬‬
‫‪−6‬‬ ‫‪−7‬‬ ‫‪−5‬‬ ‫‪0‬‬ ‫‪−9‬‬
‫‪         ‬‬
‫‪ 0  1 −1 −6  3‬‬
‫‪ , , , , ,...‬‬
‫‪ 3  3  3  3  3‬‬
‫‪         ‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬

‫ﺩﺭ ﭘﺎﻳﺎﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺟﻤﻊﺑﻨﺪﻱ ﺗﺎﻛﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎ ﻛﻤﻚ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ ﻛﻪ ﺁﻳﺎ ﺩﺳﺘﮕﺎﻩ‬
‫ﻣﺜﻠﺜﻲ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺩﺍﺭﺩ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺍﺻﻼ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺣﺎﻟﺖ ﻭﺟﻮﺩ ﺟﻮﺍﺏ‪ ،‬ﺍﻳﻦ ﺭﻭﺵ‬
‫ﻗﺎﺩﺭ ﺍﺳﺖ ﺟﻮﺍﺏ ﻭ ﻳﺎ ﺟﻮﺍﺏﻫﺎﻱ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻛﻨﺪ‪.‬‬

‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ‬ ‫�‬

‫ﺑﻌﺪ ﺍﺯ ﻣﻌﺮﻓﻲ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺜﻠﺜﻲ‪ ،‬ﺑﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻳﺎ ﺑﻪ ﺍﺧﺘﺼﺎﺭ‬
‫ﺭﻭﺵ ﮔﻮﺱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺑﺮﻣﻲﮔﺮﺩﻳﻢ‪ .‬ﺍﻳﺪﻩ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺴﻴﺎﺭ ﺳﺎﺩﻩ ﻭ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‪ ،‬ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺑﺎ ﻛﻤﻚ ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﺩﺳﺘﮕﺎﻩ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺣﻞ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺗﻐﻴﻴﺮ ﻧﻤﻲﺩﻫﺪ‪ .‬ﺷﻤﺎﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺍﺳﺖ‪:‬‬
‫ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‬ ‫ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ‬
‫‪Ax = b‬‬ ‫→‪−−−−−−−−−‬‬ ‫̄‪Ux = b‬‬ ‫ﺣﻞ →‪−−−−−−−‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺟﺎﻳﮕﺰﻳﻦ ﻣﻨﺎﺳﺒﻲ ﺑﺮﺍﻱ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺳﺘﮕﺎﻩ ﺍﺳﺖ )ﺑﻪ ﺧﺼﻮﺹ ﺩﺭ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ‬
‫ﻛﺎﻣﭙﻴﻮﺗﺮﻱ(‪ .‬ﻟﺬﺍ ﻏﺎﻟﺒﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﻭﻱ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺍﻋﻤﺎﻝ ﻣﻲﺷﻮﺩ‪ .‬ﺷﻤﺎﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﻭﻱ ﻣﺎﺗﺮﻳﺲ‬
‫ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪b1‬‬ ‫‪u11‬‬ ‫‪u12‬‬ ‫···‬ ‫‪u1n‬‬ ‫‪b̄1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪b2 ‬‬ ‫‪‬‬ ‫···‬ ‫‪b̄2 ‬‬
‫‪‬‬ ‫‪ 0‬ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪‬‬ ‫‪u22‬‬ ‫‪u2n‬‬ ‫ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ‪‬‬
‫‪ .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪−−−−−−−−−→  .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫ﺟﻮﺍﺏ →‪−−−−−−−‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪bn‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪· · · unn‬‬ ‫‪b̄n‬‬
‫ﻭ ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‬
‫‪h‬‬ ‫‪i‬‬ ‫ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‬
‫‪h‬‬ ‫‪i‬‬ ‫ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ‬
‫‪A‬‬ ‫‪b‬‬ ‫→‪−−−−−−−−−‬‬ ‫‪U‬‬ ‫̄‪b‬‬ ‫ﺣﻞ →‪−−−−−−−‬‬
‫ﺍﻛﻨﻮﻥ ﺳﻮﺍﻻﺕ ﺯﻳﺮ ﺩﺭ ﻣﻮﺭﺩ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻣﻄﺮﺡ ﻣﻲﮔﺮﺩﺩ‪:‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪ .۱‬ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﭼﮕﻮﻧﻪ ﻭ ﺑﻪ ﭼﻪ ﺗﺮﺗﻴﺒﻲ ﺑﺎﺷﺪ؟‬

‫‪ .۲‬ﺁﻳﺎ ﺑﺎ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﻫﻤﻮﺍﺭﻩ ﻣﻲﺗﻮﺍﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﻌﺎﺩﻝ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ؟‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻻﺕ ﭘﺎﺳﺦ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬

‫ﺩﺳﺘﮕﺎﻩ ‪ n‬ﻣﺘﻐﻴﺮﻩ ‪ Ax = b‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺷﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪b1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪b2 ‬‬
‫‪‬‬ ‫‪‬‬
‫‪ .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪bn‬‬
‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺭ ‪ n − 1‬ﻣﺮﺣﻠﻪ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺩﺭ ﻣﺮﺣﻠﻪ ‪i‬ﺍﻡ ﻋﻨﺎﺻﺮ‬
‫ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ‪i‬ﺍﻡ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﺪ ﻭ ﻟﺬﺍ ﺑﻌﺪ ﺍﺯ ‪ n − 1‬ﻣﺮﺣﻠﻪ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ‬
‫ﺍﻳﻦ ﻣﺮﺍﺣﻞ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ ،‬ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﻋﻤﺎﻝ ﻣﻲﺷﻮﻧﺪ ﻛﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﺻﻔﺮ‬
‫ﺷﻮﻧﺪ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪ ،‬ﻣﻮﻟﻔﻪ )‪(1, 1‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ‪ ۵۳‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﺑﺎ ﻛﻤﻚ ﺁﻥ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺳﺘﻮﻥ‬
‫ﺍﻭﻝ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺍﻟﺒﺘﻪ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻣﺤﻮﺭ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﮔﺮ ﻣﻮﻟﻔﻪ )‪ (1, 1‬ﺻﻔﺮ ﺑﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﺩﻭ‬
‫ﺳﻄﺮ )ﺩﻭ ﻣﻌﺎﺩﻟﻪ(‪ ،‬ﺳﻌﻲ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﻋﺪﺩﻱ ﻏﻴﺮ ﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ )‪ (1, 1‬ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻋﻤﻞ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫• ﺍﮔﺮ ‪ ،a11 = 0‬ﺁﻥﮔﺎﻩ ﺳﻄﺮ ﺍﻭﻝ ﺭﺍ ﺑﺎ ﺳﻄﺮ ﺩﻳﮕﺮﻱ ﺟﺎﺑﺠﺎ ﻣﻲﻛﻨﻴﻢ ﺗﺎ ﺍﻳﻦ ﻛﻪ ﻋﺪﺩﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ )‪(1, 1‬‬
‫ﻣﺎﺗﺮﻳﺲ ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪ .‬ﺍﻳﻦ ﻛﺎﺭ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ‪ ۵۴‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﻣﺤﻮﺭﮔﻴﺮﻱ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺟﺎﺑﺠﺎﻳﻲ‬
‫ﺩﻭ ﺳﻄﺮ )ﻣﻌﺎﺩﻟﻪ( ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻋﻨﺼﺮﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪.‬‬

‫• ﺑﻪ ﺍﺯﺍﻱ ‪ ،i = 2, . . . , n‬ﺑﺮﺍﻱ ﺻﻔﺮ ﺷﺪﻥ ﻣﻮﻗﻌﻴﺖ )‪ ،(i, 1‬ﻣﻲﺗﻮﺍﻥ ‪ − aai1ii‬ﺑﺮﺍﺑﺮ ﺳﻄﺮ ﺍﻭﻝ ﺭﺍ ﺑﺎ ﺳﻄﺮ ‪i‬ﺍﻡ ﺟﻤﻊ‬
‫ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺑﻌﺪ ﺍﺯ ﭘﺎﻳﺎﻥ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫)‪(1‬‬
‫‪a11‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫)‪(1‬‬
‫‪a13‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫)‪(1‬‬
‫‪a1n‬‬ ‫‪b1‬‬‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫)‪(1‬‬
‫‪a22‬‬ ‫)‪(1‬‬
‫‪a23‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a2,n−1‬‬ ‫)‪(1‬‬
‫‪a2n‬‬ ‫‪b2 ‬‬
‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫)‪(1‬‬
‫‪a32‬‬ ‫)‪(1‬‬
‫‪a33‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a3,n−1‬‬ ‫)‪(1‬‬
‫‪a3n‬‬ ‫‪b3 ‬‬
‫)‪(1‬‬

‫‪ : ‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻌﺪ ﺍﺯ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‬

‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬
‫‪‬‬ ‫···‬ ‫‪bn−1 ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪an−1,2‬‬ ‫‪an−1,3‬‬ ‫‪an−1,n−1‬‬ ‫‪an−1,n‬‬ ‫‪‬‬
‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬

‫‪0‬‬ ‫)‪(1‬‬
‫‪an2‬‬ ‫)‪(1‬‬
‫‪an3‬‬ ‫···‬ ‫)‪(1‬‬
‫‪an,n−1‬‬ ‫‪ann‬‬
‫)‪(1‬‬
‫‪bn‬‬
‫)‪(1‬‬

‫ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﻧﻴﺰ ﻫﻤﺎﻧﻨﺪ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﻋﻤﻞ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﻣﻮﻟﻔﻪ )‪ (2, 2‬ﺍﺯ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺩﺭ‬
‫ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺑﺎ ﻛﻤﻚ ﺁﻥ ﻋﻨﺎﺻﺮ ﺯﻳﺮﺵ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺍﮔﺮ ﻣﻮﻟﻔﻪﻱ )‪(2, 2‬‬
‫‪53 Pivot‬‬ ‫‪54 Pivoting‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۹۲‬‬

‫ﺻﻔﺮ ﺑﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﺩﺭ ﺻﻮﺭﺕ ﻧﻴﺎﺯ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮ ﺩﻭﻡ ﺑﺎ ﻳﻚ ﺳﻄﺮ ﺑﻌﺪ ﺁﻥ‬
‫)‪(1‬‬

‫)ﻣﺤﻮﺭﮔﻴﺮﻱ( ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺑﻪ ﺍﺯﺍﻱ ‪ − aai2 ،i = 2, . . . , n‬ﺑﺮﺍﺑﺮ ﺳﻄﺮ ﺩﻭﻡ ﺭﺍ ﺑﺎ ﺳﻄﺮ ‪i‬ﺍﻡ ﺟﻤﻊ ﻣﻲﻛﻨﻴﻢ ﻭ ﺩﺭ ﺳﻄﺮ‬
‫)‪(1‬‬
‫‪22‬‬
‫‪i‬ﺍﻡ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺩﺭ ﭘﺎﻳﺎﻥ ﺳﻄﺮ ﺩﻭﻡ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫)‪(1‬‬
‫‪a11‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫)‪(1‬‬
‫‪a13‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫)‪(1‬‬
‫‪a1n‬‬ ‫‪b1‬‬‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫)‪(2‬‬
‫‪a22‬‬ ‫)‪(2‬‬
‫‪a23‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a2,n−1‬‬ ‫)‪(2‬‬
‫‪a2n‬‬ ‫‪b2 ‬‬
‫)‪(2‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫)‪(2‬‬
‫‪a33‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a3,n−1‬‬ ‫)‪(2‬‬
‫‪a3n‬‬ ‫‪b3 ‬‬
‫)‪(2‬‬

‫‪ : ‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻌﺪ ﺍﺯ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‬

‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬
‫‪‬‬ ‫···‬ ‫‪bn−1 ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪an−1,3‬‬ ‫‪an−1,n−1‬‬ ‫‪an−1,n‬‬ ‫‪‬‬
‫)‪(2‬‬ ‫)‪(2‬‬ ‫)‪(2‬‬ ‫)‪(2‬‬

‫‪0‬‬ ‫‪0‬‬ ‫)‪(2‬‬

‫‪an3‬‬ ‫···‬ ‫)‪(2‬‬
‫‪an,n−1‬‬ ‫‪ann‬‬
‫)‪(2‬‬
‫‪bn‬‬
‫)‪(2‬‬

‫ﺍﻟﺒﺘﻪ ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﺍﺑﺘﺪﺍﻱ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﻣﺤﻮﺭ ﻭ ﺗﻤﺎﻧﻲ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﺻﻔﺮ ﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ ﻻﺯﻡ‬
‫ﻧﻴﺴﺖ ﻛﻪ ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﻛﺎﺭﻱ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻣﺤﻮﺭ ﺻﻔﺮ ﻫﺴﺘﻨﺪ‪.‬‬

‫ﻣﺮﺣﻠﻪ ‪k‬ﺍﻡ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ‪k‬ﺍﻡ ﻛﻪ ‪ ،1 ≤ k ≤ n − 1‬ﻣﺮﺍﺣﻞ ﺯﻳﺮ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ :‬ﺩﺭ ﺻﻮﺭﺕ ﻟﺰﻭﻡ‪ ،‬ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮ ‪ k‬ﺑﺎ ﺳﻄﺮﻱ‬
‫)‪(k−1‬‬

‫ﺑﻌﺪ ﺍﺯ ﺁﻥ ﺟﺎﺑﺠﺎ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﺯﺍﻱ ‪ ،i = k + 1, . . . , n‬ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ‪ ،aik ̸= 0‬ﺁﻥﮔﺎﻩ ‪ − aaik‬ﺑﺮﺍﺑﺮ ﺳﻄﺮ ‪k‬ﺍﻡ ﺭﺍ ﺑﺎ‬
‫)‪(k−1‬‬
‫)‪(k−1‬‬

‫‪kk‬‬
‫ﺳﻄﺮ ‪i‬ﺍﻡ ﺟﻤﻊ ﻣﻲﻛﻨﻴﻢ ﻭ ﺣﺎﺻﻞ ﺭﺍ ﺩﺭ ﺳﻄﺮ ‪i‬ﺍﻡ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫ﺑﻌﺪ ﺍﺯ ﻣﺮﺣﻠﻪ ‪n − 1‬ﺍﻡ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﻌﺎﺩﻝ‪ ،‬ﺑﻌﺪ ﺍﺯ ﻣﺮﺣﻠﻪ ‪n − 1‬ﺍﻡ‪ ،‬ﻣﺎﺗﺮﻳﺲ‬
‫ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫)‪(1‬‬
‫‪a11‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫)‪(1‬‬
‫‪a13‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫‪a1n‬‬ ‫)‪(1‬‬
‫‪b1‬‬ ‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫)‪(2‬‬
‫‪a22‬‬ ‫)‪(2‬‬
‫‪a23‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a2,n−1‬‬ ‫‪a2n‬‬ ‫)‪(2‬‬
‫‪b2 ‬‬ ‫)‪(2‬‬

‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫)‪(3‬‬
‫‪a33‬‬ ‫···‬ ‫)‪(3‬‬
‫‪a3,n−1‬‬ ‫‪a3n‬‬ ‫)‪(3‬‬
‫‪b3 ‬‬ ‫)‪(3‬‬

‫‪ : ‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻌﺪ ﺍﺯ ﻣﺮﺣﻠﻪ ‪n − 1‬ﺍﻡ‬

‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫···‬ ‫‪bn−1 ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪an−1,n−1‬‬ ‫‪an−1,n‬‬ ‫‪‬‬
‫)‪(n−1‬‬ ‫)‪(n−1‬‬ ‫)‪(n−1‬‬

‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪ann‬‬ ‫)‪(n−1‬‬

‫‪bn‬‬‫)‪(n−1‬‬

‫ﻭ ﺳﭙﺲ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺑﻪ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﺜﻠﺜﻲ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ‪ ،‬ﺑﺎ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ‪ ،‬ﺑﻪ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ(‬ ‫ﻣﺜﺎﻝ ‪)۸−۴‬ﺣﻞ ﯾ‬
‫ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x1 + 3x2 + 2x3 + 7x4 = −9‬‬
‫‪‬‬
‫‪‬‬
‫‪ 2x + 4x + 3x − 5x = 18‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬
‫‪‬‬
‫‪‬‬ ‫‪−x‬‬ ‫‪+‬‬ ‫‪3x‬‬ ‫‪+‬‬ ‫‪2x‬‬ ‫‪−‬‬ ‫‪4x‬‬ ‫‪4 = 11‬‬
‫‪‬‬
‫‪‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬
‫‪‬‬
‫‪3x1 + x2 + 1x3 + 3x4 = −1‬‬
‫◀ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ‪ a11 ̸= 0‬ﻭ ﻟﺬﺍ ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﻧﻴﺎﺯ ﺑﻪ ﺟﺎﺑﺠﺎﻱ ﺳﻄﺮ ﺍﻭﻝ ﻧﺪﺍﺭﻳﻢ‪ .‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺭﻭﺵ ﺣﺬﻓﻲ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﮔﻮﺱ‪ ،‬ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﻗﻄﺮ ﺳﺘﻮﻥ ﺍﻭﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7 −9‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪−9‬‬ ‫‪7‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 2 4‬‬ ‫‪3 −5 18‬‬ ‫‪ 0‬‬ ‫‪−2 −1 −19 36‬‬
‫‪‬‬ ‫‪ −−R2:=−2R1+R2‬‬
‫‪−−−−−−−−→ ‬‬ ‫‪ −−R3:=1R1+R3‬‬
‫→‪−−−−−−−‬‬
‫‪−1 3‬‬ ‫‪2 −4 11‬‬ ‫‪‬‬ ‫‪−1‬‬ ‫‪−4 11‬‬
‫‪‬‬ ‫‪‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪‬‬
‫‪3 1‬‬ ‫‪1‬‬ ‫‪3 −1‬‬ ‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪3 −1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7 −9‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7 −9‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 0 −2‬‬ ‫‪−1 −19 36‬‬ ‫‪‬‬ ‫‪0 −2 −1 −19 36‬‬
‫‪‬‬ ‫‪ −−R4:=−3R1+R4‬‬
‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬‫→‬ ‫‪‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪3 −2‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪4‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪6‬‬ ‫‪4‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪3 −1‬‬ ‫‪0 −8 −5 −18 26‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ :‬ﺩﺭ ﻣﻮﻗﻌﻴﺖ )‪ (2, 2‬ﻋﺪﺩ ‪ −2‬ﻗﺮﺍﺭ ﮔﺮﻓﺘﻪ ﻭ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺁﻥ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻭ ﺑﺎ‬
‫ﻛﻤﻚ ﺁﻥ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﺭﺍ ﻧﻴﺰ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−9‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−9‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪−2 −1‬‬ ‫‪−19‬‬ ‫‪36‬‬ ‫‪R3:=3R2+R3 ‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−1‬‬ ‫‪−19‬‬ ‫‪36‬‬
‫‪‬‬ ‫‪ −−‬‬‫‪−−−−−−→ ‬‬ ‫‪ −−‬‬
‫‪R4:=−4R2+R4‬‬
‫→‪−−−−−−−‬‬
‫‪ 0‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪−54‬‬ ‫‪‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪4‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪ 0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪110‬‬
‫‪0‬‬ ‫‪−8‬‬ ‫‪−5‬‬ ‫‪−18‬‬ ‫‪26‬‬ ‫‪0‬‬ ‫‪−8‬‬ ‫‪−5‬‬ ‫‪−18‬‬ ‫‪26‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−9‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪−2 −1‬‬ ‫‪−19‬‬ ‫‪36‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪−54‬‬ ‫‪110‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪58‬‬ ‫‪−118‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺳﻮﻡ ﻧﻴﺰ ﻧﻴﺎﺯﻱ ﺑﻪ ﺟﺎﺑﺠﺎﻳﻲ ﺩﻭ ﺳﻄﺮ )ﻣﺤﻮﺭﮔﻴﺮﻱ( ﻧﺪﺍﺭﻳﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻋﻨﺼﺮ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ‬
‫ﺳﻮﻡ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−9‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−9‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪−2‬‬ ‫‪−1‬‬ ‫‪−19‬‬ ‫‪36‬‬ ‫‪R4:=1R3+R4 ‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−1‬‬ ‫‪−19‬‬ ‫‪36‬‬
‫‪‬‬ ‫‪ −−‬‬‫‪−−−−−−→ ‬‬ ‫‪‬‬
‫‪ 0‬‬ ‫‪−54‬‬ ‫‪110‬‬ ‫‪‬‬ ‫‪−54‬‬ ‫‪110‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪58‬‬ ‫‪−118‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪4‬‬ ‫‪−8‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﺎ ﻛﻤﻚ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﺷﺪ ﻭ ﻫﻤﻴﻨﻄﻮﺭ ﺍﻋﻀﺎﻱ ﺭﻭﻱ ﻗﻄﺮ ﺍﺻﻠﻲ‬
‫ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ‪ ،‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻣﻲﺗﻮﺍﻥ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪:‬‬
‫‪‬‬
‫‪‬‬ ‫‪−8‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪x4‬‬ ‫‪= −2,‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪4‬‬
‫‪−9‬‬ ‫‪‬‬
‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪x3 = 110 + 54(−2) = 2,‬‬
‫‪‬‬
‫‪ 0‬‬ ‫‪−2‬‬ ‫‪−1‬‬ ‫‪−19‬‬ ‫‪36‬‬
‫‪‬‬ ‫ﭘﺲ‪−‬ﺭﻭ‪ −−−‬‬ ‫ﺟﺎﻳﮕﺬﺍﺭﻱ‬
‫→‪−−−−−−−‬‬
‫‪1‬‬
‫‪ 0‬‬ ‫‪−54‬‬ ‫‪110‬‬ ‫‪‬‬ ‫‪36‬‬ ‫‪+‬‬ ‫)‪19(−2‬‬ ‫)‪+ 1(2‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪‬‬ ‫= ‪ x2‬‬
‫‪‬‬ ‫‪= 0,‬‬
‫‪‬‬
‫‪‬‬ ‫‪−2‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪4‬‬ ‫‪−8‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫)‪−9 − 7(−2) − 2(2) − 3(0‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪x1‬‬ ‫‪= 1.‬‬
‫‪1‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۹۴‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﺗﻨﻬﺎ ﺑﺎ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E1‬ﺗﻮﺍﻧﺴﺘﻴﻢ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺑﻪ‬
‫ﺣﻞ ﺁﻥ ﺑﭙﺮﺩﺍﺯﻳﻢ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﺑﻌﺪ ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪) E2‬ﻣﺤﻮﺭﮔﻴﺮﻱ( ﻧﻴﺰ ﻣﻮﺭﺩ ﻧﻴﺎﺯ ﻭﺍﻗﻊ‬
‫ﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۹−۴‬ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺑﺎ ﻣﺤﻮﺭﮔﯿﺮﯼ(‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺑﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺣﻞ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1 3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2 6‬‬ ‫‪2‬‬ ‫‪−5‬‬ ‫‪14‬‬
‫‪‬‬ ‫‪‬‬
‫‪−1 3‬‬ ‫‪−4‬‬ ‫‪13‬‬
‫‪‬‬ ‫‪4‬‬ ‫‪‬‬
‫‪1 2‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪0.5‬‬
‫ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﺑﺎ ﺍﻧﺠﺎﻡ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪R2:=−2R1+R2‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2 6 2‬‬ ‫‪−5‬‬ ‫‪14‬‬ ‫‪R3:=+1R1+R3‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬
‫‪‬‬ ‫‪ −−−−‬‬
‫‪R4:=−1R1+R4‬‬
‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬
‫→‬ ‫‪‬‬ ‫‪‬‬
‫‪−1 3 4‬‬ ‫‪−4‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪13‬‬ ‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪1 2 1‬‬ ‫‪3‬‬ ‫‪0.5‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬

‫ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺳﺘﻮﻥ ﺩﻭﻡ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﻣﺎ ﻋﺪﺩ ‪ 0‬ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﺑﺎ ﺍﻳﻦ ﻣﺤﻮﺭ‬
‫ﻧﻤﻲﺗﻮﺍﻥ ﻣﻮﻟﻔﻪﻫﺎﻱ )‪ (3, 2‬ﻭ )‪ (4, 2‬ﺭﺍ ﺻﻔﺮ ﻛﺮﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﻫﺮ ﻋﺪﺩﻱ ﺩﺭ ﺻﻔﺮ ﺿﺮﺏ ﺷﻮﺩ ﺣﺎﺻﻞ ﺻﻔﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪ .‬ﺩﺭ ﭼﻨﻴﻦ‬
‫ﺣﺎﻻﺗﻲ‪ ،‬ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ )ﻋﻤﻞ ‪ ( E2‬ﻳﻚ ﻋﺪﺩ ﻏﻴﺮﺻﻔﺮ ﺭﺍ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮ ﺩﻭﻡ ﺑﺎ ﺳﻄﺮ ﺳﻮﻡ‬
‫ﻭ ﻳﺎ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮ ﺩﻭﻡ ﺑﺎ ﺳﻄﺮ ﭼﻬﺎﺭﻡ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻳﻚ ﻣﺤﻮﺭ ﻏﻴﺮﺻﻔﺮ ﺭﺳﻴﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻧﺒﺎﻳﺪ ﺳﻄﺮ ﺩﻭﻡ ﺭﺍ ﺑﺎ ﺳﻄﺮ ﺍﻭﻝ‬
‫ﺟﺎﺑﺠﺎ ﻛﺮﺩ‪ .‬ﭼﺮﺍ ﻛﻪ ﻫﺮﭼﻨﺪ ﺍﻳﻦ ﻛﺎﺭ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﺤﻮﺭ ﻏﻴﺮﺻﻔﺮ ﺷﻮﺩ‪ ،‬ﺍﻣﺎ ﺩﺭ ﻋﻮﺽ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻋﺪﺩﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ‬
‫ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﻇﺎﻫﺮ ﺷﻮﺩ‪.‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻮﺿﻴﺤﺎﺕ ﻓﻮﻕ‪ ،‬ﺳﻄﺮ ﺩﻭﻡ ﻭ ﭼﻬﺎﺭﻡ ﺭﺍ ﺟﺎﺑﺠﺎ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬ ‫ﺟﺎﺑﺠﺎﻳﻲ ‪ R2‬ﻭ ‪R4‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬
‫‪‬‬ ‫‪ −−−−‬‬‫‪−−−−−−−−→ ‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪‬‬ ‫ﻣﺤﻮﺭﮔﻴﺮﻱ‬ ‫‪0‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪12‬‬ ‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬
‫ﺣﺎﻝ ﻛﻪ ﻋﻨﺼﺮﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﮔﺮﻓﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺭﺍ ﺑﺮﺍﻱ ﺻﻔﺮ ﻛﺮﺩﻥ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ‬
‫ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪R2:=−2R1+R2‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬ ‫‪R3:=+6R2+R3‬‬‫‪‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬
‫‪‬‬ ‫‪ −−−−−−−−−−−−−→ 0‬‬
‫‪R4:= 0R2+R4‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪−21‬‬ ‫‪21‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪12‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻧﻴﺰ ﻋﺪﺩ ‪ 0‬ﺩﻭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﺍﺳﺖ ﻭ ﻧﻴﺎﺯ ﺑﻪ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮﻱ )ﻣﺤﻮﺭﮔﻴﺮﻱ( ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬ ‫ﺟﺎﺑﺠﺎﻳﻲ ‪ R3‬ﻭ ‪R4‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪1.5‬‬
‫‪‬‬ ‫‪ −−−−‬‬‫‪−−−−−−−−→ ‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−21‬‬ ‫‪‬‬ ‫ﻣﺤﻮﺭﮔﻴﺮﻱ‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪21‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−21‬‬ ‫‪21‬‬
‫ﺑﻌﺪ ﺍﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﻲﺷﻮﺩ ﻭ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫‪x4 = −1, x3 = 1.5, x2 = 1, x1 = 0.‬‬

‫ﺗﺎ ﻛﻨﻮﻥ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﺍ ﺭﻭﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻳﻲ ﺍﻋﻤﺎﻝ ﻛﺮﺩﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺁﻥﻫﺎ ﻧﺎﻣﻨﻔﺮﺩ ﺑﻮﺩ ﻭ ﻟﺬﺍ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ‬
‫ﺑﻪ ﻓﺮﺩ ﺩﺍﺷﺘﻨﺪ‪ .‬ﺩﻳﺪﻳﻢ ﻛﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﺧﻮﺑﻲ ﺟﻮﺍﺏ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬
‫ﺭﺍ ﺭﻭﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺎ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻣﻨﻔﺮﺩ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩﻫﺎ ﻳﺎ ﺟﻮﺍﺏ ﻧﺪﺍﺭﻧﺪ ﻭ ﻳﺎ ﺍﻳﻦ‬
‫ﻛﻪ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﻧﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﻧﺎﺳﺎﺯﮔﺎﺭ(‬ ‫ﻣﺜﺎﻝ ‪)۱۰−۴‬ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺭﻭﯼ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪6‬‬ ‫‪2‬‬ ‫‪−5‬‬ ‫‪14‬‬
‫‪‬‬ ‫‪‬‬
‫‪−1‬‬ ‫‪−3 4‬‬ ‫‪−4‬‬ ‫‪13‬‬
‫‪‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪−6 1‬‬ ‫‪3‬‬ ‫‪−2.5‬‬
‫ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﺑﺎ ﺍﻧﺠﺎﻡ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪6‬‬ ‫‪2‬‬ ‫‪−5‬‬ ‫‪14‬‬ ‫‪R3:=+1R1+R3‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬
‫‪‬‬ ‫‪ −−−−‬‬‫‪−−−−−−−−−→ ‬‬
‫‪R4:=+2R1+R4‬‬ ‫‪‬‬
‫‪−1‬‬ ‫‪−3 4‬‬ ‫‪−4‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪13‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪−6 1‬‬ ‫‪3‬‬ ‫‪−2.5‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪5‬‬ ‫‪17‬‬ ‫‪−4.5‬‬

‫ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ :‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻛﻠﻴﻪ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﺍﺯ ﻗﺒﻞ ﺻﻔﺮ ﺷﺪﻩﺍﻧﺪ ﻭ ﻧﻴﺎﺯﻱ ﺑﻪ ﺻﻔﺮ ﻛﺮﺩﻥ ﺁﻥﻫﺎ ﻧﻴﺴﺖ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﻧﻴﺎﺯ ﻧﻴﺴﺖ ﻛﻪ ﻋﻤﻠﻴﺎﺗﻲ ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ‪.‬‬
‫ﺍﻣﺎ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻣﺤﻮﺭ ﻧﻴﺰ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺍﺯ ﺍﻳﻦ ﻛﻪ ﻣﺤﻮﺭ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﺁﻥ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ‬
‫ﮔﺮﻓﺖ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺳﺘﮕﺎﻩ ﺻﻔﺮ ﺍﺳﺖ‪.‬‬
‫ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ‪ ،‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1 3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1 3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪‬‬
‫‪16‬‬ ‫‪5‬‬
‫‪R4:=− 6 R3+R4‬‬ ‫‪‬‬ ‫‪0 0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪16‬‬
‫‪‬‬ ‫‪−−−−−−−−‬‬ ‫‪−−−−−→ ‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪12‬‬ ‫‪0 0‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫‪5‬‬ ‫‪17‬‬ ‫‪−4.5‬‬ ‫‪0 0‬‬ ‫‪0‬‬ ‫‪14.5‬‬ ‫‪−14.5‬‬
‫ﺩﻳﺪﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﺷﺪ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﺑﺎﺭ ﺭﻭﻱ ﻗﻄﺮ ﺍﺻﻠﻲ ﻣﺎﺗﺮﻳﺲ ﻳﻚ ﺻﻔﺮ ﻇﺎﻫﺮ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻛﻪ ﻧﺸﺎﻥ ﺩﻫﻨﺪﻩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﻇﺮﺍﻳﺐ ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻳﺎ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ ﻭ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۹۶‬‬

‫ﺟﻮﺍﺏ ﺩﺍﺭﺩ‪ .‬ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺍﺻﻼ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪.‬‬
‫ﺩﺍﺭﻳﻢ‪:‬‬

‫• ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﭼﻬﺎﺭﻡ ﺩﺍﺭﻳﻢ‪ 14.5x4 = −14.5 :‬ﻭ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ‬

‫‪x4 = −1.‬‬

‫• ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺳﻮﻡ ﺩﺍﺭﻳﻢ‪ ،6x3 + 3x4 = 12 :‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ‬

‫‪12 − 3x4‬‬ ‫)‪12 − 3(−1‬‬
‫= ‪x3‬‬ ‫=‬ ‫‪= 2.5‬‬
‫‪6‬‬ ‫‪6‬‬
‫• ﻣﻘﺪﺍﺭ ‪ x2‬ﺭﺍ ﺑﺎﻳﺪ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﺍﺯ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺩﺍﺭﻳﻢ‪ .0x2 − 2x3 − 19x4 = 16 :‬ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ‬
‫ﺩﺍﺭﻳﻢ‪:‬‬
‫‪0x2 − 2(2.5) − 19(−1) = 16 ⇒ 0x2 + 14 = 16‬‬
‫ﺍﻣﺎ ﺑﻪ ﺍﺯﺍﻱ ﻫﻴﭻ ﻣﻘﺪﺍﺭﻱ ﺍﺯ ‪ x2‬ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺑﺎ ﻣﻌﺎﺩﻻﺕ ﺳﻮﻡ ﻭ ﭼﻬﺎﺭﻡ ﻧﺎﺳﺎﺯﮔﺎﺭ ﺍﺳﺖ‪.‬‬
‫ﻳﻌﻨﻲ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﻓﻮﻕ ﺩﻳﺪﻳﻢ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﻧﺎﺳﺎﺯﮔﺎﺭ ﺗﻮﺳﻂ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﺷﺪ ﻭﻟﻲ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ‬
‫ﭘﺲﺭﻭ ﻣﺘﻮﺟﻪ ﺷﺪﻳﻢ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﺑﻌﺪ ﺩﺳﺘﮕﺎﻫﻲ ﺑﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ‬
‫ﺳﭙﺲ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ‪ ،‬ﺟﻮﺍﺏﻫﺎ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﺑﯽﻧﻬﺎﯾﺖ ﺟﻮﺍﺏ(‬ ‫ﻣﺜﺎﻝ ‪)۱۱−۴‬ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺭﻭﯼ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪6‬‬ ‫‪2‬‬ ‫‪−5‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪‬‬
‫‪−1‬‬ ‫‪−3 4‬‬ ‫‪−4‬‬ ‫‪13‬‬
‫‪‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪−6 1‬‬ ‫‪3‬‬ ‫‪−2.5‬‬
‫ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﻧﻈﺮﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۰-۴‬ﺍﺳﺖ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﺳﻤﺖ ﺭﺍﺳﺖ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺍﺯ ‪ 14‬ﺑﻪ ‪12‬‬
‫ﺗﻐﻴﻴﺮ ﻳﺎﻓﺘﻪ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺗﻐﻴﻴﺮ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺑﺎﺷﺪ‪.‬‬
‫ﻫﻤﺎﻧﻨﺪ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﺜﻠﺜﻲ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬ ‫‪1 3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪−5‬‬ ‫‪12‬‬ ‫‪0 0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪14‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪2‬‬ ‫ﮔﻮﺱ‪ −−−‬‬ ‫ﺭﻭﺵ ﺣﺬﻓﻲ‬
‫‪−−−−−−−−→ ‬‬ ‫‪‬‬
‫‪−1‬‬ ‫‪−3‬‬ ‫‪−4‬‬ ‫‪‬‬ ‫‪0 0‬‬ ‫‪12‬‬
‫‪‬‬ ‫‪4‬‬ ‫‪13‬‬ ‫‪‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪−6‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪−2.5‬‬ ‫‪0 0‬‬ ‫‪0‬‬ ‫‪14.5‬‬ ‫‪−14.5‬‬
‫ﻭﺟﻮﺩ ﺻﻔﺮ ﺭﻭﻱ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻧﺸﺎﻥ ﺍﺯ ﻣﻨﻔﺮﺩ ﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺩ‪ .‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬

‫• ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﭼﻬﺎﺭﻡ ﻭ ﺳﻮﻡ‪ ،‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬

‫‪x4 = −1, x3 = 2.5‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫• ﺑﺮﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩﻥ ‪ x2‬ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ‪ x4‬ﻭ ‪ x3‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪0x2 − 2(2.5) − 19(−1) = 14 ⇒ 0x2 + 14 = 14‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﻣﻘﺪﺍﺭﻱ ﺍﺯ ‪ ،x2‬ﻣﻌﺎﺩﻟﻪ ﻓﻮﻕ )ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ( ﺑﺮﻗﺮﺍﺭ ﻣﻲﺑﺎﺷﺪ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‬
‫ﻣﻲﺗﻮﺍﻥ ‪ x2‬ﺭﺍ ﺩﻟﺨﻮﺍﻩ ﺍﻧﺘﺨﺎﺏ ﻛﺮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺯﺍﻳﺪ ﺍﺳﺖ‪.‬‬

‫• ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ‪ x4‬ﻭ ‪ x3‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ﺩﺍﺭﻳﻢ‪ x1 + 3x2 + 2(2.5) + 7(−1) = −1 :‬ﻭ ﻟﺬﺍ‬
‫‪x1 = 1 − 3x2‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ‪ x2‬ﺭﺍ ﺩﻟﺨﻮﺍﻩ ﻣﻲﺗﻮﺍﻥ ﺍﻧﺘﺨﺎﺏ ﻧﻤﻮﺩ ﻭ ﺑﻪ ﺗﺒﻊ ﺁﻥ ﻣﻘﺪﺍﺭﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ‪ x1‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪.‬‬

‫ﺍﺯ ﺭﻭﺍﻝ ﺟﺎﻳﮕﺬﺍﺭﻱ ﻓﻮﻕ ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺍﺳﺖ‪ .‬ﻣﺠﻤﻮﻋﻪ ﺟﻮﺍﺏﻫﺎﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪h‬‬ ‫⊤‪i‬‬ ‫‬
‫‪1 − 3x2‬‬ ‫‪x2‬‬ ‫‪2.5‬‬ ‫‪−1‬‬ ‫‪: x2 ∈ R‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺗﻤﺎﻣﻲ ﺟﻮﺍﺏﻫﺎ‪ ،‬ﻣﻘﺪﺍﺭ ‪ x4‬ﻭ ‪ x3‬ﻳﻜﺴﺎﻥ ﺍﺳﺖ ﻭﻟﻲ ‪ x2‬ﻭ ﺑﻪ ﺗﺒﻊ ﺁﻥ ‪ x1‬ﻣﻘﺎﺩﻳﺮ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺗﻮﺍﻧﻨﺪ ﺍﺧﺘﻴﺎﺭ‬
‫ﻛﻨﻨﺪ‪.‬‬

‫■ ﺟﻤﻊﺑﻨﺪﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬

‫• ﺍﮔﺮ ‪ Ax = b‬ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻣﻲﺗﻮﺍﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ‬
‫ﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻛﻠﻴﻪ ﺿﺮﺍﻳﺐ ﺭﻭﻱ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ‪ ،‬ﺭﻭﺍﻝ‬
‫ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﺪ‪.‬‬
‫• ﺍﮔﺮ ‪ Ax = b‬ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎﺯ ﻫﻢ ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻣﻲﺗﻮﺍﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ‬
‫ﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪ ،‬ﻣﻨﺘﻬﻲ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﺣﺪﺍﻗﻞ ﻳﻜﻲ ﺍﺯ ﺿﺮﺍﻳﺐ ﺭﻭﻱ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‪،‬‬
‫ﺭﻭﺍﻝ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﺩﺍﺭﺩ ﻳﺎ ﺧﻴﺮ ﻭ ﺩﺭ ﺻﻮﺭﺕ ﻭﺟﻮﺩ ﺟﻮﺍﺏ‪ ،‬ﻓﺮﻡ ﻛﻠﻲ ﺟﻮﺍﺏﻫﺎ‬
‫ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﺪ‪.‬‬

‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ‬ ‫�‬

‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ ۵۵‬ﻫﻤﺎﻧﻨﺪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺍﺯ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪ ،‬ﺍﻣﺎ‬
‫ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ‪ ،‬ﺑﺎ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ )ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻫﻤﺎﻧﻲ(‬
‫ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺷﻤﺎﻱ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪b1‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪b̄1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪b2 ‬‬ ‫‪‬‬ ‫···‬ ‫‪b̄2 ‬‬
‫‪‬‬ ‫‪ 0 1‬ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪ .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪−−−−−−−−−→  . .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪ .. ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪bn‬‬ ‫‪0 0‬‬ ‫···‬ ‫‪1‬‬ ‫‪b̄n‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﻌﺪ ﺍﺯ ﺗﺒﺪﻳﻞ ﻓﻮﻕ‪ ،‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﺳﺘﻮﻥ ﺁﺧﺮ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ‬
‫ﺻﻮﺭﺗﻲ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﺒﺪﻳﻞ ﻓﻮﻕ ﻣﻤﻜﻦ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬
‫‪55 Gauss-Jordan‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۹۸‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻣﺎ ﻧﺤﻮﻩ ﺗﺒﺪﻳﻞ ﺩﺳﺘﮕﺎﻩ‬
‫ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ ﺑﻪ ﺻﻮﺭﺕﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺗﻮﺍﻧﺪ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ .‬ﺑﻪ ﻫﻤﻴﻦ ﻋﻠﺖ‪ ،‬ﺩﺭ ﻣﻨﺎﺑﻊ ﻣﺘﻔﺎﻭﺕ‪ ،‬ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ‬
‫ﺑﺮﺍﻱ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺩﻭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﺍﻳﺞ ﺭﺍ ﺑﺮ ﻣﻲﺷﻤﺎﺭﻳﻢ‪.‬‬

‫‪ GJ1‬ﺩﺭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺍﻭﻝ‪ ،‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﺗﺎ ﺁﺧﺮ‪ ،‬ﺍﺑﺘﺪﺍ ﻋﻨﺼﺮ ﻣﺤﻮﺭﻱ ﺳﺘﻮﻥ ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻋﻨﺎﺻﺮ ﺯﻳﺮ‬
‫ﺁﻥ ﺻﻔﺮ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ﻳﻚ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺳﭙﺲ ﺍﺯ‬
‫ﺳﺘﻮﻥ ‪n‬ﺍﻡ ﺷﺮﻭﻉ ﻣﻲﻛﻨﻴﻢ ﻭ ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﻗﻄﺮ ﺭﺍ ﻧﻴﺰ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫‪ GJ2‬ﺩﺭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺩﻭﻡ‪ ،‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﺗﺎ ﺁﺧﺮ‪ ،‬ﺍﺑﺘﺪﺍ ﻋﻨﺼﺮ ﻣﺤﻮﺭﻱ ﺳﺘﻮﻥ ﺑﺮﺍﺑﺮ ‪ 1‬ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ‬
‫ﻣﺤﻮﺭ ﻭ ﻫﻤﻴﻦ ﻃﻮﺭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺑﺎﻻﻱ ﻣﺤﻮﺭ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺑﻌﺪ ﺍﺯ ‪ n‬ﻣﺮﺣﻠﻪ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ‬
‫ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻟﮕﻮﺭﻳﺘﻢ ﮔﻮﺱ ﺟﺮﺩﻥ ‪ GJ1‬ﺭﺍ ﺑﻪ ﻃﻮﺭ ﻛﺎﻣﻞ ﺗﺸﺮﻳﺢ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫■ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪GJ1‬‬

‫ﺩﺭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ ،GJ1‬ﺩﺭ ﺍﺑﺘﺪﺍ ﺩﺭ ‪ n‬ﻣﺮﺣﻠﻪ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺳﭙﺲ ﺩﺭ ‪ n − 1‬ﻣﺮﺣﻠﻪ ﺩﻳﮕﺮ‪ ،‬ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﻗﻄﺮ ﺍﺻﻠﻲ ﺑﻪ ﺗﺮﺗﻴﺐ ﺍﺯ ﺳﺘﻮﻥ ‪n‬ﺍﻡ ﺗﺎ ﺳﺘﻮﻥ ﺩﻭﻡ ﻧﻴﺰ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪ .‬ﻛﻠﻴﻪ ‪2n − 1‬‬
‫ﻣﺮﺣﻠﻪ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ GJ1‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ ،GJ1‬ﺩﺭ ﺻﻮﺭﺕ ﻧﻴﺎﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﺗﺎ ﻋﺪﺩﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ‬
‫)‪ (1, 1‬ﻗﺮﺍﺭ ﮔﻴﺮﺩ)ﭼﻮﻥ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﻓﺮﺽ ﺷﺪﻩ‪ ،‬ﻟﺬﺍ ﻫﻤﻮﺍﺭﻩ ﺍﻳﻦ ﻛﺎﺭ ﻣﻤﻜﻦ ﺍﺳﺖ(‪ .‬ﺳﭙﺲ ﻣﻮﻟﻔﻪ )‪ (1, 1‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ‬
‫ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﺍﺑﺘﺪﺍ ﺑﺎ ﻛﻤﻚ ﻋﻤﻞ ﺳﻄﺮﻱ ‪ E3‬ﻣﺤﻮﺭ ﺭﺍ ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12 · · · a1n‬‬ ‫‪b1‬‬ ‫‪1‬‬ ‫‪a12 · · · a1n‬‬
‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬
‫‪b1‬‬ ‫‪1 a12 · · · a1n‬‬
‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬
‫‪b1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a21 a22 · · · a2n‬‬ ‫‪b2 ‬‬ ‫‪ a21 a22 · · · a2n‬‬ ‫‪b2 ‬‬ ‫‪0 a22 · · · a2n‬‬‫)‪(1‬‬ ‫)‪(1‬‬
‫‪b2 ‬‬
‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪ .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪−‬‬‫‪−‬‬‫→‬
‫‪..  E3 ‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪−‬‬‫‪−‬‬‫→‬
‫‪..  E1 ‬‬
‫‪‬‬ ‫‪.. ..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬ ‫‪. ‬‬ ‫‪.‬‬
‫‪‬‬ ‫‪ .‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪an1 an2 · · · ann‬‬ ‫‪bn‬‬ ‫‪an1 an2 · · · ann‬‬ ‫‪bn‬‬ ‫‪0 an2 · · · ann‬‬
‫)‪(1‬‬ ‫)‪(1‬‬
‫‪bn‬‬
‫)‪(1‬‬

‫ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﺗﺎ )‪(n − 1‬ﺍﻡ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﻣﻮﻟﻔﻪ )‪ (2, 2‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﮔﺮ ﺍﻳﻦ ﻣﻮﻟﻔﻪ ﺻﻔﺮ ﺑﻮﺩ‪ ،‬ﺑﺎ‬
‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﻋﻨﺼﺮﻱ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺳﭙﺲ ﻣﺤﻮﺭ ﺭﺍ ﺑﻪ ﻳﻚ ﺗﺒﺪﻳﻞ ﻛﺮﺩﻩ ﻭ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﺭﺍ ﺻﻔﺮ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ‪ ،‬ﺩﺭ ﻣﺮﺍﺣﻞ ﺳﻮﻡ ﺗﺎ )‪ (n − 1‬ﺑﻪ ‪ 1‬ﻛﺮﺩﻥ ﻣﺤﻮﺭ ﻭ ﺻﻔﺮ ﻛﺮﺩﻥ ﻋﻨﺎﺻﺮ ﺯﻳﺮﺵ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﺑﻌﺪ ﺍﺯ ﺍﻧﺠﺎﻡ‬
‫‪ n − 1‬ﻣﺮﺣﻠﻪ‪ ،‬ﺩﺭﻧﻬﺎﻳﺖ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻓﺮﻡ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻣﺮﺣﻠﻪ ‪n‬ﺍﻡ‪ :‬ﺩﺭ ﻣﺮﺣﻠﻪ ‪n‬ﺍﻡ ﺑﺎ ﻋﻤﻞ ﺳﻄﺮﻱ ‪ E3‬ﻣﻮﻟﻔﻪ )‪ (n, n‬ﺭﺍ ﻧﻴﺰ ‪ 1‬ﻣﻲﻛﻨﻴﻢ‪ ،‬ﺗﺎ ﺩﺭ ﻧﻬﺎﻳﺖ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻣﺎﺗﺮﻳﺴﻲ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ‪ 1‬ﺗﺒﺪﻳﻞ ﺷﻮﺩ‪ .‬ﺷﻤﺎﻱ ‪ n − 1‬ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺭﻭﺵ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12 · · · a1,n−1 a1n‬‬ ‫‪b1‬‬ ‫‪1 a12 · · · a1,n−1 a1n‬‬
‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬
‫‪b1‬‬‫)‪(1‬‬

‫‪‬‬ ‫‪‬‬ ‫‪ n‬ﻣﺮﺣﻠﻪﻱ‬ ‫‪‬‬ ‫‪‬‬

‫‪ a21 a22 · · · a2,n−1 a2n b2 ‬‬ ‫‪0‬‬ ‫‪1 · · · a2,n−1 a2n‬‬
‫)‪(2‬‬ ‫)‪(2‬‬
‫‪b2 ‬‬
‫)‪(2‬‬

‫‪‬‬ ‫‪ ‬ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪‬‬ ‫‪‬‬

‫‪ ..‬‬ ‫‪.‬‬
‫‪..‬‬ ‫‪.‬‬
‫‪..‬‬ ‫‪.‬‬
‫‪..‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪..  −−−−−−−−−−−−−→ ‬‬ ‫‪.‬‬ ‫‪.. . .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ .‬‬ ‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ E3 ، E2‬ﻭ ‪E1‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪an−1,1 an−1,2 · · · an−1,n−1 an−1,n bn−1 ‬‬ ‫‪0‬‬ ‫··· ‪0‬‬ ‫‪1‬‬ ‫‪an−1,n‬‬
‫)‪(n−1‬‬
‫‪bn−1 ‬‬
‫)‪(n−1‬‬

‫‪an,1 an,2 · · · an,n−1 ann‬‬ ‫‪bn‬‬ ‫··· ‪0 0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪bn‬‬‫)‪(n‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۹۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻣﺮﺣﻠﻪ )‪(n + 1‬ﺍﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺑﺎﻻﻱ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ‪n‬ﺍﻡ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺍﺯ ﻣﻮﻟﻔﻪ )‪(n, n‬ﺍﻡ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ ،‬ﻛﻪ ﻣﻘﺪﺍﺭ ﺁﻥ ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫)‪(1‬‬
‫‪a1n‬‬ ‫‪b1‬‬‫)‪(1‬‬
‫‪1‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫‪0‬‬ ‫)‪(n+1‬‬
‫‪b1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a2,n−1‬‬ ‫)‪(2‬‬
‫‪a2n‬‬ ‫‪b2 ‬‬
‫)‪(2‬‬
‫‪0‬‬ ‫‪1‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a2,n−1‬‬ ‫‪0‬‬ ‫‪b2 ‬‬
‫)‪(n+1‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪ ..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪.‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬‫‪−−→ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪ E1 ‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪1‬‬ ‫‪an−1,n‬‬
‫)‪(n−1‬‬
‫‪bn−1 ‬‬
‫)‪(n−1‬‬
‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪bn−1 ‬‬
‫)‪(n+1‬‬

‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪bn‬‬‫)‪(n‬‬

‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪bn‬‬‫)‪(n‬‬

‫ﻣﺮﺣﻠﻪ )‪(n + 2‬ﺍﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻣﻮﻟﻔﻪ )‪(n − 1, n − 1‬ﺍﻡ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﺑﺎ ﻛﻤﻚ ﺁﻥ ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ‬
‫ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ )‪(n − 1‬ﺍﻡ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫)‪(1‬‬
‫‪a12‬‬ ‫···‬ ‫)‪(1‬‬
‫‪a1,n−1‬‬ ‫‪0‬‬ ‫‪b1‬‬‫)‪(n+1‬‬
‫‪1‬‬ ‫‪a12‬‬ ‫)‪(1‬‬
‫···‬ ‫‪0‬‬ ‫‪0‬‬ ‫)‪(n+2‬‬
‫‪b1‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬ ‫···‬ ‫)‪(2‬‬
‫‪a2,n−1‬‬ ‫‪0‬‬ ‫‪b2 ‬‬
‫)‪(n+1‬‬
‫‪0‬‬ ‫‪1‬‬ ‫···‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪b2 ‬‬
‫)‪(n+2‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪ ..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪ ..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪−‬‬‫‪−‬‬‫→‬
‫‪.  E1 ‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪bn−1 ‬‬
‫)‪(n+1‬‬
‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪bn−1 ‬‬
‫)‪(n+1‬‬

‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪bn‬‬‫)‪(n‬‬

‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0 1‬‬ ‫‪bn‬‬ ‫)‪(n‬‬

‫ﻣﺮﺣﻠﻪ )‪(n + 3‬ﺍﻡ ﺍﻟﻲ )‪(2n − 1‬ﺍﻡ ‪ :‬ﺩﺭ ﻫﺮ ﻣﺮﺣﻠﻪ ﺍﺯ ﻣﺮﺍﺣﻞ )‪(n + 3‬ﺍﻡ ﺍﻟﻲ )‪(2n − 1‬ﺍﻡ‪ ،‬ﻣﻮﻟﻔﻪﻫﺎﻱ ﺑﺎﻻﻱ ﻗﻄﺮ‬
‫ﺩﺭ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ‪ ،‬ﺍﺯ ﺁﺧﺮ ﺑﻪ ﺗﺎ ﺳﺘﻮﻥ ﺩﻭﻡ‪ ،‬ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﻌﺪ ﺍﺯ ﺍﻳﻦ ﻣﺮﺍﺣﻞ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪0‬‬ ‫)‪(2n−1‬‬
‫‪b1‬‬
‫‪‬‬ ‫‪‬‬
‫‪0 1‬‬ ‫···‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪‬‬
‫)‪(2n−2‬‬
‫‪b2‬‬
‫‪‬‬ ‫‪‬‬
‫‪ .. ..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪. .‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫···‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪bn−1 ‬‬
‫)‪(n+1‬‬

‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪bn‬‬ ‫)‪(n‬‬

‫ﺍﻛﻨﻮﻥ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﺳﺘﻮﻥ ﺁﺧﺮ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺣﺎﺻﻞ ﺩﺭ ﺗﻜﺮﺍﺭ ﺁﺧﺮ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪.‬‬
‫ﺩﺭﺍﺩﺍﻣﻪ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ GJ1‬ﺑﺮﺍﻱ ﺣﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﺭﻭﺵ ﮔﻮﺱ⁃ﺟﺮﺩﻥ ‪(GJ1‬‬ ‫ﻣﺜﺎﻝ ‪)۱۲−۴‬ﺣﻞ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪2 1‬‬ ‫‪2‬‬ ‫‪10‬‬
‫‪‬‬ ‫‪‬‬
‫‪1 2‬‬ ‫‪1‬‬ ‫‪8‬‬
‫‪‬‬ ‫‪‬‬
‫‪3 1‬‬ ‫‪−1‬‬ ‫‪2‬‬
‫ﺩﺭ ﺯﻳﺮ ﻣﺮﺍﺣﻞ ﺍﻟﮕﻮﺭﻳﺘﻢ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ GJ1‬ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬
‫◀ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪:‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰۰‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪2‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪10‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5 R2:=−1R1+R2 1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪ R1:= 21 R1 ‬‬ ‫‪ R3:=−3R1+R3 ‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪8‬‬ ‫‪‬‬ ‫‪8‬‬ ‫‪‬‬ ‫‪3‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪ −−−−−→ 1‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪ −−−−−−−−→ 0‬‬ ‫‪1.5‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪ R2:= 3 R2 ‬‬
‫‪2‬‬
‫‪ R3:= 2 R2+R3 ‬‬
‫‪1‬‬
‫‪‬‬
‫‪0‬‬ ‫‪1.5‬‬ ‫‪0‬‬ ‫‪3‬‬‫‪‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬ ‫→‬ ‫‪‬‬ ‫‪‬‬ ‫‪−‬‬
‫‪−‬‬ ‫‪−‬‬‫‪−‬‬ ‫‪−‬‬ ‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫→‬ ‫‪‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬ ‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−4‬‬ ‫‪−12‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻣﻮﻟﻔﻪ )‪ (3, 3‬ﻧﻴﺰ ‪ 1‬ﻣﻲﺷﻮﺩ‪.‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1 0.5‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪ R3:=− 14 R3 ‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪1 0 2‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪ −−−−−−→ 0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−4‬‬ ‫‪−12‬‬ ‫‪0‬‬ ‫‪0 1 3‬‬

‫ﺗﺎ ﺍﻳﻦﺟﺎ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ ﻗﻄﺮﻱ ‪ 1‬ﺗﺒﺪﻳﻞ ﮔﺮﺩﻳﺪ‪ .‬ﺩﺭ ﺩﻭ ﻣﺮﺣﻠﻪ ﺑﻌﺪ‪ ،‬ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﻗﻄﺮ ﻧﻴﺰ ﺻﻔﺮ‬
‫ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫◀ ﻣﺮﺣﻠﻪ ﭼﻬﺎﺭﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ‪ ،‬ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺳﻮﻡ ﺻﻔﺮ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪1 0.5 0‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪ R1:=−1R3+R1 ‬‬
‫‪R2:=0R3+R2‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫→‬ ‫‪0‬‬ ‫‪1 0‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪0‬‬ ‫‪0 1‬‬ ‫‪3‬‬

‫◀ ﻣﺮﺣﻠﻪ ﭘﻨﺠﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1 0.5 0‬‬ ‫‪2‬‬ ‫‪1 0 0‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪ R1:=− 21 R2+R1 ‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1 0‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2 −−−−−−−−−→ 0 1 0‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0 1‬‬ ‫‪3‬‬ ‫‪0 0 1‬‬ ‫‪3‬‬

‫ﺣﺎﻝ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﺳﺘﻮﻥ ﺁﺧﺮ ﻣﺎﺗﺮﻳﺲ ﻓﻮﻕ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﻳﻌﻨﻲ‪:‬‬
‫‪x1 = 1, x2 = 2, x3 = 3.‬‬

‫■ ∗ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪GJ2‬‬

‫ﺍﻟﮕﻮﺭﻳﺘﻢ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ ،GJ2‬ﺩﺭ ‪ n‬ﻣﺮﺣﻠﻪ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ ﻣﺮﺣﻠﻪ ‪k‬ﺍﻡ ﺍﻳﻦ ﺍﻟﮕﻮﺭﻳﺘﻢ‪،‬‬
‫ﻣﻮﻟﻔﻪ )‪ (k, k‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺻﻮﺭﺕ ﻧﻴﺎﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﺗﺎ ﻳﻚ ﻋﺪﺩ ﻏﻴﺮﺻﻔﺮ ﺩﺭ ﺍﻳﻦ‬
‫ﻣﻮﻗﻌﻴﺖ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺑﺘﺪﺍ ﻣﺤﻮﺭ ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﻭ ﻫﻤﻴﻦ ﻃﻮﺭ ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﺁﻥ ﺻﻔﺮ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺟﺰﺋﻴﺎﺕ ﻣﺮﺍﺣﻞ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺁﻭﺭﺩﻩ ﻧﻤﻲﺷﻮﺩ ﻭ ﺍﻳﻦ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺑﺎ ﻳﻚ ﻣﺜﺎﻝ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۰۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﺭﻭﺵ ﮔﻮﺱ⁃ﺟﺮﺩﻥ ‪(GJ2‬‬ ‫ﻣﺜﺎﻝ ‪)۱۳−۴‬ﺣﻞ ﯾ‬

‫ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۳-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺍﻳﻦﺑﺎﺭ ﺑﺎ ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ GJ2‬ﺁﻥ ﺭﺍ ﺣﻞ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ‬
‫ﻣﺮﺍﺣﻞ ﺍﻟﮕﻮﺭﻳﺘﻢ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪ GJ2‬ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬
‫◀ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ :‬ﻣﻮﻟﻔﻪ )‪ (1, 1‬ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪10‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5 R2:=−1R1+R2 1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪ R1:= 21 R1 ‬‬ ‫‪ R3:=−3R1+R3 ‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪8‬‬ ‫‪‬‬ ‫‪8‬‬ ‫‪‬‬ ‫‪3‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪ −−−−−→ 1‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪ −−−−−−−−→ 0‬‬ ‫‪1.5‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪1‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ :‬ﻣﻮﻟﻔﻪ ﻣﺤﻮﺭﻱ )‪ (2, 2‬ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﻭ ﺑﺎﻻﻱ ﺁﻥ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫‪5 R3:=− 21 R2+R3 1 0‬‬ ‫‪1‬‬ ‫‪4‬‬
‫‪‬‬ ‫‪ R2:= 3 R2 ‬‬
‫‪2‬‬
‫‪ R1:=+ 12 R2+R1 ‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪3‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫‪−‬‬ ‫→‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪1.5‬‬ ‫‪0‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪ −−−−−−−−−→ 0 1‬‬ ‫‪0‬‬ ‫‪2‬‬
‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬ ‫‪0‬‬ ‫‪−0.5‬‬ ‫‪−4‬‬ ‫‪−13‬‬ ‫‪0 0‬‬ ‫‪−4‬‬ ‫‪−12‬‬

‫◀ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ :‬ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻣﻮﻟﻔﻪ )‪ (3, 3‬ﻧﻴﺰ ‪ 1‬ﻣﻲﺷﻮﺩ ﻭ ﻋﻨﺎﺻﺮ ﺑﺎﻻﻱ ﺁﻥ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪4‬‬ ‫‪R2:=0R3+R2 1 0‬‬ ‫‪0‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪ R3:=− 41 R3 ‬‬ ‫‪ R1:=−1R3+R1‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪2 −−−−−−−−→ ‬‬
‫‪‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪ −−−−−−→ 0‬‬ ‫‪1‬‬ ‫‪0‬‬ ‫‪0 1‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−4‬‬ ‫‪−12‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪0 0‬‬ ‫‪1‬‬ ‫‪3‬‬

‫ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﺳﺘﻮﻥ ﺁﺧﺮ ﻣﺎﺗﺮﻳﺲ ﻓﻮﻕ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪.‬‬

‫■ ﻧﻜﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ‬

‫• ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺩﺭ ‪ n‬ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺭﻭﺵﻫﺎﻱ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺍﺯ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E3‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﻫﺮ ﺳﻪ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻳﺎﺩﺁﻭﺭﻱ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ‬
‫ﮔﻮﺱ‪ ،‬ﻓﻘﻂ ﺍﺯ ﺍﻋﻤﺎﻝ ‪ E2‬ﻭ ‪ E1‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫• ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻧﺎﻣﻨﻔﺮﺩ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﮔﺮ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﻭﻟﻲ ﺍﮔﺮ ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺭ ﻳﻚ ﻣﺮﺣﻪ ﺑﻪ ﻣﺤﻮﺭ‬
‫ﺻﻔﺮ ﺑﺮﻣﻲﺧﻮﺭﻳﻢ ﻛﻪ ﻛﻠﻴﻪ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﻧﻴﺰ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺗﺒﻌﺎ ﺩﺭ ﭼﻨﻴﻦ ﺣﺎﻟﺘﻲ‪ ،‬ﺗﺒﺪﻳﻞ ﻣﺎﺗﺮﻳﺲ ﺑﻪ ﻓﺮﻡ ﻫﻤﺎﻧﻲ ﻣﻤﻜﻦ‬
‫ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻭﻗﺘﻲ ﺩﺳﺘﮕﺎﻩ ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﻣﻲﺗﻮﺍﻥ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﻪ ﻓﺮﻡ ﺳﻄﺮﻱ ﭘﻠﻜﺎﻧﻲ ‪ ۵۶‬ﺗﺒﺪﻳﻞ‬
‫ﻧﻤﻮﺩ‪ .‬ﺑﻌﺪ ﺍﺯ ﺗﺒﺪﻳﻞ ﺑﻪ ﻓﺮﻡ ﺳﻄﺮﻱ ﭘﻠﻜﺎﻧﻲ‪ ،‬ﺑﻪ ﺳﺎﺩﮔﻲ ﻣﻲﺗﻮﺍﻥ ﺗﺸﺨﻴﺺ ﺩﺍﺩ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ‬
‫ﺟﻮﺍﺏ ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻓﺮﻡ ﻋﻤﻮﻣﻲ ﺟﻮﺍﺏﻫﺎ ﺭﺍ ﻧﻴﺰ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪.‬‬
‫ﺧﻮﺍﻧﻨﺪﮔﺎﻥ ﻋﻼﻗﻪﻣﻨﺪ ﺑﻪ ]؟ [ ﻣﺮﺍﺟﻌﻪ ﻛﻨﻨﺪ‪.‬‬

‫• ﺩﺭ ﺑﺨﺶﻫﺎﻱ ﺁﺗﻲ ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ GJ1‬ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﺑﺎ ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ GJ2‬ﺑﻪ ﺗﻌﺪﺍﺩ ﻣﺤﺎﺳﺒﺎﺕ ﻛﻤﺘﺮﻱ ﻧﻴﺎﺯ ﺩﺍﺭﺩ‬
‫ﻭ ﻟﺬﺍ ﺑﻪ ﻟﺤﺎﻅ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻣﻨﺎﺳﺐﺗﺮ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺑﻪ ﻟﺤﺎﻅ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ‪ ،‬ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ GJ2‬ﺳﺎﺩﻩﺗﺮ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪56 Row‬‬ ‫‪echelon‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰۲‬‬

‫∗ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ‬ ‫‪۴.۴‬‬

‫ﻓﺮﻡ ﻛﻠﻲ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ‪ ۵۷‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪AX = B‬‬ ‫)‪(۱۴.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ A ∈ Rn×n‬ﻭ ‪ B ∈ Rn×m‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻠﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ‪ X ∈ Rn×m‬ﻣﺎﺗﺮﻳﺴﻲ ﻣﺠﻬﻮﻝ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺍﮔﺮ ‪ ،k = 1‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﻣﻌﻤﻮﻝ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ﻣﻘﺎﺑﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺎﺗﺮﻳﺴﻲ ﺑﻪ‬
‫ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﺑﺮﺩﺍﺭﻱ ‪ ۵۸‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ‬
‫ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪A‬‬ ‫‪B‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ )‪ (۱۴.۴‬ﺭﺍ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺮﺑﻌﻲ ﺩﺭ ﻧﻈﺮ ﺧﻮﺍﻫﻴﻢ ﮔﺮﻓﺖ‪ ،‬ﻫﺮﭼﻨﺪ ﻛﻪ‬
‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺎﺗﺮﻳﺴﻲ ﺑﺎ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻏﻴﺮ ﻣﺮﺑﻌﻲ ﻧﻴﺰ ﻣﻄﺮﺡ ﻣﻲﺑﺎﺷﻨﺪ ﻭﻟﻲ ﺩﺭ ﺍﻳﻨﺠﺎ ﺑﻪ ﺁﻥ ﻧﻤﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬
‫ﺍﺯ ﻳﻚ ﺩﻳﺪﮔﺎﻩ‪ ،‬ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺎﺗﺮﻳﺴﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻣﺠﻤﻮﻋﻪﺍﻱ ﺍﺯ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺮﺩﺍﺭﻱ ﺑﺎ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻳﻜﺴﺎﻥ ﺩﺭ ﻧﻈﺮ‬
‫ﮔﺮﻓﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ )‪ (۱۴.۴‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ‪ k‬ﺩﺳﺘﮕﺎﻩ ﺑﺮﺩﺍﺭﻱ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪:‬‬
‫‪Axj = bj , j = 1, . . . , k,‬‬ ‫)‪(۱۵.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ xj‬ﻭ ‪ bj‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺳﺘﻮﻥﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ‪ X‬ﻭ ‪ B‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﭼﻨﺪﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺩﺍﺷﺘﻪ‬
‫ﺑﺎﺷﻴﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺗﻤﺎﻡ ﺩﺳﺘﮕﺎﻩﻫﺎ ﻳﻜﺴﺎﻥ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺭ ﻭﺍﻗﻊ ﺑﺎ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻣﻮﺍﺟﻪ ﻫﺴﺘﻴﻢ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ‬
‫ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۴−۴‬ﺩﺳﺘﮕﺎﻩ ﺧﻄ ﻣﺎﺗﺮﯾﺴ ﻣﻌﺎﺩﻝ ﺑﺎ ﺩﻭ ﺩﺳﺘﮕﺎﻩ ﺧﻄ ﺑﺮﺩﺍﺭﯼ(‬

‫ﺩﻭ ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪3x1 + x2 + x3 = 5‬‬ ‫‪‬‬
‫‪3x1 + x2 + x3 = 2‬‬
‫‪‬‬ ‫‪‬‬
‫‪6x1 − 12x2 − x3 = −7‬‬ ‫‪6x1 − 12x2 − x3 = 7‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪3x1 + 15x2 − 5x3 = 13‬‬ ‫‪3x1 + 15x2 − 5x3 = 8‬‬
‫ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻫﺮ ﺩﻭ ﺩﺳﺘﮕﺎﻩ ﻳﻜﺴﺎﻥ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﻳﻦ ﺩﻭ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ‪ AX = B‬ﺩﺭ ﻧﻈﺮ‬
‫ﮔﺮﻓﺖ ﻛﻪ‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪5‬‬ ‫‪x11‬‬ ‫‪x12‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪A := ‬‬
‫‪6‬‬ ‫‪−12 −1‬‬
‫‪‬‬ ‫‪B := ‬‬
‫‪−7 7‬‬
‫‪‬‬ ‫‪X := ‬‬
‫‪x21‬‬ ‫‪x22 ‬‬
‫‪‬‬
‫‪3‬‬ ‫‪15 −5‬‬ ‫‪13 8‬‬ ‫‪x31‬‬ ‫‪x32‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺍﻭﻝ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ‪ X‬ﻭ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﻭﻡ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ‪ X‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻣﺎﺗﺮﻳﺲ‬
‫ﺍﻓﺰﻭﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪‬‬
‫‪6‬‬ ‫‪−12 −1‬‬ ‫‪−7‬‬ ‫‪7‬‬ ‫)‪(۱۶.۴‬‬
‫‪‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪15‬‬ ‫‪−5‬‬ ‫‪13‬‬ ‫‪8‬‬

‫‪57 System‬‬ ‫‪of matrix linear equations‬‬ ‫‪58 System‬‬ ‫‪of vector linear equations‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۰۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻫﻤﺎﻧﻨﺪ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺑﺮﺩﺍﺭﻱ‪ ،‬ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ﻧﻴﺰ ﺳﻪ ﺣﺎﻟﺖ ﺩﺭ ﻣﻮﺭﺩ ﺟﻮﺍﺏ ﻣﻲﺗﻮﺍﻧﺪ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ‬
‫ﺑﺎﺷﺪ‪ .‬ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﻭﺟﻮﺩ ﻭ ﻣﻨﺤﺼﺮﺑﻪ ﻓﺮﺩﯼ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎەﻫﺎﯼ ﺧﻄ ﻣﺎﺗﺮﯾﺴ‬ ‫ﻗﻀﯿﻪ ‪۹−۴‬‬

‫‪ b ∈ R‬ﺩﺍﺭﻳﻢ‪:‬‬ ‫‪n×m‬‬
‫‪A∈R‬ﻭ‬ ‫‪n×n‬‬
‫ﺩﺭ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ‪ AX = B‬ﻛﻪ‬

‫ﺍﻟﻒ( ﺍﮔﺮ ‪ A‬ﻣﺎﺗﺮﻳﺴﻲ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ‪ X = A−1 B‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺏ( ﺍﮔﺮ ‪ A‬ﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻳﺎ ﺟﻮﺍﺏ ﻧﺪﺍﺭﺩ ﻭ ﻳﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻮﺍﺏ ﺩﺍﺭﺩ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ )‪ (۱۴.۴‬ﺭﺍ ﺑﺎ ‪ k‬ﺩﺳﺘﮕﺎﻩ )‪ (۱۵.۴‬ﻣﺘﻨﺎﻇﺮ ﻧﻤﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺟﻮﺍﺏ‬
‫ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺭﺍ ﺑﺎ ﺣﻞ ‪ k‬ﺩﺳﺘﮕﺎﻩ ﻣﺘﻨﺎﻇﺮﺵ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺘﻨﺎﻇﺮ ﺩﺍﺭﺍﻱ ﻣﺎﺗﺮﻳﺲ‬
‫ﺿﺮﺍﻳﺐ ﻳﻜﺴﺎﻥ ‪ A‬ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﺍﻳﻦ ﻧﻜﺘﻪ ﻣﻲﺗﻮﺍﻧﺪ ﺩﺭ ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎﻱ ﻛﺎﺭﺍ ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ‪ k‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﺭﻭﺩ‪.‬‬

‫■ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺎﺗﺮﻳﺴﻲ‬

‫‪h‬‬ ‫‪i‬‬
‫ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻧﺎﻣﻨﻔﺮﺩ ‪ ،AX = B‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ‪ A B‬ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ ﻭ ﺑﺎ‬
‫‪h‬‬ ‫‪i‬‬
‫ﻛﻤﻚ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E2‬ﻭ ‪ ، E1‬ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺭﺍ ﺑﻪ ﻓﺮﻡ ̄‪ U B‬ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ‪ U‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ‬
‫ﺑﺎﻻﻣﺜﻠﺜﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﺯﻳﺮ‪ ،‬ﺍﻳﻦ ﻓﺮﺍﻳﻨﺪ ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11 a12 · · · a1,n-1 a1n b11 · · · b1m‬‬ ‫‪u11 u12 · · · u1,n-1 u1n b̄11 · · · b̄1m‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a21 a22 · · · a2,n-1 a2n b21 · · · b2m ‬‬ ‫‪‬‬ ‫‪0 u22 · · · u2,n-1 u2n b̄21 · · · b̄2m ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ ..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..  −−−→ ‬‬
‫‪‬‬ ‫‪.. .. . .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ .‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.  E1,E2 ‬‬
‫‪‬‬ ‫‪. .‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ n-1,1 n-1,2‬‬
‫‪a‬‬ ‫‪a‬‬ ‫·‬ ‫·‬ ‫·‬ ‫‪a‬‬ ‫‪n-1,n-1‬‬ ‫‪a‬‬ ‫‪n-1,n‬‬ ‫‪b‬‬ ‫‪n-1,1‬‬ ‫·‬ ‫·‬ ‫·‬ ‫‪b‬‬ ‫‪n-1,m ‬‬ ‫‪‬‬ ‫‪0 0 · · · un-1,n-1 un-1,n b̄n-1,1 · · · b̄n-1,m ‬‬
‫‪an,1 an,2 · · · an,n-1 ann bn1 · · · bnm‬‬ ‫‪0‬‬ ‫··· ‪0‬‬ ‫‪0‬‬ ‫‪unn b̄n1 · · · b̄nm‬‬
‫ﺣﺎﻝ ﺑﺎ ﺗﻌﻤﻴﻤﻲ ﺍﺯ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﻣﻮﺳﻮﻡ ﺑﻪ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺑﺮﺩﺍﺭﻱ ﻣﻲﺗﻮﺍﻥ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺭﺍﻳﻪ‬
‫ﻧﻤﻮﺩ‪:‬‬

‫‪b̄′n‬‬
‫= ‪ :x′n‬ﺳﻄﺮ ‪n‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬
‫‪unn‬‬
‫‪b̄′ − un-1,n x′n‬‬
‫‪ :x′n-1 = n-1‬ﺳﻄﺮ ‪n − 1‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬
‫‪un-1,n-1‬‬
‫‪b̄′n-2 − un-2,n-1 x′n-1 − un-2,n x′n‬‬
‫= ‪ :x′n-2‬ﺳﻄﺮ ‪n − 2‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬
‫‪un-2,n−2‬‬
‫‪..‬‬
‫‪:‬‬ ‫‪.‬‬
‫‪b̄′ − u12 x′2 − u13 x′3 − · · · − u1n x′n‬‬
‫‪ :x′1 = 1‬ﺳﻄﺮ ‪1‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬
‫‪u11‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪b̄′i := b̄i1‬‬ ‫‪...‬‬ ‫ﻛﻪ ‪ x′i‬ﺳﻄﺮ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ﻣﺠﻬﻮﻝ ‪ X‬ﻣﻲﺑﺎﺷﺪ ﻭ ‪ b̄i′‬ﻧﻴﺰ ﺳﻄﺮ ‪i‬ﺍﻡ ﻣﺎﺗﺮﻳﺲ ̄‪ b‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ‪b̄im .‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻫﻤﺎﻥ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺍﺳﺖ ﻛﻪ ﺭﻭﻱ ﺑﺮﺩﺍﺭﻫﺎﻱ ﺳﻄﺮﻱ )ﻭ ﻧﻪ ﺍﺳﻜﺎﻟﺮ( ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰۴‬‬

‫ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﯾﺴ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ(‬ ‫ﻣﺜﺎﻝ ‪)۱۵−۴‬ﺣﻞ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ )‪ (۱۶.۴‬ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۴-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﺣﻞ ﺁﻥ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﻣﺮﺍﺣﻞ‬
‫ﺭﻭﺵ ﺩﺭ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5 2‬‬ ‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5 2‬‬ ‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5 2‬‬
‫‪‬‬ ‫‪ R2:=−2R1+R2‬‬
‫‪R3:=−1R1+R3 ‬‬ ‫‪ R3:=1R2+R3 ‬‬ ‫‪‬‬
‫‪6 −12 −1 −7 7 −‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪ −−−−−−−→ 0 −14 −3 −17 3 −−−−−−−→ 0 −14 −3 −17 3‬‬
‫‪3 15 −5 13 8‬‬ ‫‪0 14 −6‬‬ ‫‪8 6‬‬ ‫‪0‬‬ ‫‪0 −9 −9 9‬‬
‫ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﮔﺮﺩﻳﺪ‪ .‬ﺍﻛﻨﻮﻥ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺑﺮﺩﺍﺭﻱ ﺭﺍ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫] ‪[ −9 9‬‬
‫= ‪ :x′3‬ﺳﻄﺮ ﺳﻮﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬ ‫] ‪= [ 1 −1‬‬
‫‪−9‬‬
‫] ‪[ −17 3 ] − (−3)[ 1 −1‬‬ ‫] ‪[ −14 0‬‬
‫= ‪ :x′2‬ﺳﻄﺮ ﺩﻭﻡ ﻣﺎﺗﺮﻳﺲ ‪X‬‬ ‫=‬ ‫]‪= [1 0‬‬
‫‪−14‬‬ ‫‪−14‬‬
‫] ‪[ 5 2 ] − [ 1 0 ] − [ 1 −1‬‬ ‫]‪[3 3‬‬
‫= ‪ :x′1‬ﺳﻄﺮ ﺍﻭﻝ ﻣﺎﺗﺮﻳﺲ ‪X‬‬ ‫=‬ ‫]‪= [1 1‬‬
‫‪3‬‬ ‫‪3‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪،‬ﺳﻄﺮﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﻣﺠﻬﻮﻝ ‪ X‬ﺑﻪ ﺩﺳﺖ ﺁﻣﺪ ﻭ ﻟﺬﺍ‬
‫‪  ‬‬ ‫‪‬‬
‫‪′‬‬
‫‪x1‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪  ‬‬ ‫‪‬‬
‫‪X=‬‬ ‫‪′‬‬
‫‪x 2  = 1‬‬
‫‪‬‬ ‫‪0‬‬‫‪‬‬
‫‪x′3‬‬ ‫‪1‬‬ ‫‪−1‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺍﻭﻝ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۴-۴‬ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ‪ X‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺩﻭﻡ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪.‬‬

‫■ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺎﺗﺮﻳﺴﻲ‬

‫ﺧﻄﻲ ﻣﺎﺗﺮﻳﺴﻲ ‪ AX = B‬ﺍﺯ ﺳﻪ ﻋﻤﻞ ﻣﻘﺪﻣﺎﺗﻲ ﺳﻄﺮﻱ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ ﺗﺎ‬ ‫ﺩﺳﺘﮕﺎﻩ ‪h‬‬‫ﺑﺮﺍﻱ ﺣﻞ ‪i‬‬
‫ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ‪h‬‬
‫‪i‬‬ ‫ﺩﺭ ﺭﻭﺵ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ‪ A B‬ﺑﻪ ﻓﺮﻡ ̄‪ I B‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻝ ﻣﺎﺗﺮﻳﺲ ̄‪ B‬ﻫﻤﺎﻥ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺣﺎﻟﺘﻲ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ ﺍﻳﻦ ﺗﺒﺪﻳﻞ ﻣﻤﻜﻦ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻫﻤﺎﻧﻨﺪ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻋﺎﺩﻱ )ﺑﺮﺩﺍﺭﻱ(‪ ،‬ﺩﺭ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺮﺩﺍﺭﻱ ﻧﻴﺰ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺗﻮﺍﻥ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺭﺍ ﺑﻪ ﻓﺮﻡ‬
‫‪h‬‬ ‫‪i‬‬
‫̄‪ I B‬ﺗﺒﺪﻳﻞ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻨﺠﺎ ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﻣﻄﺎﺑﻖ ﺑﺎ ﺭﻭﻳﻜﺮﺩ ‪ GJ1‬ﻭ ﻳﺎ ‪ GJ2‬ﺑﻪ ﺗﺒﺪﻳﻞ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﭘﺮﺩﺍﺧﺖ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺧﻄ ﻣﺎﺗﺮﯾﺴ ﺑﺎ ﺭﻭﺵ ﮔﻮﺱ⁃ﺟﺮﺩﻥ(‬ ‫ﻣﺜﺎﻝ ‪)۱۶−۴‬ﺣﻞ ﯾ‬

‫ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۵-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺩﺭ ﺍﻳﻨﺠﺎ ﺁﻥ ﺭﺍ ﺑﺎ ﺭﻭﺵ ‪ GJ1‬ﺣﻞ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﻣﺮﺍﺣﻞ ﺣﻞ ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪2‬‬
‫‪3‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪5 2‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪ R1:= 13 R1 ‬‬ ‫‪3‬‬ ‫‪‬‬ ‫‪3‬‬ ‫‪3‬‬ ‫‪3  R2:=−6R1+R2‬‬ ‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬
‫‪6 −12 −1 −7 7 −‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪ R3:=−3R1+R3‬‬
‫‪‬‬
‫‪‬‬ ‫‪ −−−−→ 6 −12 −1 −7 7 −−−−−−−−→ 0 −14 −3 −17 3‬‬
‫‪3 15 −5 13 8‬‬ ‫‪3 15 −5 13 8‬‬ ‫‪0 14 −6‬‬ ‫‪8 6‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1 13 13 53 23‬‬ ‫‪1 1 1 5 2‬‬
‫‪1‬‬
‫‪R2:= −14‬‬ ‫‪R2 ‬‬ ‫‪ R3:=−14R2+R3  3 3 3 3  R2:= −14‬‬‫‪1‬‬
‫‪R2‬‬
‫‪−−−−−−−→ ‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪3 17 −3  −‬‬
‫‪14 14 14 ‬‬
‫‪−‬‬ ‫‪−‬‬ ‫‪−‬‬‫‪−‬‬‫‪−‬‬‫‪−‬‬‫‪−‬‬ ‫‪−‬‬‫→‬ ‫→‪0 1 3 17 −3  −−−−−−−‬‬
‫‪‬‬ ‫‪14 14 14 ‬‬
‫‪0 14 −6‬‬ ‫‪8‬‬ ‫‪6‬‬ ‫‪0 0 −9 −9‬‬ ‫‪9‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۰۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬

‫‪1‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪2‬‬ ‫‪3‬‬
‫‪R2:=− 14‬‬ ‫‪R3+R2‬‬ ‫‪1‬‬ ‫‪4‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬‫‪0‬‬ ‫‪1 0 0 1 1‬‬
‫‪‬‬ ‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬
‫‪R1:=− 31 R3+R1‬‬ ‫‪‬‬ ‫‪3‬‬ ‫‪ R1:=− 13 R2+R1 ‬‬
‫‪3‬‬ ‫‪‬‬
‫‪0 1‬‬ ‫→‪−3  −−−−−−−−−‬‬ ‫‪0 1 0 1 0 −‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ −−−−−−−−→ 0 1 0 1 0‬‬
‫‪3‬‬ ‫‪17‬‬
‫‪‬‬ ‫‪14‬‬ ‫‪14‬‬ ‫‪14 ‬‬ ‫‪‬‬
‫‪0 0‬‬ ‫‪1‬‬ ‫‪1 −1‬‬ ‫‪0 0 1 1 −1‬‬ ‫‪0 0 1 1 −1‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪X = 1‬‬ ‫‪0‬‬
‫‪.‬‬
‫‪1‬‬ ‫‪−1‬‬

‫ﺧﻄﺎ ﻭ ﻛﻨﺘﺮﻝ ﺧﻄﺎ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫‪۵.۴‬‬

‫ﺗﺎ ﺍﻳﻦ ﺟﺎ ﺭﻭﺵﻫﺎﻱ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﻭ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺮﻓﻲ ﺷﺪﻧﺪ‪ .‬ﺍﻣﺎ ﺩﺭ ﺧﺼﻮﺹ‬
‫ﻣﻼﺣﻈﺎﺕ ﺧﻄﺎﻱ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺻﺤﺒﺘﻲ ﺑﻪ ﻋﻤﻞ ﻧﻴﺎﻣﺪ‪ .‬ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺗﻮﺳﻂ ﻳﻚ ﻣﺎﺷﻴﻦ ﺑﺮﺍﻱ ﺣﻞ‬
‫ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﻧﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻫﺎﻱ ﺭﻭﻧﺪﻛﺮﺩﻥ‪ ،‬ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﻴﻨﻲ ﻭ ﺍﻧﺘﺸﺎﺭ ﺭﺥ ﻣﻲﺩﻫﺪ‪ .‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ‬
‫ﺑﺎ ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻻﺗﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺑﺎﺷﻨﺪ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬

‫‪ .۱‬ﺧﻄﺎﻱ ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ )ﻭ ﻳﺎ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ( ﺗﺎ ﭼﻪ ﺍﻧﺪﺍﺯﻩ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺰﺭﮒ ﺑﺎﺷﺪ؟‬

‫‪ .۲‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺗﺮﻓﻨﺪﻱ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻭ ﻳﺎ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺑﻪ ﻛﺎﺭ ﺑﺮﻳﻢ‪ ،‬ﺗﺎ ﻧﺘﺎﻳﺞ ﻣﺎﺷﻴﻨﻲ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﻗﻴﻖﺗﺮ ﺑﺎﺷﺪ؟‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻻﺕ ﭘﺎﺳﺦ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺭﺍﺳﺘﺎﻱ ﺳﻮﺍﻝ ﺍﻭﻝ‪ ،‬ﺩﻭ ﻣﺜﺎﻝ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻭ ﺍﻧﺪﺍﺯﻩ ﺧﻄﺎﻱ ﻧﺘﺎﻳﺞ‬
‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﻭﻱ ﻣﺎﺷﻴﻦﻫﺎ ﮔﺰﺍﺭﺵ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ]؟ [(‬ ‫ﻣﺎﺷﯿﻦ ﺑﺮﺍﯼ ﺣﻞ ﯾ‬ ‫ﻣﺜﺎﻝ ‪)۱۷−۴‬ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺭﻭﯼ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪5.42 16.78 0.78‬‬ ‫‪38.0014‬‬
‫‪‬‬ ‫‪‬‬
‫‪0.423‬‬ ‫‪2.3 23.46‬‬ ‫‪53.97833‬‬
‫‪‬‬ ‫‪‬‬
‫‪26.73‬‬ ‫‪1.27 2.45‬‬ ‫‪68.74938‬‬
‫ﺟﻮﺍﺏ ﺩﻗﻴﻖ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪x1 = 2.31, x2 = 1.42, x3 = 2.12.‬‬
‫ﺣﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﺍ ﺭﻭﻱ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﺍﻋﺪﺍﺩ ﻣﺎﺷﻴﻨﻲ )‪ FS (10, 7, −8, 7‬ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪ .‬ﺑﺎ ﻋﻤﻞ‬
‫ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ ، E1‬ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﺜﻠﺜﻲ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪5.420000‬‬ ‫‪16.78000 0.7800000‬‬ ‫‪38.00140‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪0.9904170‬‬ ‫‪23.39913‬‬ ‫‪51.01254‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪1923.623‬‬ ‫‪4078.082‬‬
‫ﺍﻟﺒﺘﻪ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻓﺮﺍﻳﻨﺪ ﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺩﺭ ﻣﺎﺷﻴﻦ ﻣﺒﺘﻨﻲ ﺑﺮ )‪ FS (10, 7, −8, 7‬ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺍﺳﺖ ﻭ ﻛﻠﻴﻪ‬
‫ﻣﺤﺎﺳﺒﺎﺕ ﻣﻴﺎﻧﻲ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﻭﻱ ﺍﻳﻦ ﻣﺎﺷﻴﻦ ﺍﻧﺠﺎﻡ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻣﻮﻟﻔﻪ )‪ (2, 2‬ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺍﺯ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰۶‬‬

‫‪ − 0.423‬ﺑﺮﺍﺑﺮ ﺳﻄﺮ ﺍﻭﻝ ﺑﺎ ﺳﻄﺮ ﺩﻭﻡ ﺩﺭ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺍﻭﻟﻴﻪ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﻳﻌﻨﻲ‪:‬‬
‫ﺟﻤﻊ ‪5.42‬‬
‫‪0.423‬‬
‫‪a22 := −‬‬
‫)‪(1‬‬
‫‪16.78 + 2.3‬‬
‫‪5.42‬‬
‫ﺍﻣﺎ ﺩﺭ ﻣﺎﺷﻴﻦ‪ ،‬ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫‬ ‫‬
‫‪0.423‬‬
‫‪a22 := fl −‬‬
‫)‪(1‬‬
‫‪16.78 + 2.3‬‬
‫‪5.42‬‬
‫ﻛﻪ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 7, −8, 7‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪m := 0.423 ⊘ 5.42 = 0.4230000 e+0 ⊘ 0.5420000e+1 = 0.7804428 e−1,‬‬
‫‪z1 := −m ⊗ 16.78 = (−0.7804428 e−1) ⊗ (0.5420000 e+1) = −0.1309583e+1,‬‬
‫‪a22 := z1 ⊕ 2.3 = (−0.1309583 e+1) ⊕ 0.2300000 e+1 = 0.9904170 e+0‬‬
‫)‪(1‬‬

‫ﺍﻟﺒﺘﻪ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻳﻦ ﻋﺒﺎﺭﺕ ﺑﺮﺍﺑﺮ · · · ‪ 0.9904169741697417‬ﺍﺳﺖ ﻛﻪ ﻣﻲﺑﻴﻨﻴﻢ ﻣﻘﺪﺍﺭ ﻣﺎﺷﻴﻨﻲ ﺑﺎ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﺗﻔﺎﻭﺕ‬
‫ﺍﻧﺪﻛﻲ ﺩﺍﺭﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺗﻘﺮﻳﺒﺎ ﺗﻤﺎﻡ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺷﺪﻩ ﺑﺎ ﺧﻄﺎ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫ﺑﻌﺪ ﺍﺯ ﻣﺜﻠﺜﻲ ﺷﺪﻥ‪ ،‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺑﻪ ﺣﻞ ﻣﺴﺎﻟﻪ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﻧﻴﺰ ﺩﺭ ﻣﺎﺷﻴﻦ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻭ ﺗﺒﻌﺎ ﺧﻄﺎﻱ‬
‫ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﻧﻴﺰ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﻧﺘﺎﻳﺞ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 7, −8, 7‬ﺩﺭ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ‬
‫‬ ‫‬ ‫ﺍﺳﺖ‪:‬‬
‫‪4078.082‬‬
‫‪x3 = fl‬‬ ‫‪= 2.120001,‬‬
‫‪1923.623‬‬
‫‬ ‫‬
‫)‪51.01254 − (23.39913)(2.120001‬‬
‫‪x2 = fl‬‬ ‫‪= 1.419968,‬‬
‫‪0.9904170‬‬
‫‬ ‫‬
‫)‪38.00140 − (16.78000)(1.419968) − (0.7800000)(2.120001‬‬
‫‪x1 = fl‬‬ ‫‪= 2.310100.‬‬
‫‪5.420000‬‬
‫ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺟﻮﺍﺏ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﮔﻮﺱ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 7, −8, 7‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫|‪|2.31 − 2.3101‬‬
‫‪ :‬ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ‪x1‬‬ ‫‪× 100 = 4.33 e−3 %,‬‬
‫‪2.31‬‬
‫|‪|1.42 − 1.419968‬‬
‫‪ :‬ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ‪x2‬‬ ‫‪× 100 = 2.25 e−3 %,‬‬
‫‪1.42‬‬
‫|‪|2.12 − 2.12001‬‬
‫‪ :‬ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺩﺭ ‪x3‬‬ ‫‪× 100 = 5.72 e−5 %.‬‬
‫‪2.12‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻣﺜﺎﻝ ﻓﻮﻕ ﺩﻳﺪﻳﻢ‪ ،‬ﺧﻄﺎﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﭼﻨﺪﺍﻥ ﭼﺸﻤﮕﻴﺮ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺍﻟﺒﺘﻪ ﺭﺿﺎﻳﺖ ﺑﺨﺶ ﻧﻴﺰ ﻧﻴﺴﺖ‪.‬‬
‫ﺍﻣﺎ ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻧﻴﺰ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺧﻄﺎﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺑﺎﺷﺪ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۸−۴‬ﻣﻘﺎﯾﺴﻪ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻗﯿﻖ ﻭ ﻣﺤﺎﺳﺒﺎﺕ ﻣﺎﺷﯿﻨ (‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫"‬ ‫‪#‬‬
‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬
‫‪0.234 −1.317‬‬ ‫‪1.029‬‬
‫ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ‪ x1 = 10, x2 = 1‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﺍ ﻳﻚ ﺑﺎﺭ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﺩﻗﻴﻖ ﻭ ﻭﺍﻗﻌﻲ )ﺑﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺭﻗﻢ ﺍﻋﺸﺎﺭ( ﻭ ﻳﻚ ﺑﺎﺭ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 4, −5, 5‬ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪ .‬ﻧﺘﺎﻳﺞ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۰۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺩﺭ ﺯﻳﺮ ﻭ ﺩﺭ ﺩﻭ ﺳﺘﻮﻥ ﻣﺠﺰﺍ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬

‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺭ ﻣﺎﺷﻴﻦ )‪FS (10, 4, −5, 5‬‬ ‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻗﻴﻖ‬

‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬

‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬ ‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬
‫‪0.2346‬‬ ‫‪−1.317‬‬ ‫‪1.029‬‬ ‫‪0.2346‬‬ ‫‪−1.317‬‬ ‫‪1.029‬‬

‫‪m=fl( 0.2346‬‬
‫)‪0.0002 )=1173, R2:=fl(−mR1+R2‬‬
‫‪m= 0.2346 =1173, R2:=−mR1+R2‬‬
‫→‪−−−−−−−−−−−−−−−−−−−−−−−−−‬‬ ‫‪−−−−‬‬‫→‪−−−−−−−−−−−−−−−−‬‬
‫‪0.0002‬‬

‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬

‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬ ‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬
‫‪0‬‬ ‫‪−1726‬‬ ‫‪−1727‬‬ ‫‪0‬‬ ‫‪−1726.8‬‬ ‫‪−1726.8‬‬
‫ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬ ‫ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬
‫‬ ‫‬ ‫‪−1726.8‬‬
‫‪−1727‬‬ ‫= ‪x2‬‬ ‫‪= 1,‬‬
‫‪x2 = fl‬‬ ‫‪= 1.001,‬‬ ‫‪−1726.8‬‬
‫‪−1726‬‬
‫‬ ‫‬ ‫)‪1.473 − 1.471(1‬‬
‫)‪1.473 − (1.471)(1.001‬‬ ‫= ‪x1‬‬ ‫‪= 10.‬‬
‫‪x1 = fl‬‬ ‫‪= 5.000.‬‬ ‫‪0.0002‬‬
‫‪0.0002‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﻣﺤﺎﺳﺒﺎﺕ ﺩﻗﻴﻖ ﺑﻪ ﺟﻮﺍﺏ ﺩﻗﻴﻖ )‪ (x1 , x2 ) = (10, 1‬ﻣﻨﺠﺮ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﻣﺎ‬
‫ﺭﻭﺵ ﮔﻮﺱ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 4, −5, 5‬ﺑﻪ ﺟﻮﺍﺑﻲ ﻣﺘﻔﺎﻭﺕ )‪ (x1 , x2 ) = (5.000, 1.001‬ﻣﻨﺠﺮ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻣﺸﺎﻫﺪﻩ‬
‫ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺟﻮﺍﺏ ﻣﺎﺷﻴﻨﻲ ﺧﻄﺎﻱ ﻛﻤﻲ ﺑﺮﺍﻱ ‪ x2‬ﺍﺗﻔﺎﻕ ﺍﻓﺘﺎﺩﻩ ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺧﻄﺎﻱ ‪ x1‬ﭼﺸﻤﮕﻴﺮ )‪ (50%‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻫﻨﮕﺎﻡ‬
‫ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ‪ ،‬ﺩﺭ ﺍﺑﺘﺪﺍ ‪ x2‬ﻭ ﺳﭙﺲ ‪ x1‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﻣﻘﺪﺍﺭ ‪ x2‬ﻫﻤﺮﺍﻩ ﺑﺎ ﺧﻄﺎ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ ﻛﻪ ﺩﺭ ﺍﻳﻨﺠﺎ ﺍﻳﻦ ﺧﻄﺎ‬
‫ﻧﺴﺒﺘﺎ ﻛﻮﭼﻚ ﺍﺳﺖ‪ .‬ﺳﭙﺲ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ x1‬ﺍﺯ ‪ x2‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺧﻄﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ‪ x2‬ﺑﻪ ‪ x1‬ﻧﻴﺰ ﺳﺮﺍﻳﺖ‬
‫ﻣﻲﻛﻨﺪ‪ .‬ﺍﻣﺎ ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻫﻨﮕﺎﻡ ﻣﺤﺎﺳﺒﻪ ‪ ،x1‬ﺧﻄﺎﻱ ﻛﻮﭼﻚ ﻣﻮﺟﻮﺩ ﺩﺭ ‪ x2‬ﺗﻘﻮﻳﺖ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺧﻄﺎﻱ‬
‫ﺑﺰﺭﮔﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﺩﻭ ﻣﺜﺎﻝ ﻓﻮﻕ ﺩﻳﺪﻳﻢ‪ ،‬ﺧﻄﺎ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺘﺎﻳﺞ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎ ﺑﺎ ﺧﻄﺎ‬
‫ﻫﻤﺮﺍﻩ ﺍﺳﺖ ﻭ ﮔﺎﻫﻲ ﺍﻳﻦ ﺧﻄﺎ ﭼﺸﻤﮕﻴﺮ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺗﺎﺛﻴﺮ ﺧﻄﺎ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﻛﺎﻫﺶ ﺩﺍﺩ؟ ﺩﺭ ﺑﺨﺶ ﺑﻌﺪ ﺧﻮﺍﻫﻴﻢ ﺩﻳﺪ ﻛﻪ ﺑﺎ ﺗﺮﻓﻨﺪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺍﻳﻦ ﻣﻬﻢ ﺩﺳﺖ ﻳﺎﻓﺖ‪.‬‬

‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫�‬

‫ﺍﻳﺪﻩ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺍﺯ ﻗﻀﻴﻪ ﺯﻳﺮ ﻧﺎﺷﻲ ﻣﻲ ﺷﻮﺩ‪.‬‬

‫ﺗﺎﺛﯿﺮ ﺍﻧﺪﺍﺯﻩ ﻣﺤﻮﺭ ﺑﺮ ﺧﻄﺎﯼ ﺭﻭﺵ ﮔﻮﺱ‬ ‫ﻗﻀﯿﻪ ‪۱۰−۴‬‬

‫ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‪ ،‬ﻫﺮﭼﻪ ﺍﻧﺪﺍﺯﻩ ﻣﺤﻮﺭﻫﺎ ﻛﻮﭼﻚ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﺧﻄﺎﻱ ﺑﻴﺸﺘﺮﻱ ﻫﻤﺮﺍﻩ ﺍﺳﺖ‪.‬‬
‫ﺑﺮﻋﻜﺲ‪ ،‬ﻫﺮﭼﻪ ﺍﻧﺪﺍﺯﻩ ﻣﺤﻮﺭﻫﺎ ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻛﻢﺗﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬

‫ﻧﻜﺘﻪ ﻣﻮﺟﻮﺩ ﺩﺭ ﻗﻀﻴﻪ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺩﺭ ﻣﺜﺎﻝﻫﺎﻱ ‪ ۱۷-۴‬ﻭ ‪ ۱۸-۴‬ﻣﻲﺗﻮﺍﻥ ﻣﺸﺎﻫﺪﻩ ﻛﺮﺩ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۸-۴‬ﺍﻧﺪﺍﺯﻩ ﻣﺤﻮﺭ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۰۸‬‬

‫ﻛﻮﭼﻚ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺧﻄﺎﻱ ﺯﻳﺎﺩﻱ ﺑﻪ ﻭﺟﻮﺩ ﺁﻣﺪ‪ .‬ﺍﻣﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﺳﺖ ﻛﻪ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮﻫﺎ‪ ،‬ﺗﺎ‬
‫ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﻋﺪﺩﻱ ﺑﺎ ﺍﻧﺪﺍﺯﻩ ﺑﺰﺭﮒ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﺩﺍﺩ‪ .‬ﺑﺎ ﺍﻳﻦ ﻛﺎﺭ‪ ،‬ﻛﻪ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ‪ ۵۹‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫ﻣﺤﻮﺭﮔﯿﺮﯼ ﺟﺰﺋ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ‬ ‫ﺗﻌﺮﯾﻒ ‪۱۱−۴‬‬
‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺍﻳﻦ ﻛﻪ ﺩﺭ ﻫﺮ ﻣﺮﺣﻠﻪ ﺍﺯ ﺭﻭﺵ‪ ،‬ﻣﻌﺎﺩﻟﻪ )ﺳﻄﺮ( ﻣﺤﻮﺭ ﺑﺎ ﻳﻚ ﻣﻌﺎﺩﻟﻪ‬
‫)ﺳﻄﺮ( ﺯﻳﺮ ﺧﻮﺩﺵ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺟﺎﺑﺠﺎ ﺷﻮﺩ ﻛﻪ ﻣﻮﻟﻔﻪﺍﻱ ﺑﺎ ﺑﻴﺸﺘﺮﻳﻦ ﺍﻧﺪﺍﺯﻩ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﮔﻴﺮﺩ‪.‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﺑﺮﺍﻱ ﺑﺮﺭﺳﻲ ﺗﺎﺛﻴﺮ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‪ ،‬ﻣﺜﺎﻝﻫﺎﻱ ‪ ۱۷-۴‬ﻭ ‪ ۱۸-۴‬ﻛﻪ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻣﻌﻤﻮﻟﻲ ‪ ۶۰‬ﺣﻞ ﺷﺪﻩ‬
‫ﺑﻮﺩﻧﺪ ﺭﺍ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻫﻤﺮﺍﻩ ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺣﻞ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ]؟ [(‬ ‫ﻣﺎﺷﯿﻦ ﺑﺮﺍﯼ ﺣﻞ ﯾ‬ ‫ﻣﺜﺎﻝ ‪ )۱۹−۴‬ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺑﺎ ﻣﺤﻮﺭﮔﯿﺮﯼ ﺟﺰﺋ ﺭﻭﯼ ﯾ‬
‫ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۷-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺭﻭﻱ ﻣﺎﺷﻴﻦ )‪،FS (10, 7, −8, 7‬‬
‫ﺟﻮﺍﺏ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻣﻌﻤﻮﻟﻲ ﺑﺎ ‪ 2.25 e−3 ،4.33 e−3‬ﻭ ‪ 5.72 e−5‬ﺩﺭﺻﺪ ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﻱ ‪ x2 ،x1‬ﻭ‬
‫‪ x3‬ﻫﻤﺮﺍﻩ ﺑﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﻫﻤﺮﺍﻩ ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺭﺍ ﺭﻭﻱ ﺩﺳﺘﮕﺎﻩ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ‬
‫ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪5.420000 16.78000 .7800000‬‬ ‫‪38.00140‬‬
‫‪‬‬ ‫‪‬‬
‫‪.4230000 2.300000‬‬ ‫‪23.46000‬‬ ‫‪53.97833‬‬
‫‪‬‬ ‫‪‬‬
‫‪26.73000 1.270000‬‬ ‫‪2.450000‬‬ ‫‪68.74938‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮ ﺍﻭﻝ ﺑﺎ ﺳﻄﺮ ﺳﻮﻡ‪ ،‬ﺑﺰﺭﮒﺗﺮﻳﻦ ﻋﺪﺩ ﺩﺭ‬
‫ﺳﺘﻮﻥ ﺍﻭﻝ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪26.73000‬‬ ‫‪1.270000‬‬ ‫‪2.450000‬‬ ‫‪68.74938‬‬
‫‪‬‬ ‫‪‬‬
‫‪−−−−−−−−−−−→ ‬‬ ‫‪53.97833‬‬
‫‪R1⇆R3‬‬
‫‪.4230000 2.300000‬‬
‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬
‫‪23.46000‬‬ ‫‪‬‬
‫‪5.420000 16.78000 .7800000‬‬ ‫‪38.00140‬‬
‫ﺣﺎﻝ ﺑﺎ ﺍﻳﻦ ﻣﺤﻮﺭ ﺟﺪﻳﺪ )ﻛﻪ ﺍﻧﺪﺍﺯﻩﺍﺵ ﺍﺯ ﺍﻧﺪﺍﺯﻩ ﻋﻨﺎﺻﺮ ﺯﻳﺮﺵ ﺑﺰﺭﮒﺗﺮ ﺍﺳﺖ( ﺑﻪ ﺻﻔﺮ ﻛﺮﺩﻥ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺯﻳﺮ ﻗﻄﺮ ﭘﺮﺩﺍﺧﺘﻪ‬
‫ﻣﻲﺷﻮﺩ‪:‬‬
‫‪R2:=−fl( 0.4230000‬‬ ‫‪e+0‬‬ ‫‪‬‬ ‫‪‬‬
‫)‪0.2673000 e+2 R1+R2‬‬
‫‪26.73000‬‬ ‫‪1.270000 2.450000‬‬ ‫‪68.74938‬‬
‫‪R3:=−fl( 0.5420000‬‬ ‫‪e+1‬‬
‫)‪0.2673000 e+2 R1+R3‬‬
‫‪‬‬ ‫‪‬‬
‫‪−−−−−−−−−−−−−−−−−−→ ‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪2.279839 23.42123‬‬ ‫‪52.89038‬‬
‫‪‬‬
‫‪0‬‬ ‫‪16.52167 .2832174‬‬ ‫‪24.06120‬‬
‫ﺩﺭ ﺍﻳﻦ ﻣﺮﺣﻠﻪ ﻧﻴﺰ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻧﻴﺎﺯ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﮔﺮ ﺳﻄﺮ ﺩﻭﻡ ﻭ ﺳﻮﻡ ﺭﺍ ﺟﺎﺑﺠﺎ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻋﺪﺩﻱ ﺑﺎ‬
‫ﺑﻴﺸﺘﺮﻳﻦ ﺍﻧﺪﺍﺯﻩ ﺩﺭ ﻣﻮﻗﻌﻴﺖ ﻣﺤﻮﺭ ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪26.73000‬‬ ‫‪1.270000‬‬ ‫‪2.450000‬‬ ‫‪68.74938‬‬
‫‪‬‬ ‫‪‬‬
‫‪−−−−−−−−−−−→ ‬‬ ‫‪24.06120‬‬
‫‪R2⇆R3‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪16.52167 .2832174‬‬ ‫‪‬‬
‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬
‫‪0‬‬ ‫‪2.279839‬‬ ‫‪23.42123‬‬ ‫‪52.89038‬‬

‫‪59 Partial‬‬ ‫‪pivoting‬‬ ‫‪60 Naive‬‬ ‫‪Gaussian elimination method‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۰۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺣﺎﻝ ﺑﺎ ﺍﻧﺠﺎﻡ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‪ ،‬ﻣﻮﻟﻔﻪ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪26.73000‬‬ ‫‪1.270000 2.450000‬‬ ‫‪68.74938‬‬
‫(‪R3:=−fl‬‬ ‫‪) ‬‬
‫‪0.2279839 e+1‬‬
‫‪0.1652167 e+2 R2+R3‬‬
‫‪‬‬
‫‪−−−−−−−−−−−−−−−−−−→ ‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪16.52167 .2832174‬‬ ‫‪24.06120‬‬
‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪23.38215‬‬ ‫‪49.57016‬‬
‫ﺍﻛﻨﻮﻥ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﮔﺮﺩﻳﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﺲﺭﻭ ﺟﻮﺍﺏ ﻣﺴﺎﻟﻪ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪:‬‬
‫‬ ‫‬
‫‪49.57016‬‬
‫‪x3 = fl‬‬ ‫‪= 2.120000,‬‬
‫‪23.38215‬‬
‫‬ ‫‬
‫)‪24.06120 − (.2832174)(2.120000‬‬
‫‪x2 = fl‬‬ ‫‪= 1.420000,‬‬
‫‪16.52167‬‬
‫‬ ‫‬
‫)‪68.74938 − (1.270000)(1.420000) − (2.450000)(2.120000‬‬
‫‪x1 = fl‬‬ ‫‪= 2.310000.‬‬
‫‪26.73000‬‬
‫ﺟﻮﺍﺏﻫﺎﻱ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻫﻤﺎﻥ ﺟﻮﺍﺏﻫﺎﻱ ﺩﻗﻴﻖ ﻣﺴﺎﻟﻪ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺧﻄﺎﻳﻲ ﻧﺪﺍﺭﻳﻢ‪ .‬ﺍﻟﺒﺘﻪ‬
‫ﺍﻳﻦ ﺣﺎﻟﺖ ﺍﻳﺪﻩﺁﻝ ﻫﻤﻴﺸﻪ ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﻧﺠﺎﻡ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺑﺎﻋﺚ ﻧﻤﻲﺷﻮﺩ ﻛﻪ ﺧﻄﺎ ﺻﻔﺮ ﺷﻮﺩ‪ ،‬ﺑﻠﻜﻪ‬
‫ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺧﻄﺎ ﻛﻢ ﻭ ﻣﻌﻘﻮﻝ ﮔﺮﺩﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۲۰−۴‬ﻣﻘﺎﯾﺴﻪ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﯿﺮﯼ ﻭ ﺑﺎ ﻣﺤﻮﺭﮔﯿﺮﯼ(‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﻣﻄﺎﻟﻌﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۸-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺩﺭ ﻣﺎﺷﻴﻦ )‪ FS (10, 4, −5, 5‬ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ‬
‫ﮔﻮﺱ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻭ ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺑﻪ ﺣﻞ ﺁﻥ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬ ‫ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬

‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬

‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬ ‫‪0.0002‬‬ ‫‪1.471‬‬ ‫‪1.473‬‬
‫‪0.2346 −1.317 1.029‬‬ ‫‪0.2346‬‬ ‫‪−1.317‬‬ ‫‪1.029‬‬
‫"‬ ‫‪#‬‬
‫‪R1⇆R2‬‬ ‫‪0.2346 −1.317 1.029‬‬
‫→‪−−−−−−−‬‬
‫ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬ ‫‪0.0002‬‬ ‫‪1.471 1.473‬‬

‫◀ ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺍﺭﻳﻢ‪:‬‬ ‫◀ ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺍﺭﻳﻢ‪:‬‬
‫(‪m:=fl‬‬ ‫)‪)=0.0008525, R2:=fl(−mR1+R2‬‬
‫‪0.0002‬‬
‫(‪m:=fl‬‬ ‫)‪)=1173, R2:=fl(−mR1+R2‬‬
‫‪0.2346‬‬

‫→‪−−−−−−−−−−−−−−−−−−−−−−−−−−−−‬‬ ‫→‪−−−−−−−−−−−−−−−−−−−−−−−−−‬‬
‫‪0.2346‬‬ ‫‪0.0002‬‬

‫"‬ ‫‪#‬‬ ‫"‬ ‫‪#‬‬

‫‪0.2346 −1.317 1.029‬‬ ‫‪0.0002 1.471 1.473‬‬
‫‪0‬‬ ‫‪1.472‬‬ ‫‪1.474‬‬ ‫‪0‬‬ ‫‪1726‬‬ ‫‪−1727‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۱۰‬‬

‫◀ ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬ ‫◀ ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ ﭘﺲﺭﻭ ﺩﺍﺭﻳﻢ‪:‬‬

‫‬ ‫‬ ‫‬ ‫‬
‫‪1.474‬‬ ‫‪−1727‬‬
‫‪x2 = fl‬‬ ‫‪= 1.001,‬‬ ‫‪x2 = fl‬‬ ‫‪= 1.001,‬‬
‫‪1.472‬‬ ‫‪1726‬‬
‫‬ ‫‬ ‫‬ ‫‬
‫)‪1.029 − (−1.317)(1.001‬‬ ‫)‪1.473 − (1.471)(1.001‬‬
‫‪x1 = fl‬‬ ‫‪x1 = fl‬‬
‫‪0.2346‬‬ ‫‪0.0002‬‬
‫‪= 10.00.‬‬ ‫‪= 5.000.‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ‪ ،‬ﺟﻮﺍﺏ ﺣﺎﺻﻞ ﺑﺮﺍﻱ ‪ x1‬ﻭ ‪ x2‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺩﺍﺭﺍﻱ ‪ 50%‬ﻭ ‪0.1%‬‬
‫ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﻭﻗﺘﻲ ﺍﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﺧﻄﺎﻱ ﻧﺴﺒﻲ ﺑﺮﺍﻱ ‪ x1‬ﻭ ‪ x2‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺩﺍﺭﺍﻱ ‪ 0%‬ﻭ‬
‫‪ 0.1%‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ‪ ،‬ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻣﺎﻧﻊ ﺗﻘﻮﻳﺖ ﺷﺪﻳﺪ ﺧﻄﺎ ﻣﻲﺷﻮﺩ ﻭ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﻳﮕﺮ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫‪۶.۴‬‬

‫ﺗﺎ ﺍﻳﻨﺠﺎ‪ ،‬ﻛﺎﺭﺑﺮﺩ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﺍ ﺩﺭ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺩﻳﺪﻳﻢ‪ .‬ﺍﻣﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺩﺭ ﺣﻞ ﻣﺴﺎﻳﻞ ﺩﻳﮕﺮﻱ ﻧﻴﺰ‬
‫ﻛﺎﺭﺑﺮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺑﺮﺧﻲ ﺍﺯ ﺁﻥ ﻛﺎﺭﺑﺮﺩﻫﺎ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫• ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‬
‫• ﻣﺤﺎﺳﺒﻪ ﻭﺍﺭﻭﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‬
‫• ﺗﺸﺨﻴﺺ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﻓﻮﻕ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺍﺭﺍﻳﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫�‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ‪ A‬ﻣﺎﺗﺮﻳﺴﻲ ﻣﺮﺑﻌﻲ ﺑﺎ ﺑﻌﺪ ‪ n‬ﺑﺎﺷﺪ‪ .‬ﻣﺸﺎﺑﻪ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﺎ‬
‫ﻛﻤﻚ ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E1‬ﻭ ‪ E2‬ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺭﺍ ﺑﻪ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﺎﻧﻨﺪ ‪ U‬ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪.‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪a11‬‬ ‫‪a12‬‬ ‫···‬ ‫‪a1n‬‬ ‫‪u11‬‬ ‫‪u12‬‬ ‫···‬ ‫‪u1n‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ a21‬‬ ‫‪a22‬‬ ‫···‬ ‫‪a2n‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪u22‬‬ ‫···‬ ‫‪u2n ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪A :=  .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪ −−−−−−−−→ U := ‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪‬‬ ‫‪ E1‬ﻭ ‪E2‬‬ ‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪an1‬‬ ‫‪an2‬‬ ‫···‬ ‫‪ann‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫···‬ ‫‪unn‬‬
‫ﺣﺎﻝ ﺳﻪ ﻧﻜﺘﻪ ﺯﻳﺮ ﺭﺍ ﻳﺎﺩﺁﻭﺭﻱ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫• ﺍﻧﺠﺎﻡ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E1‬ﺑﺎﻋﺚ ﻧﻤﻲﺷﻮﺩ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺗﻐﻴﻴﺮ ﻛﻨﺪ‪،‬‬
‫• ﺍﻧﺠﺎﻡ ﻋﻤﻞ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪ E2‬ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺩﺭ ‪ −1‬ﺿﺮﺏ ﺷﻮﺩ )ﻗﺮﻳﻨﻪ ﺷﻮﺩ(‪،‬‬
‫• ﺩﺗﺮﻣﻴﻨﺎﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺮﺍﺑﺮ ﺣﺎﺻﻞﺿﺮﺏ ﻋﻨﺎﺻﺮ ﺭﻭﻱ ﻗﻄﺮ ﺍﺻﻠﻲ ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ‪.‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺳﻪ ﻧﻜﺘﻪ ﻓﻮﻕ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‪:‬‬
‫‪det(A) = (−1)r × u11 × u22 × · · · × unn‬‬ ‫)‪(۱۷.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ r‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻌﺪﺍﺩ ﺩﻓﻌﺎﺗﻲ ﻛﻪ ﻋﻤﻞ ‪ E2‬ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ‪ r‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻌﺪﺍﺩ ﺟﺎﺑﺠﺎﻳﻲ‬
‫ﺳﻄﺮﻫﺎ ﺩﺭ ﺧﻼﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ‪.A‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻣﺎﺗﺮﯾﺲ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ(‬ ‫ﻣﺜﺎﻝ ‪)۲۱−۴‬ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﯿﻨﺎﻥ ﯾ‬

‫ﻣﺎﺗﺮﻳﺲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬
‫‪‬‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪6‬‬ ‫‪2‬‬ ‫‪−5‬‬
‫‪A := ‬‬
‫‪−1‬‬
‫‪‬‬
‫‪‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪−4‬‬‫‪‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪3‬‬
‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﺎ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪ ،‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‪ ،‬ﻋﻨﺎﺻﺮ‬
‫ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﺭﺍ ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1 3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪R2:=−2R1+R2‬‬
‫‪‬‬
‫‪ 2 6 2‬‬ ‫‪−5‬‬ ‫‪‬‬
‫‪R3:=+1R1+R3‬‬
‫‪−2‬‬ ‫‪−19‬‬
‫‪‬‬ ‫‪ −−−−−−−−−→ 0‬‬
‫‪R4:=−1R1+R4‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪−1 3 4‬‬ ‫‪−4‬‬‫‪‬‬ ‫‪0‬‬ ‫‪3‬‬
‫‪‬‬ ‫‪E1‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪‬‬
‫‪1 2 1‬‬ ‫‪3‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﺑﻌﺪ ﺍﺯ ﺍﻧﺠﺎﻡ ﻣﺤﻮﺭﮔﻴﺮﻱ‪ ،‬ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺩﻭﻡ ﻧﻴﺰ ﺻﻔﺮ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬ ‫‪R3:=+6R2+R3 ‬‬ ‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬
‫‪−−−−−→ ‬‬ ‫‪ −−‬‬‫‪−−−−−−−→ ‬‬ ‫‪‬‬
‫‪R2⇆R4‬‬
‫‪0‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪−21‬‬
‫‪E2‬‬
‫‪‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪E1‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺳﻮﻡ‪ ،‬ﺗﻨﻬﺎ ﺑﻪ ﻳﻚ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪2‬‬ ‫‪7‬‬
‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪−1‬‬ ‫‪−1‬‬ ‫‪−4‬‬
‫‪−−−−−→ ‬‬ ‫‪‬‬
‫‪R3⇆R4‬‬
‫‪0‬‬ ‫‪−2‬‬ ‫‪−19‬‬
‫‪E2‬‬
‫‪‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪−21‬‬
‫ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺩﺭ ﻣﺮﺍﺣﻞ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺑﻪ ﺩﻭﺑﺎﺭ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻧﻴﺎﺯ ﭘﻴﺪﺍ ﻛﺮﺩﻳﻢ‪ .‬ﻟﺬﺍ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ )‪ (۱۷.۴‬ﺑﻪ‬
‫ﺍﺯﺍﻱ ‪ r = 2‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪det(A) = (−1)2 × (1) × (−1) × (−2) × (−21) = −42‬‬

‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ‪:‬‬

‫» ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ؟ «‬

‫ﭘﺎﺳﺦ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺑﻪ ﻣﻨﻈﻮﺭ ﻛﺎﻫﺶ ﺧﻄﺎ ﺩﺭ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺩﺗﺮﻣﻴﻨﺎﻥ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﻣﻨﺘﻬﻲ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ‪ r‬ﺩﺭ )‪ (۱۷.۴‬ﺑﺮﺍﺑﺮ ﺗﻌﺪﺍﺩ ﺩﻓﻌﺎﺕ ﺍﻧﺠﺎﻡ ﻣﺤﻮﺭﮔﻴﺮﻱ ﺟﺰﺋﻲ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﺤﺎﺳﺒﻪ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫�‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺑﻪ ﻧﺤﻮﻩ ﺍﺳﺘﺨﺮﺍﺝ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﻋﻤﻞ ﻣﺤﺎﺳﺒﻪ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺗﻮﺻﻴﻪ ﻧﻤﻲﺷﻮﺩ‪ .‬ﭼﺮﺍ‬
‫ﻛﻪ ﺩﺭ ﺣﻞ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻣﺴﺎﻳﻞ‪ ،‬ﺍﮔﺮ ﻣﻌﻜﻮﺱ ﻣﺎﺗﺮﻳﺲ ﻇﺎﻫﺮ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺪﻭﻥ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻌﻜﻮﺱ ﻣﺎﺗﺮﻳﺲ ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﻣﺴﺎﻟﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۱۲‬‬

‫ﺭﺍ ﺣﻞ ﻛﺮﺩ‪ .‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻭﻳﻠﻜﻴﻨﺴﻮﻥ ‪ ۶۱‬ﻣﻲﮔﻮﻳﺪ ”ﻫﺮ ﻣﺴﺎﻟﻪﺍﻱ ﺭﺍ ﻛﻪ ﺑﺘﻮﺍﻥ ﺑﺎ ﻣﻌﻜﻮﺱﮔﻴﺮﻱ ﺣﻞ ﻛﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺁﻥ ﻣﺴﺎﻟﻪ ﺭﺍ‬
‫ﻣﻲﺗﻮﺍﻥ ﺑﺪﻭﻥ ﻣﻌﻜﻮﺱﮔﻴﺮﻱ ﻧﻴﺰ ﺣﻞ ﻛﺮﺩ‪ “.‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺣﻞ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﻌﻜﻮﺱ ‪ A‬ﻭ‬
‫ﺑﻪ ﺻﻮﺭﺕ ‪ A−1 b‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺑﺪﻭﻥ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻌﻜﻮﺱ ‪ A‬ﻧﻴﺰ ﺭﻭﺵﻫﺎﻳﻲ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ‬
‫ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﺎ ﺍﻳﻦ ﻭﺟﻮﺩ‪ ،‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻧﺤﻮﻩ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻌﻜﻮﺱ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‪ ،‬ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ A‬ﻣﺎﺗﺮﻳﺴﻲ ﻣﺮﺑﻌﻲ ﻭ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻌﻜﻮﺱ ﺁﻥ ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﻣﻌﻜﻮﺱ ﻣﺎﺗﺮﻳﺲ‬
‫‪ A‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺯﻳﺮ ﺻﺪﻕ ﻣﻲﻛﻨﺪ‪:‬‬
‫‪AX = I‬‬ ‫)‪(۱۸.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ I‬ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺍﺳﺖ ﻭ ‪ X‬ﻣﺎﺗﺮﻳﺲ ﻣﺠﻬﻮﻝ ﺍﺳﺖ‪ .‬ﺧﺎﻃﺮ ﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﭼﻮﻥ ﻣﺠﻬﻮﻝ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﻳﻚ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﺍﻳﻦ ﻣﻌﺎﺩﻟﻪ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻣﻌﻜﻮﺱ ‪ A‬ﻛﺎﻓﻲ ﺍﺳﺖ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ ﻓﻮﻕ ﺭﺍ ﺣﻞ ﻛﻨﻴﻢ‪ .‬ﺍﻣﺎ ﺩﺳﺘﮕﺎﻩ ﻣﺎﺗﺮﻳﺴﻲ )‪ (۱۸.۴‬ﻣﻌﺎﺩﻝ ‪n‬‬
‫ﺩﺳﺘﮕﺎﻩ )ﺑﺮﺩﺍﺭﻱ( ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Axi = ei , i = 1, . . . , n.‬‬ ‫)‪(۱۹.۴‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ ei‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪i‬ﺍﻣﻴﻦ ﺳﺘﻮﻥ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ‪ ،i‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ‪i‬ﺍﻣﻴﻦ ﺳﺘﻮﻥ‬
‫ﻣﺎﺗﺮﻳﺲ ‪ X‬ﺍﺳﺖ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ‪ n‬ﺳﺘﻮﻥ ﻣﺎﺗﺮﻳﺲ ‪ X‬ﻛﺎﻓﻲ ﺍﺳﺖ ‪ n‬ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺭﺍ ﺣﻞ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ‪ n‬ﺩﺳﺘﮕﺎﻩ ﻣﻲﺗﻮﺍﻥ ﺍﺯ‬
‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ‪ n‬ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ n ،‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺯﻳﺮ ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪A‬‬ ‫‪ei , i = 1, . . . , n‬‬
‫ﻭ ﺳﭙﺲ ﺑﺎ ﺍﻋﻤﺎﻝ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‪ ،‬ﺍﻳﻦ ‪ n‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺭﺍ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪I‬‬ ‫‪xi , i = 1, . . . , n‬‬
‫ﺍﻣﺎ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺭ ﺗﻤﺎﻣﻲ ‪ n‬ﺩﺳﺘﮕﺎﻩ ﻳﻜﺴﺎﻥ ﺍﺳﺖ ﻭ ﻟﺬﺍ ‪ n‬ﺳﺘﻮﻥ ﺍﻭﻝ ﺗﻤﺎﻣﻲ ‪ n‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻳﻜﺴﺎﻥ‬
‫ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺩﺭ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ‪ ،‬ﺩﻗﻴﻘﺎ ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻳﻜﺴﺎﻧﻲ ﺭﻭﻱ ﺗﻤﺎﻣﻲ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪،‬‬
‫ﮔﻮﺱ‪-‬ﺟﺮﺩﻥ ﺭﺍ ﺑﻪ ‪h‬ﻃﻮﺭ ﻫﻤﺰﻣﺎﻥ ﺭﻭﻱ ﺍﻳﻦ ‪n‬‬
‫‪i‬‬ ‫ﻻﺯﻡ ﻧﻴﺴﺖ ﻛﻪ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺭﺍ ﺑﻪ ﻃﻮﺭ ﻣﺠﺰﺍ ﺣﻞ ﻛﻨﻴﻢ‪ .‬ﺑﻠﻜﻪ ﻣﻲﺗﻮﺍﻧﻴﻢ ﺭﻭﺵ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺍﻋﻤﺎﻝ ﻧﻤﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪ ،‬ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﺍﻓﺰﻭﺩﻩ ‪ A ei , i = 1, . . . , n‬ﺭﺍ ﺩﺭ ﻫﻢ ﺍﺩﻏﺎﻡ ﻧﻤﻮﺩﻩ ﻭ‬
‫ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺯﻳﺮ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﻳﻢ‪:‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪A‬‬ ‫‪I .‬‬
‫ﺣﺎﻝ ﺑﺎ ﺍﻋﻤﺎﻝ ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ‪ n ،‬ﺳﺘﻮﻥ ﺍﻭﻝ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪h‬‬ ‫‪ h‬ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ ‪i‬‬ ‫‪i‬‬
‫‪A‬‬ ‫‪I −−−−−−−−−−→ I‬‬ ‫‪B .‬‬
‫‪ E2 ، E1‬ﻭ ‪E3‬‬

‫ﺣﺎﻝ ﻣﺎﺗﺮﻳﺲ ‪ B‬ﻫﻤﺎﻥ ﻭﺍﺭﻭﻥ ‪ A‬ﺍﺳﺖ‪.‬‬

‫ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻭﺍﺭﻭﻥ ‪ A‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻋﻤﻞ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪h‬‬ ‫‪i‬‬
‫‪ .۱‬ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ‪ A I‬ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪.‬‬
‫‪h‬‬ ‫‪i‬‬ ‫‪h‬‬ ‫‪i‬‬
‫‪ .۲‬ﻋﻤﻠﻴﺎﺕ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ )ﻫﺮ ﺳﻪ ﻋﻤﻞ( ﺭﺍ ﺑﺮﺍﻱ ﺗﺒﺪﻳﻞ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ‪ A I‬ﺑﻪ ‪ I B‬ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪.‬‬
‫‪ .۳‬ﺍﮔﺮ ﺗﺒﺪﻳﻞ ﻗﺪﻡ ‪ ۲‬ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ‪ B‬ﻫﻤﺎﻥ ﻭﺍﺭﻭﻥ ‪ A‬ﺍﺳﺖ‪.‬‬
‫‪61 James‬‬ ‫‪H. Wilkinson‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪ .۴‬ﺍﻣﺎ ﺍﮔﺮ ﻧﺘﻮﺍﻳﻢ ﺗﺒﺪﻳﻞ ﻗﺪﻡ ‪ ۲‬ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ‪ A‬ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﺎﺗﺮﯾﺲ(‬ ‫ﻣﺜﺎﻝ ‪)۲۲−۴‬ﺍﺳﺘﺨﺮﺍﺝ ﻣﻌﮑﻮﺱ ﯾ‬

‫ﻣﺎﺗﺮﻳﺲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪1‬‬ ‫‪−1‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪A := 1‬‬ ‫‪0‬‬ ‫‪−1‬‬‫‪‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪2‬‬
‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﻭﺍﺭﻭﻥ ﻣﺎﺗﺮﻳﺲ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪.‬‬
‫ﺑﻌﺪ ﺍﺯ ﺗﺸﻜﻴﻞ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ‪ ،‬ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ‪ ،‬ﺍﺑﺘﺪﺍ‪ ،‬ﻣﺤﻮﺭ ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﻗﻄﺮ ﺩﺭ ﺳﺘﻮﻥ ﺍﻭﻝ ﺻﻔﺮ‬
‫ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2 1 -1 1 0 0‬‬ ‫‪1 12 - 12 12 0 0 R2:=-1R1+R2 1 21 - 12‬‬ ‫‪1‬‬
‫‪0‬‬ ‫‪0‬‬
‫‪‬‬ ‫‪ R1:= 21 R1 ‬‬ ‫‪ R3:=-1R1+R3 ‬‬ ‫‪2‬‬ ‫‪‬‬
‫‪1 0 -1 0 1 0 −‬‬ ‫‪−−→ ‬‬ ‫‪‬‬ ‫‪−−−→ ‬‬ ‫‪0 - 21 - 12 - 12 1 0‬‬
‫‪‬‬ ‫‪ −−E3‬‬ ‫‪1 0 -1 0 1 0 −−−−−‬‬ ‫‪E1‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2 -2 0 1‬‬
‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬
‫‪1 1 2 0 0 1‬‬ ‫‪1 1 2 0 0 1‬‬ ‫‪0 2‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ‪ ،‬ﻣﺤﻮﺭ ﺑﻪ ‪ 1‬ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﻋﻨﺎﺻﺮ ﺯﻳﺮ ﺁﻥ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪-‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪0 0‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪- 12‬‬ ‫‪1‬‬
‫‪0 0‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪- 12‬‬ ‫‪0 0‬‬‫‪1‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪R2:=-2R2‬‬ ‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪R3:= -1 R2+R3‬‬ ‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬‫‪‬‬
‫‪0 -‬‬ ‫‪- 1 0‬‬
‫‪-‬‬ ‫‪−−→ ‬‬ ‫‪‬‬ ‫‪−−−→ ‬‬ ‫‪1 1 -2 0‬‬
‫‪ −−−E3‬‬ ‫‪0 1 1 1 -2 0 −−−−−‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪E1‬‬ ‫‪0 1‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1‬‬
‫‪2‬‬ ‫‪- 0 1‬‬ ‫‪5‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬ ‫‪0 2 2 -2 0 1‬‬
‫‪1‬‬ ‫‪5‬‬ ‫‪1‬‬
‫‪0 0‬‬ ‫‪2 -1 1 1‬‬
‫ﺗﺎ ﺍﻳﻨﺠﺎ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺍﻛﻨﻮﻥ ﻣﻮﻟﻔﻪ )‪ (3, 3‬ﻣﺤﻮﺭ ﺍﺳﺖ‪ .‬ﻣﺤﻮﺭ ﺭﺍ ﺑﻪ ﻳﻚ ﺗﺒﺪﻳﻞ ﻛﺮﺩﻩ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺑﺎﻻﻳﺶ ﺭﺍ‬
‫ﺻﻔﺮ ﻣﻲﻛﻨﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪- 12‬‬ ‫‪1‬‬
‫‪0 0‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪- 12‬‬ ‫‪1‬‬
‫‪0 0‬‬ ‫‪R2:=-1R3+R2‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪ R3:= 12 R3 ‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪‬‬ ‫‪R1:= 12 R3+R1‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫‪4‬‬
‫‪0 1‬‬ ‫‪1 -2 0 ‬‬ ‫‪−−→ ‬‬ ‫‪1 -2 0 ‬‬ ‫‪−−−→ ‬‬ ‫‪-‬‬ ‫‪-‬‬ ‫‪1‬‬
‫‪ −−−E3‬‬ ‫‪ −−−−−‬‬
‫‪3‬‬ ‫‪5‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪0 1‬‬ ‫‪1‬‬ ‫‪0 1 0‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪E1‬‬
‫‪0 0‬‬ ‫‪2 -1 1 1‬‬ ‫‪0 0‬‬ ‫‪1 -2 2 2‬‬
‫‪1‬‬ ‫‪1 1‬‬
‫‪0 0 1 -‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬

‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺁﺧﺮ‪ ،‬ﻣﻮﻟﻔﻪ )‪ (2, 2‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺤﻮﺭ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﻣﻮﻟﻔﻪ ﺑﺎﻻﻱ ﺁﻥ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪0‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪4  R2:=-11R3+R2‬‬ ‫‪1 0‬‬ ‫‪0‬‬ ‫‪- 12‬‬ ‫‪3‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪4‬‬
‫‪R1:= 2 R3+R1‬‬ ‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪0‬‬ ‫‪1 0‬‬ ‫‪3‬‬
‫‪-‬‬ ‫‪5‬‬
‫‪-‬‬ ‫‪1 −‬‬‫→‪−−−−−−−‬‬ ‫‪- -‬‬ ‫‪0 1‬‬ ‫‪0‬‬ ‫‪3‬‬ ‫‪5‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪E1‬‬ ‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪0‬‬ ‫‪-‬‬ ‫‪0 1‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬ ‫‪1 -‬‬ ‫‪0 0‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬

‫ﺩﺭ ﻧﻬﺎﻳﺖ‪ ،‬ﺳﻪ ﺳﺘﻮﻥ ﺍﻭﻝ ﻣﺎﺗﺮﻳﺲ ﺍﻓﺰﻭﺩﻩ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻭ ﻟﺬﺍ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥ ﺩﺭ ﺳﻪ ﺳﺘﻮﻥ ﺁﺧﺮ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪- 12‬‬ ‫‪3‬‬ ‫‪1‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪A−1 = ‬‬
‫‪‬‬
‫‪3‬‬
‫‪2‬‬ ‫‪-‬‬ ‫‪5‬‬
‫‪2‬‬ ‫‪-‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪-‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬
‫‪1‬‬
‫‪2‬‬

‫ﺗﺸﺨﻴﺺ )ﻧﻴﻤﻪ(ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﻮﺩﻥ ﻣﺎﺗﺮﻳﺲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ‬ ‫�‬
‫ﺩﻳﺪﻳﻢ ﻛﻪ ﺗﺸﺨﻴﺺ ﻣﻌﻴﻦ ﻳﺎ ﻧﻤﻴﻪﻣﻌﻴﻦ ﺑﻮﺩﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ‪ ،‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻌﺮﻳﻒ‪ ،‬ﺳﺎﺩﻩ ﻧﻴﺴﺖ‪ .‬ﺧﺼﻮﺻﺎ ﺍﮔﺮ ﺑﻌﺪ ﻣﺎﺗﺮﻳﺲ ﺑﺰﺭﮒ‬
‫ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﻗﻀﻴﻪ ﺯﻳﺮ ﺧﺎﺻﻴﺘﻲ ﺍﺯ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻴﻦ ﻳﺎ ﻧﻴﻤﻪﻣﻌﻴﻦ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ ﻛﻪ ﺑﻪ ﻛﻤﻚ ﺁﻥ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻃﻮﺭ ﺳﺎﺩﻩﺗﺮﻱ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۱۴‬‬

‫ﺗﺸﺨﻴﺺ ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ﻣﻌﻴﻦ ﻭ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪.‬‬

‫ﻣﺎﺗﺮﯾﺲﻫﺎﯼ ﻣﻌﯿﻦ ﻭ ﻧﯿﻤەﻣﻌﯿﻦ ﻣﺜﺒﺖ ﻭ ﺭﻭﺵ ﺣﺬﻓ ﮔﻮﺱ‬ ‫ﻗﻀﯿﻪ ‪۱۲−۴‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ ﺑﺎﺷﺪ‪ .‬ﺁﻥﮔﺎﻩ‪:‬‬

‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪ ،‬ﺍﮔﺮ ﻭ ﺗﻨﻬﺎ ﺍﮔﺮ‪ ،‬ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺭﻭﻱ ‪ A‬ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻧﻴﺎﺯﻱ‬
‫ﻣﺜﺒﺖ‬
‫ﻧﺎﻣﻨﻔﻲ ﺑﺎﺷﻨﺪ‪.‬‬ ‫ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﻋﻨﺎﺻﺮ ﻣﺤﻮﺭﻱ‬
‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ A ،‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪ ،‬ﻫﺮﮔﺎﻩ ﺗﻨﻬﺎ ﺑﺎ ﻋﻤﻞ ‪ E1‬ﻣﺎﺗﺮﻳﺲ ‪ A‬ﺑﻪ ﻣﺎﺗﺮﻳﺴﻲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ‬
‫ﻣﺜﺒﺖ‬
‫ﻧﺎﻣﻨﻔﻲ ﺗﺒﺪﻳﻞ ﮔﺮﺩﺩ‪.‬‬ ‫ﻗﻄﺮﻱ‬

‫ﻣﻄﺎﺑﻖ ﺑﺎ ﻗﻀﻴﻪ ﻓﻮﻕ‪ ،‬ﺑﺮﺍﻱ ﺗﺸﺨﻴﺺ ﻣﻌﻴﻦ ﻳﺎ ﻧﻤﻴﻪﻣﻌﻴﻦ ﺑﻮﺩﻥ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﺘﻘﺎﺭﻥ‪ ،‬ﻛﺎﻓﻲ ﺍﺳﺖ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﺎ‬
‫ﻋﻤﻞ ‪ E1‬ﺭﺍ ﺭﻭﻱ ﻣﺎﺗﺮﻳﺲ ﺍﻋﻤﺎﻝ ﻛﻨﻴﻢ‪ .‬ﺍﮔﺮ ﺩﺭ ﻳﻚ ﻣﺮﺣﻠﻪ ﺍﺯ ﺭﻭﺵ ﺣﺬﻓﻲ ﮔﻮﺱ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻧﻴﺎﺯ ﺩﺍﺷﺘﻴﻢ ﻭ ﻳﺎ ﺑﻪ ﻣﺤﻮﺭ‬
‫ﻣﻨﻔﻲ ﺑﺮﺧﻮﺭﺩﻳﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻭ ﻳﺎ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺑﺎ ﻋﻨﺎﺻﺮ‬
‫ﻣﻌﻴﻦ ﻣﺜﺒﺖ‬ ‫ﻣﺜﺒﺖ‬
‫ﻧﺎﻣﻨﻔﻲ ﺗﺒﺪﻳﻞ ﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬ ‫ﻗﻄﺮﻱ‬

‫ﻣﺎﺗﺮﯾﺲ(‬ ‫ﻣﺜﺎﻝ ‪)۲۳−۴‬ﺗﺸﺨﯿﺺ ﻣﻌﯿﻦ ﯾﺎ ﻧﯿﻤەﻣﻌﯿﻦ ﻭ ﯾﺎ ﻧﺎﻣﻌﯿﻦ ﺑﻮﺩﻥ ﯾ‬

‫ﻣﺎﺗﺮﻳﺲﻫﺎﻱ ‪ C ،B ،A‬ﻭ ‪ D‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪4 −2‬‬ ‫‪2 4 −2‬‬ ‫‪2 4 −2‬‬ ‫‪2 4 −2‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪A := ‬‬
‫‪ 4‬‬ ‫‪8 9‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ , B :=  4 1 −3 , C :=  4 9 −3 , D :=  4 9 −3‬‬
‫‪‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪9 10‬‬ ‫‪−2 −3 3‬‬ ‫‪−2 −3 7‬‬ ‫‪−2 −3 3‬‬
‫◀ ﺩﺭﻣﺮﺣﻠﻪ ﺍﻭﻝ ﺍﺯ ﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ‪ ،A‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬
‫‪‬‬ ‫‪ R2:=−2R1+R2‬‬
‫‪R3:=+1R1+R3 ‬‬ ‫‪‬‬
‫‪A := ‬‬
‫‪ 4‬‬ ‫‪8‬‬ ‫‪9‬‬
‫‪ −−−−−−−→ 0‬‬
‫‪−‬‬‫‪−‬‬ ‫‪‬‬ ‫‪0‬‬ ‫‪13‬‬‫‪‬‬
‫‪E1‬‬
‫‪−2‬‬ ‫‪9‬‬ ‫‪10‬‬ ‫‪0 13‬‬ ‫‪8‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﺑﻪ ﻣﺤﻮﺭ ﺻﻔﺮ ﺑﺮﺧﻮﺭﺩﻳﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻋﻨﺼﺮ ﺯﻳﺮ ﻣﺤﻮﺭ ﻏﻴﺮﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺑﻪ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬
‫ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻧﺎﻣﻌﻴﻦ ﺍﺳﺖ‪.‬‬
‫◀ ﺩﺭ ﺍﻭﻟﻴﻦ ﻣﺮﺣﻠﻪ ﺍﺯ ﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ‪ B‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪−2 R2:=−2R1+R2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬
‫‪‬‬ ‫‪ R3:=+1R1+R3 ‬‬ ‫‪‬‬
‫‪B := ‬‬
‫‪ 4‬‬ ‫‪1‬‬ ‫‪‬‬
‫‪−3 −−−−−−−−−→ 0‬‬‫‪‬‬ ‫‪−7‬‬ ‫‪1‬‬‫‪‬‬
‫‪E1‬‬
‫‪−2‬‬ ‫‪−3‬‬ ‫‪3‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪1‬‬
‫ﺩﺭ ﻣﺮﺣﻠﻪ ﺍﻭﻝ ﻣﺤﻮﺭ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﻣﺮﺣﻠﻪ ﺩﻭﻡ ﻣﺤﻮﺭ ﻋﺪﺩﻱ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﻧﺎﻣﻌﻴﻦ ﺍﺳﺖ‪.‬‬
‫◀ ﺩﺭ ﺯﻳﺮ ﻣﺮﺍﺣﻞ ﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ‪ C‬ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪−2 R2:=−2R1+R2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬
‫‪‬‬ ‫‪ R3:=+1R1+R3 ‬‬ ‫‪ R3:=−1R2+R3 ‬‬ ‫‪‬‬
‫‪C := ‬‬
‫‪ 4‬‬ ‫‪9‬‬ ‫‪‬‬
‫‪−3 −−−−−−−−−→ 0‬‬‫‪‬‬ ‫‪1‬‬ ‫‪‬‬
‫‪1 −−−−−−−−−→ 0‬‬‫‪‬‬ ‫‪1‬‬ ‫‪1‬‬‫‪‬‬
‫‪E1‬‬ ‫‪E1‬‬
‫‪−2‬‬ ‫‪−3‬‬ ‫‪7‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪5‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪4‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻭ ﺗﻨﻬﺎ ﺑﺎ ﻋﻤﻞ ‪ E1‬ﻣﺎﺗﺮﻳﺲ ‪ C‬ﺑﻪ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﺗﺒﺪﻳﻞ ﺷﺪ‪ .‬ﭼﻮﻥ ﺍﻋﻀﺎﻱ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱۵‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﺭﻭﻱ ﻗﻄﺮ ﻣﺎﺗﺮﻳﺲ ﺑﺎﻻﻣﺜﻠﺜﻲ ﻣﺜﺒﺖ ﺍﺳﺖ‪ ،‬ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ‪ C‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪.‬‬
‫◀ ﻣﺮﺍﺣﻞ ﻣﺜﻠﺜﻲ ﻛﺮﺩﻥ ‪ D‬ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪2‬‬ ‫‪4‬‬ ‫‪−2‬‬ ‫‪2 4‬‬ ‫‪−2‬‬ ‫‪2 4‬‬ ‫‪−2‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪R2:=−2R1+R2‬‬ ‫‪ R3:=−1R2+R3 ‬‬ ‫‪‬‬
‫‪D := ‬‬ ‫‪−3‬‬ ‫‪‬‬ ‫‪1‬‬ ‫‪‬‬ ‫‪1‬‬
‫‪R3:=+1R1+R3‬‬
‫‪ 4‬‬ ‫‪9‬‬ ‫‪ −−−−−−−−−→ 0 1‬‬
‫‪E1‬‬ ‫‪ −−−−−−−−→ 0 1‬‬
‫‪E1‬‬ ‫‪‬‬
‫‪−2‬‬ ‫‪−3‬‬ ‫‪3‬‬ ‫‪0 1‬‬ ‫‪1‬‬ ‫‪0 0‬‬ ‫‪0‬‬
‫ﻣﺎﺗﺮﻳﺲ ‪ D‬ﺑﺪﻭﻥ ﻣﺤﻮﺭﮔﻴﺮﻱ ﻣﺜﻠﺜﻲ ﺷﺪ‪ .‬ﻳﻚ ﻋﻀﻮ ﻗﻄﺮﻱ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﻣﺜﻠﺜﻲ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﻭ ﺑﻘﻴﻪ ﻣﺜﺒﺖ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪،‬‬
‫ﻣﺎﺗﺮﻳﺲ ‪ D‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻧﻴﻤﻪﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺍﺳﺖ‪.‬‬

‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ‬ ‫‪۷.۴‬‬

‫ﺗﺎ ﺍﻳﻨﺠﺎ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪ‪ .‬ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﺁﻥ ﺩﺳﺘﻪ ﺭﻭﺵﻫﺎﻳﻲ ﻫﺴﺘﻨﺪ ﻛﻪ‬
‫ﺗﻌﺪﺍﺩ ﻋﻤﻠﻴﺎﺕ ﻻﺯﻡ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺍﺯ ﺍﺑﺘﺪﺍ ﻣﺸﺨﺺ ﺍﺳﺖ‪ .‬ﻣﺜﻼ ﺩﺭ ﺭﻭﺵ ﮔﻮﺱ ﻛﻪ ﻳﻚ ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ﺍﺳﺖ‪ ،‬ﻛﺎﻣﻼ ﻣﺸﺨﺺ‬
‫ﺍﺳﺖ ﻛﻪ ﭼﻪ ﺗﻌﺪﺍﺩ ﺟﻤﻊ ﻭ ﺗﻔﺮﻳﻖ ﻭ ﭼﻪ ﺗﻌﺪﺍﺩ ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺑﺮﺍﻱ ﺣﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ‪ n × n‬ﻻﺯﻡ ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﻣﻘﺎﺑﻞ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻗﺮﺍﺭ ﺩﺍﺭﻧﺪ ﻛﻪ ﺑﺎ ﻛﻤﻚ ﺗﻜﺮﺍﺭ ﺑﻪ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻣﻲﭘﺮﺩﺍﺯﻧﺪ‪ .‬ﺩﺭ‬
‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ‪ ،‬ﻋﻤﻮﻣﺎ ﺗﻌﺪﺍﺩ ﺗﻜﺮﺍﺭﻫﺎﻱ ﻻﺯﻡ ﺑﺮﺍﻱ ﺭﺳﻴﺪﻥ ﺑﻪ ﻳﻚ ﺩﻗﺖ ﺧﺎﺹ ﺍﺯ ﻗﺒﻞ ﻣﺸﺨﺺ ﻧﻴﺴﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﻋﻠﺖ‬
‫ﺩﺭ ﺍﺑﺘﺪﺍﻱ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻣﺸﺨﺺ ﻧﻤﻲﺑﺎﺷﺪ ﻛﻪ ﭼﻪ ﺗﻌﺪﺍﺩ ﻋﻤﻠﻴﺎﺕ ﺑﺎﻳﺪ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻧﻜﺘﻪ ﻭﺟﻪ ﺗﻤﺎﻳﺰ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻭ‬
‫ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺍﺯ ﺗﻜﺮﺍﺭ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻣﻲﻛﻨﻨﺪ‪ .‬ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻣﻲﮔﻴﺮﻧﺪ ﻭ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺭﺍ ﺩﺭ ﺗﻜﺮﺍﺭﻫﺎﻱ ﻣﺘﻮﺍﻟﻲ ﺑﻬﺒﻮﺩ ﻣﻲﺩﻫﻨﺪ‬
‫ﺗﺎ ﺑﻪ ﻳﻚ ﺟﻮﺍﺏ ﻧﺴﺒﺘﺎ ﺧﻮﺏ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﺑﺮﺳﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ‪ ،‬ﻳﻚ ﺩﻧﺒﺎﻟﻪ‬
‫ﺍﺯ ﺑﺮﺩﺍﺭﻫﺎ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ ﻛﻪ ﺩﺭﺻﻮﺭﺗﻲ ﻛﻪ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻫﻤﮕﺮﺍ ﮔﺮﺩﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻣﺘﻔﺎﻭﺗﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻫﺮﻛﺪﺍﻡ ﺑﻪ ﺻﻮﺭﺕ ﻣﺘﻔﺎﻭﺗﻲ ﺩﻧﺒﺎﻟﻪ ﺫﻛﺮ ﺷﺪﻩ ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺍﺯ ﺟﻤﻠﻪ ﺭﻭﺵﻫﺎﻱ‬
‫ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻣﻮﺍﺭﺩ ﺯﻳﺮ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪:‬‬
‫ﮊﺍﻛﻮﺑﻲ ‪۶۲‬‬ ‫• ﺭﻭﺵ‬

‫ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ‪۶۳‬‬ ‫• ﺭﻭﺵ‬

‫ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‬ ‫�‬

‫ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬

‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪a11 x1 + a12 x2 + · · · + a1n xn = b1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪a21 x1 + a22 x2 + · · · + a2n xn = b2‬‬
‫‪..‬‬ ‫‪..‬‬ ‫)‪(۲۰.۴‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪an1 x1 + an2 x2 + · · · + ann xn = bn‬‬

‫‪62 Jacobi‬‬ ‫‪63 Gauss-Seidel‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۱۶‬‬

‫ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺎﺷﺪ‪ .‬ﺑﺪﻭﻥ ﺍﺯ ﺩﺳﺖ ﺩﺍﺩﻥ ﻛﻠﻴﺖ ﻣﺴﺎﻟﻪ ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺿﺮﻳﺐ ‪ xi‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ‪i‬ﺍﻡ‬
‫ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺑﺎﺷﺪ )‪ .(aii ̸= 0‬ﭼﺮﺍ ﻛﻪ ﭼﻨﻴﻦ ﺣﺎﻟﺘﻲ ﺩﺭ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻧﺎﻣﻨﻔﺮﺩ ﺑﺮﻗﺮﺍﺭ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﻣﻌﺎﺩﻻﺕ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻟﻲ ﺭﺳﻴﺪ ﻛﻪ ﺿﺮﻳﺐ ‪ xi‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ‪i‬ﺍﻡ ﻣﺨﺎﻟﻒ ﺻﻔﺮ ﺑﺎﺷﺪ‪.‬‬
‫ﺣﺎﻝ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ‪ x1‬ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﻴﻢ ﻭ ﻫﻤﭽﻨﻴﻦ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﺗﺎ ‪n‬ﺍﻡ ﺑﻪ ﺗﺮﺗﻴﺐ ‪ x2‬ﺗﺎ ‪ xn‬ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬
‫‪‬‬ ‫‪b1 − a12 x2 − · · · − a1n xn‬‬
‫‪‬‬
‫‪‬‬ ‫= ‪x1‬‬
‫‪‬‬
‫‪‬‬ ‫‪a11‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪b‬‬ ‫‪−‬‬ ‫‪a‬‬ ‫‪x‬‬ ‫‪− · · · − a2n xn‬‬
‫= ‪ x2‬‬
‫‪‬‬
‫‪2‬‬ ‫‪21‬‬ ‫‪1‬‬
‫‪a22‬‬ ‫)‪(۲۱.۴‬‬
‫‪ ..‬‬
‫‪‬‬
‫‪‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪b − an1 x1 − · · · − an,n-1 xn-1‬‬
‫‪ xn = n‬‬
‫‪ann‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﻌﺎﺩﻻﺕ ﻓﻮﻕ ﺍﺯ ﺩﺳﺘﻜﺎﺭﻱ ﺩﺳﺘﮕﺎﻩ )‪ (۲۰.۴‬ﺣﺎﺻﻞ ﺷﺪﻩ ﺍﺳﺖ ﻭ ﺑﺎ ﺁﻥ ﻣﻌﺎﺩﻝ ﺍﺳﺖ‪ .‬ﺗﺒﻌﺎ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ )‪(۲۰.۴‬‬
‫ﺩﺭ ﻣﻌﺎﺩﻻﺕ ﻓﻮﻕ ﻧﻴﺰ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‪ ،‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺍﺑﻂ ﻓﻮﻕ‪ ،‬ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪b1 − a12 x2 − · · · − a1n xn‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪ 1‬‬ ‫‪a11‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪b2 − a21 x1 − · · · − a2n xn‬‬
‫)‪x(k+1‬‬
‫‪2‬‬ ‫=‬
‫‪a22‬‬ ‫)‪(۲۲.۴‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪bn − an1 x1 − · · · − an,n-1 xn-1‬‬
‫‪‬‬
‫)‪x(k+1‬‬ ‫=‬
‫‪n‬‬
‫‪ann‬‬
‫)‪ x(0) := [x(0‬ﻳﻚ ﺗﻘﺮﻳﺐ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ )‪ (۲۰.۴‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ‬ ‫ﺣﺎﻝ ﺍﮔﺮ ] ‪1 , x2 , . . . , xn‬‬
‫)‪(0‬‬ ‫⊤ )‪(0‬‬

‫ﺍﻳﻦ ﺟﻮﺍﺏ ﺩﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﻓﻮﻕ ﺑﻪ ﺍﺯﺍﻱ ‪ ،k = 0‬ﺑﻪ ﺗﻘﺮﻳﺐ ﺩﻳﮕﺮ ⊤] ‪x(1) := [x1 , x2 , . . . , xn‬‬
‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬

‫ﺧﻮﺍﻫﻴﻢ ﺭﺳﻴﺪ‪ .‬ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ )‪ x(1‬ﺗﻘﺮﻳﺐ ﺑﻬﺘﺮﻱ ﻧﺴﺒﺖ ﺑﻪ )‪ x(0‬ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺑﺎﺷﺪ‪ .‬ﺣﺎﻝ ﻣﺠﺪﺩ‪ ،‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ‬
‫ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﺑﻪ ﺍﺯﺍﻱ ‪ ،k = 1‬ﻣﻲﺗﻮﺍﻥ )‪ x(1‬ﺭﺍ ﺑﻪ ﺗﻘﺮﻳﺐ )‪ x(2‬ﺗﺒﺪﻳﻞ ﻛﺮﺩ ﻭ ﺑﻪ ﻫﻤﻴﻦ ﺗﺮﺗﻴﺐ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺍﺯ ﺑﺮﺩﺍﺭﻫﺎ ﺑﻪ ﺻﻮﺭﺕ‬
‫‬ ‫∞‬
‫‪ x(k) k=0‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ∗‪ x‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ∗‪ x‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ‬
‫)‪ (۲۰.۴‬ﺍﺳﺖ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﺎ ﺭﻭﺵ ﮊﺍﮐﻮﺑﯽ(‬ ‫ﻣﺜﺎﻝ ‪)۲۴−۴‬ﺣﻞ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪2x1 + x2 + x3 − x4 = 3‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪x1 + 5x2 + x3 + 2x4 = 22‬‬
‫‪‬‬
‫‪‬‬ ‫‪x2 + 2x3 + x4 = 12‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪x − x + 4x = 14‬‬
‫‪1‬‬ ‫‪3‬‬ ‫‪4‬‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ⊤]‪ x(0) := [0, 0, 0, 0‬ﺭﺍ ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻛﺎﺭ ﺑﺒﺮﻳﻢ‪ .‬ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﺑﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱۷‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪3 − x2 − x3 + x4‬‬
‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬
‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪ (k+1‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪22‬‬ ‫‪−‬‬ ‫‪x‬‬ ‫‪−‬‬ ‫‪x3 − 2x4‬‬
‫‪ x2‬‬ ‫=‬ ‫‪1‬‬
‫‪5‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪x(k+1) = 12 − x2 − x4‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪3‬‬
‫‪‬‬
‫‪‬‬ ‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫)‪ (k+1‬‬
‫‪‬‬
‫)‪(k‬‬
‫‪14 − x1 + x3‬‬
‫)‪(k‬‬
‫‪ x4‬‬ ‫=‬
‫‪4‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﭼﻨﺪ ﺗﻜﺮﺍﺭ ﺍﺯ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺭﺍ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ‪.‬‬
‫◀ ﺗﻜﺮﺍﺭ ﺍﻭﻝ‪ :‬ﺩﺭ ﺍﺑﺘﺪﺍﻱ ﺗﻜﺮﺍﺭ ﺍﻭﻝ‪ ،‬ﺗﻘﺮﻳﺐ ﺍﻭﻟﻴﻪ ⊤]‪ x := [0, 0, 0, 0‬ﺩﺭ ﺩﺳﺖ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺑﺎ ﻛﻤﻚ ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﻓﻮﻕ‬
‫)‪(0‬‬

‫ﺑﻪ ﺍﺯﺍﻱ ‪ k = 0‬ﺗﻘﺮﻳﺐ ﺩﻳﮕﺮ )‪ x(1‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﻴﻢ‪:‬‬

‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(0‬‬ ‫)‪(0‬‬
‫‪3 − x2 − x3 + x4‬‬
‫)‪(0‬‬
‫‪3−0−0+0‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(1‬‬
‫=‬ ‫=‬ ‫‪= 1.5‬‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬
‫‪2‬‬ ‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(0‬‬ ‫)‪(0‬‬ ‫)‪(0‬‬
‫)‪ (1‬‬
‫‪‬‬ ‫‪22 − x1 − x3 − 2x4‬‬ ‫)‪22 − 0 − 0 − 2(0‬‬
‫= ‪ x2‬‬ ‫=‬ ‫‪= 4.4‬‬
‫‪5‬‬ ‫‪5‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(0‬‬
‫‪12 − x2 − x4‬‬
‫)‪(0‬‬
‫‪12 − 0 − 0‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(1‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬ ‫‪3‬‬ ‫=‬ ‫=‬ ‫‪= 6.0‬‬
‫‪‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(0‬‬
‫‪14 − x1 + x3‬‬
‫)‪(0‬‬
‫‪14 − 0 + 0‬‬
‫)‪x(1‬‬
‫‪4‬‬ ‫=‬ ‫=‬ ‫‪= 3.5‬‬
‫‪4‬‬ ‫‪4‬‬
‫ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺗﻘﺮﻳﺐ ⊤]‪ x(0) := [0, 0, 0, 0‬ﺑﻪ ﺗﻘﺮﻳﺐ ﺟﺪﻳﺪ ⊤]‪ x(1) := [1.5, 4.4, 6.0, 3.5‬ﻣﻨﺠﺮ ﺷﺪ‪.‬‬
‫◀ ﺗﻜﺮﺍﺭ ﺩﻭﻡ‪ :‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﻌﺎﺩﻻﺕ ﺗﻜﺮﺍﺭ ﺑﻪ ﺍﺯﺍﻱ ‪ k = 1‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(1‬‬ ‫)‪(1‬‬
‫‪3 − x2 − x3 + x4‬‬
‫)‪(1‬‬
‫‪3 − 4.4 − 6.0 + 3.5‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(2‬‬
‫=‬ ‫=‬ ‫‪= −1.95‬‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬
‫‪2‬‬ ‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬ ‫)‪(1‬‬
‫)‪ (2‬‬
‫‪‬‬ ‫‪22 − x1 − x3 − 2x4‬‬ ‫)‪22 − 1.5 − 6.0 − 2(3.5‬‬
‫= ‪ x2‬‬ ‫=‬ ‫‪= 1.5‬‬
‫‪5‬‬ ‫‪5‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(1‬‬
‫‪12 − x2 − x4‬‬
‫)‪(1‬‬
‫‪12 − 4.4 − 3.5‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(2‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬ ‫‪3‬‬ ‫=‬ ‫=‬ ‫‪= 2.05‬‬
‫‪‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(1‬‬
‫‪14 − x1 + x3‬‬
‫)‪(1‬‬
‫‪14 − 1.5 + 6.0‬‬
‫)‪x(2‬‬
‫‪4‬‬ ‫=‬ ‫=‬ ‫‪= 4.625‬‬
‫‪4‬‬ ‫‪4‬‬
‫ﻟﺬﺍ ﺩﺭ ﺗﻜﺮﺍﺭ ﺩﻭﻡ‪ ،‬ﺗﻘﺮﻳﺐ ⊤]‪ x(2) := [−1.95, 1.5, 2.05, 4.625‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﺎ ﺍﺩﺍﻣﻪ ﺗﻜﺮﺍﺭﻫﺎ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺑﺮﺩﺍﺭﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪ ‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪1.5‬‬ ‫‪−1.95‬‬ ‫‪2.0375‬‬ ‫‪1.01625‬‬ ‫‪1.3175000‬‬ ‫‪0.6612187‬‬ ‫‪1.0601094‬‬ ‫‪0.9824391‬‬
‫‪ ‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0 (1) 4.4 (2)  1.500  (3) 2.5300 (4) 1.60500 (5) 2.2097500 (6) 1.9226250 (7) 2.0737000 (8) 1.9506469‬‬
‫‪x(0)=‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫· · · ‪ , x =6.0, x = 2.050 , x =2.9375, x =2.48500, x =3.3350000, x =2.9615313, x =3.0365000, x =2.9256109 ,‬‬
‫‪0‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪0‬‬ ‫‪3.5‬‬ ‫‪4.625‬‬ ‫‪4.5000‬‬ ‫‪3.72500‬‬ ‫‪3.8671875‬‬ ‫‪4.0043750‬‬ ‫‪4.0750781‬‬ ‫‪3.9940977‬‬
‫ﺑﺮﺍﻱ ﻧﻤﺎﻳﺶ ﻣﻨﺎﺳﺐﺗﺮ ﻣﺮﺍﺣﻞ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺣﺎﺻﻞ ﺩﺭ ﻫﺮ ﺗﻜﺮﺍﺭ ﺭﺍ ﺩﺭ ﺳﻄﺮﻫﺎﻱ ﻳﻚ ﺟﺪﻭﻝ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮﮔﺰﺍﺭﺵ ﻧﻤﻮﺩ‪:‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۱۸‬‬

‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬

‫‪k‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫ﻧﺘﺎﻳﺞ ﻫﺮ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺩﺭ ﻳﻚ ﺳﻄﺮ‬
‫‪0‬‬ ‫‪0.0000000‬‬ ‫‪0.0000000‬‬ ‫‪0.0000000‬‬ ‫‪0.0000000‬‬
‫‪1‬‬ ‫‪1.5000000‬‬ ‫‪4.4000000‬‬ ‫‪6.0000000‬‬ ‫‪3.5000000‬‬ ‫ﺍﺯ ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺑﻪ ﺧﺎﻃﺮ‬
‫‪2‬‬ ‫‪-1.950000‬‬ ‫‪1.5000000‬‬ ‫‪2.0500000‬‬ ‫‪4.6250000‬‬ ‫ﺧﻼﺻﻪ ﻧﻮﻳﺴﻲ ﻓﻘﻂ ‪ ۷‬ﺭﻗﻢ ﺍﻋﺸﺎﺭ ﺍﺯ ﺍﻋﺪﺍﺩ‬
‫‪3‬‬ ‫‪2.0375000‬‬ ‫‪2.5300000‬‬ ‫‪2.9375000‬‬ ‫‪4.5000000‬‬
‫‪4‬‬ ‫‪1.0162500‬‬ ‫‪1.6050000‬‬ ‫‪2.4850000‬‬ ‫‪3.7250000‬‬ ‫ﺩﺭ ﺟﺪﻭﻝ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ‬
‫‪5‬‬ ‫‪1.3175000‬‬ ‫‪2.2097500‬‬ ‫‪3.3350000‬‬ ‫‪3.8671875‬‬ ‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺗﻜﺮﺍﺭﻫﺎﻱ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‬
‫‪6‬‬ ‫‪0.6612187‬‬ ‫‪1.9226250‬‬ ‫‪2.9615313‬‬ ‫‪4.0043750‬‬ ‫ﺩﺭ ﺣﺎﻝ ﻫﻤﮕﺮﺍﻳﻲ ﺑﻪ ﺑﺮﺩﺍﺭ‬
‫‪7‬‬ ‫‪1.0601094‬‬ ‫‪2.0737000‬‬ ‫‪3.0365000‬‬ ‫‪4.0750781‬‬ ‫‪ ‬‬
‫‪8‬‬ ‫‪0.9824391‬‬ ‫‪1.9506469‬‬ ‫‪2.9256109‬‬ ‫‪3.9940977‬‬ ‫‪1‬‬
‫‪9‬‬ ‫‪1.0589199‬‬ ‫‪2.0207509‬‬ ‫‪3.0276277‬‬ ‫‪3.9857930‬‬ ‫‪ ‬‬
‫‪2‬‬
‫‪10‬‬ ‫‪0.9687071‬‬ ‫‪1.9883733‬‬ ‫‪2.9967280‬‬ ‫‪3.9921770‬‬ ‫‪x∗ = ‬‬
‫‪3‬‬
‫‪‬‬
‫‪11‬‬ ‫‪1.0035378‬‬ ‫‪2.0100422‬‬ ‫‪3.0097249‬‬ ‫‪4.0070052‬‬ ‫‪ ‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪4‬‬
‫‪25‬‬ ‫‪1.0000067‬‬ ‫‪2.0000059‬‬ ‫‪3.0000048‬‬ ‫‪4.0000036‬‬ ‫ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﻫﻤﺎﻥ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪29‬‬ ‫‪1.0000007‬‬ ‫‪2.0000008‬‬ ‫‪3.0000008‬‬ ‫‪4.0000003‬‬ ‫ﺩﺳﺘﮕﺎﻩ ﺍﺳﺖ‪.‬‬

‫■ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﺸﺎﻫﺪﻩ ﺷﺪ‪ ،‬ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺍﺯ ﺑﺮﺩﺍﺭﻫﺎ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﺪ ﻭ ﺍﮔﺮ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﻋﻤﻮﻣﺎ ﺩﺭ ﺑﻲﻧﻬﺎﻳﺖ‬
‫ﺑﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺮﺍﻱ ﺭﺳﻴﺪﻥ ﺑﻪ ﺟﻮﺍﺏ ﺩﻗﻴﻖ ﺩﺳﺘﮕﺎﻩ ﺑﺎﻳﺪ ﺑﻲﻧﻬﺎﻳﺖ ﺗﻜﺮﺍﺭ ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ‬
‫ﻣﺎﺷﻴﻦﻫﺎ ﻛﻪ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺍﻋﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﺍﺳﺖ‪ ،‬ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺑﺎﻳﺪ ﺁﻧﻘﺪﺭ ﺗﻜﺮﺍﺭﻫﺎ ﺭﺍ ﺍﺩﺍﻣﻪ ﺩﻫﻴﻢ ﻛﻪ ﺗﻤﺎﻡ ﺍﺭﻗﺎﻡ ﻣﺘﻐﻴﺮﻫﺎ ﻳﻜﺴﺎﻥ‬
‫ﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺗﻜﺮﺍﺭﻫﺎ ﺗﺎ ﺁﻥﺟﺎ ﺍﺩﺍﻣﻪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ ﻛﻪ ﺣﺎﻟﺖ ﺍﻳﺪﻩﺁﻝ ﻣﺎﺷﻴﻨﻲ ﺯﻳﺮ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪:‬‬
‫)‪(k+1‬‬ ‫)‪(k‬‬
‫‪xi‬‬ ‫‪= xi , i = 1, . . . , n.‬‬
‫ﺍﻣﺎ ﻧﻜﺘﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺧﺎﻃﺮ ﻭﺟﻮﺩ ﺧﻄﺎﻱ ﻣﺎﺷﻴﻨﻲ‪ ،‬ﺣﺎﻟﺖ ﻓﻮﻕ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺗﻔﺎﻕ ﻧﻴﺎﻓﺘﺪ ﻭ ﻳﺎ ﺍﻳﻦ ﻛﻪ ﺑﺮﺍﻱ ﺭﺳﻴﺪﻥ ﺑﻪ ﺣﺎﻟﺖ‬
‫ﻓﻮﻕ ﺑﻪ ﺗﻜﺮﺍﺭﻫﺎﻱ ﺑﺴﻴﺎﺭ ﺑﺴﻴﺎﺭ ﺯﻳﺎﺩﻱ ﻧﻴﺎﺯ ﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺩﺭ ﻋﻤﻞ‪ ،‬ﺍﺯ ﺍﻳﺪﻩﺁﻝ ﻓﻮﻕ ﺻﺮﻑﻧﻈﺮ ﻣﻲﻛﻨﻴﻢ ﻭ ﺩﺭ ﻋﻮﺽ ﺁﻥ ﺍﺯ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪۶۴‬‬

‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﻪ ﻧﻮﻋﻲ ﺍﻧﺘﻈﺎﺭ ﻣﺎ ﺭﺍ ﺍﺯ ﺩﻗﺖ ﺗﻘﺮﻳﺐ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ ﻭ ﻫﻤﻴﻨﻄﻮﺭ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ ﻛﻪ ﺗﺎ ﻛﺠﺎ ﺗﻜﺮﺍﺭﻫﺎ ﺑﺎﻳﺪ‬
‫ﺍﺩﺍﻣﻪ ﻳﺎﺑﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‪ ،‬ﺩﺭ ﭘﺎﻳﺎﻥ ﻫﺮ ﺗﻜﺮﺍﺭ‪ ،‬ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻛﻨﺘﺮﻝ ﻭ ﺑﺮﺭﺳﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺍﮔﺮ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﺮﻗﺮﺍﺭ ﺑﻮﺩ ﻛﻪ ﺗﻜﺮﺍﺭﻫﺎ‬
‫ﺭﺍ ﻣﺘﻮﻗﻒ ﻣﻲﻛﻨﻴﻢ ﻭ ﺗﻘﺮﻳﺐ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺩﺭ ﺁﺧﺮﻳﻦ ﺗﻜﺮﺍﺭ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺟﻮﺍﺏ ﺗﻘﺮﻳﺒﻲ ﻣﻌﺮﻓﻲ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺩﺭ ﻏﻴﺮ ﺍﻳﻦ ﺻﻮﺭﺕ‪،‬‬
‫ﺗﻜﺮﺍﺭﻫﺎ ﺭﺍ ﺍﺩﺍﻣﻪ ﻣﻲﺩﻫﻴﻢ‪.‬‬
‫ﺍﮔﺮ ∗‪ x‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﻭ ﻫﺮ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺩﻳﮕﺮ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ‪ ،Ax = b‬ﻳﻚ ﻣﻌﻴﺎﺭ‬
‫ﺗﻮﻗﻒ ﻣﻨﺎﺳﺐ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ ﻫﺮﮔﺎﻩ ‪. x(k) − x∗ < ε‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ ε‬ﻳﻚ ﻋﺪﺩ ﻛﻮﭼﻚ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺗﻠﺮﺍﻧﺲ ﺧﻄﺎ ‪ ۶۵‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﻃﺒﻴﻌﻲ ﺍﺳﺖ ﻛﻪ ﻫﺮ ﭼﻪ ﺗﻠﺮﺍﻧﺲ ﺧﻄﺎ ﺭﺍ ﻛﻮﭼﻚﺗﺮ‬
‫ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻘﺮﻳﺐ ﺑﻬﺘﺮﻱ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ∗‪ x‬ﻣﻌﻠﻮﻡ ﻧﻴﺴﺖ‪ ،‬ﻟﺬﺍ ﺩﺭ ﻋﻤﻞ ﻧﻤﻲﺗﻮﺍﻥ ﺍﺯ ﻣﻌﻴﺎﺭ‬
‫ﺗﻮﻗﻒ ﻓﻮﻕ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺩﺭ ﻋﻮﺽ ﺍﺯ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﺩﻳﮕﺮﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪:‬‬
‫‪ ۱‬ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ ﻫﺮ ﮔﺎﻩ ‪Ax(k) − b < ε‬‬
‫‪64 Stopping‬‬ ‫‪Criteria‬‬ ‫‪65 Error‬‬ ‫‪tolerance‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۱۹‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫‪ ۲‬ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ ﻫﺮ ﮔﺎﻩ ‪x(k) − x(k−1) < ε‬‬

‫∥ )‪∥x(k) −x(k−1‬‬
‫‪ ۳‬ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ ﻫﺮ ﮔﺎﻩ ‪< ε‬‬
‫∥ )‪∥x(k‬‬

‫ﺩﺭ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﻓﻮﻕ‪ ،‬ﻫﺮ ﻧﻮﻉ ﻧﺮﻣﻲ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﻟﺤﺎﻅ ﺳﺎﺩﮔﻲ ﺩﺭ ﻣﺤﺎﺳﺒﻪ‪ ،‬ﻧﺮﻡ ﺑﻲﻧﻬﺎﻳﺖ ﭘﻴﺸﻨﻬﺎﺩ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﺮﻗﺮﺍﺭﻱ ﻳﻚ ﺍﺯ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﻓﻮﻕ‪ ،‬ﺍﻳﺠﺎﺏ ﻧﻤﻲﻛﻨﺪ ﻛﻪ ‪ x(k) − x∗ < ε‬ﺍﻣﺎ ﺑﻪ ﻫﺮﺣﺎﻝ‪،‬‬
‫ﺑﺮﻗﺮﺍﺭﻱ ﻫﺮ ﻳﻚ ﺍﺯ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﻓﻮﻕ ﺑﻪ ﺍﺯﺍﻱ ‪ ε‬ﻛﻮﭼﻚ ﻣﺒﻴﻦ ﻧﺰﺩﻳﻚ ﺷﺪﻥ ﺑﻪ ﺟﻮﺍﺏ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ ﺍﺯ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻜﺮﺩﻳﻢ ﻭ ﺗﻨﻬﺎ ‪ ۲۹‬ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﺍﺩﻳﻢ‪ .‬ﺩﺭ ﻣﺜﺎﻝ ﺑﻌﺪ ﺑﺎ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻦ ﻣﻌﻴﺎﺭ‬
‫ﺗﻮﻗﻒ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺑﺎ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺣﻞ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ(‬ ‫ﻣﺜﺎﻝ ‪)۲۵−۴‬ﺭﻭﺵ ﮊﺍﮐﻮﺑﯽ ﺑﺮﺍﯼ ﺣﻞ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪10x1 − x2 + 2x3 = 6‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪ − x1 + 10x2 − x3 + 3x4 = 25‬‬
‫‪‬‬
‫‪‬‬ ‫‪2x1 − x2 + 5x3 − x4 = −11‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪3x − x + 10x = 15‬‬
‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬

‫)‪ x(k) − x(k−1‬ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ .‬ﻣﻌﺎﺩﻻﺕ‬ ‫‪2‬‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺎ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﻭ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪< ε = 1 e − 4‬‬
‫ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬
‫‪6 + x2 − 2x3‬‬
‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪ 1‬‬ ‫‪10‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪ (k+1‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪25‬‬ ‫‪+‬‬ ‫‪x‬‬ ‫‪1 + x3 − 3x4‬‬
‫‪ x2‬‬ ‫=‬
‫‪10‬‬
‫‪‬‬
‫‪‬‬ ‫‪−11‬‬ ‫‪−‬‬
‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(k+1‬‬ ‫‪2x‬‬ ‫‪1 + x2 + x4‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬ ‫‪3‬‬ ‫=‬
‫‪‬‬
‫‪‬‬ ‫‪5‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪15‬‬ ‫‪−‬‬ ‫‪3x‬‬
‫)‪(k‬‬
‫‪+‬‬
‫)‪(k‬‬
‫‪x3‬‬
‫)‪x(k+1‬‬
‫‪4‬‬ ‫=‬ ‫‪2‬‬
‫‪10‬‬
‫)‪ [x1 , x(0‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ‬
‫)‪(0‬‬
‫‪2‬‬ ‫‪,‬‬ ‫‪x‬‬
‫)‪(0‬‬
‫‪3‬‬ ‫‪,‬‬ ‫‪x‬‬
‫⊤ )‪(0‬‬
‫‪4‬‬ ‫]‬ ‫ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﻪ ﺍﺯﺍﻱ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ⊤]‪:= [0, 0, 0, 0‬‬
‫ﺍﻳﻦ ﺟﺪﻭﻝ ﻳﻚ ﺳﺘﻮﻥ ﺑﺮﺍﻱ ﻣﻘﺪﺍﺭ )‪ x(k) − x(k−1‬ﻧﻴﺰ ﺍﺿﺎﻓﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲۰‬‬

‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬

‫‪k‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫)‪x(k) − x(k−1‬‬ ‫‪2‬‬
‫‪0‬‬ ‫‪0.6000000‬‬ ‫‪2.5000000‬‬ ‫‪-2.2000000‬‬ ‫‪1.5000000‬‬ ‫–‬
‫‪1‬‬ ‫‪1.2900000‬‬ ‫‪1.8900000‬‬ ‫‪-1.6400000‬‬ ‫‪0.5300000‬‬ ‫‪1.45 e+00‬‬
‫‪2‬‬ ‫‪1.1170000‬‬ ‫‪2.3060000‬‬ ‫‪-2.2320000‬‬ ‫‪0.7690000‬‬ ‫‪7.81 e-01‬‬
‫‪3‬‬ ‫‪1.2770000‬‬ ‫‪2.1578000‬‬ ‫‪-2.0318000‬‬ ‫‪0.5850000‬‬ ‫‪3.49 e-01‬‬
‫‪4‬‬ ‫‪1.2221400‬‬ ‫‪2.2490200‬‬ ‫‪-2.1622400‬‬ ‫‪0.6494800‬‬ ‫‪1.80 e-01‬‬
‫‪5‬‬ ‫‪1.2573500‬‬ ‫‪2.2111460‬‬ ‫‪-2.1091560‬‬ ‫‪0.6090700‬‬ ‫‪8.44 e-02‬‬
‫‪6‬‬ ‫‪1.2429458‬‬ ‫‪2.2320984‬‬ ‫‪-2.1388968‬‬ ‫‪0.6257406‬‬ ‫‪4.25 e-02‬‬
‫‪7‬‬ ‫‪1.2509892‬‬ ‫‪2.2226827‬‬ ‫‪-2.1256105‬‬ ‫‪0.6164808‬‬ ‫‪2.04 e-02‬‬
‫‪8‬‬ ‫‪1.2473904‬‬ ‫‪2.2275936‬‬ ‫‪-2.1325630‬‬ ‫‪0.6206341‬‬ ‫‪1.01 e-02‬‬
‫‪9‬‬ ‫‪1.2492720‬‬ ‫‪2.2252925‬‬ ‫‪-2.1293106‬‬ ‫‪0.6184656‬‬ ‫‪4.91 e-03‬‬
‫‪10‬‬ ‫‪1.2483914‬‬ ‫‪2.2264565‬‬ ‫‪-2.1309572‬‬ ‫‪0.6194812‬‬ ‫‪2.42 e-03‬‬
‫‪11‬‬ ‫‪1.2488371‬‬ ‫‪2.2258991‬‬ ‫‪-2.1301690‬‬ ‫‪0.6189673‬‬ ‫‪1.18 e-03‬‬
‫‪12‬‬ ‫‪1.2486237‬‬ ‫‪2.2261766‬‬ ‫‪-2.1305615‬‬ ‫‪0.6192134‬‬ ‫‪5.81 e-04‬‬
‫‪13‬‬ ‫‪1.2487300‬‬ ‫‪2.2260422‬‬ ‫‪-2.1303715‬‬ ‫‪0.6190909‬‬ ‫‪2.84 e-04‬‬
‫‪14‬‬ ‫‪1.2486785‬‬ ‫‪2.2261086‬‬ ‫‪-2.1304654‬‬ ‫‪0.6191502‬‬ ‫‪1.39 e-04‬‬
‫‪15‬‬ ‫‪1.2487039‬‬ ‫‪2.2260763‬‬ ‫‪-2.1304197‬‬ ‫‪0.6191209‬‬ ‫‪6.81 e-05‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ‪ ،‬ﺑﻌﺪ ﺍﺯ ‪ ۱۵‬ﺗﻜﺮﺍﺭ ﺑﻪ ﺗﻘﺮﻳﺐ ﺯﻳﺮ ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺭﺳﻴﺪﻳﻢ‪:‬‬
‫⊤]‪x(15) = [1.2487039, 2.2260763, −2.1304197, 0.6191209‬‬
‫ﺟﻮﺍﺏ ﺩﻗﻴﻖ ﺩﺳﺘﮕﺎﻩ )ﺗﺎ ‪ ۱۵‬ﺭﻗﻢ ﺑﺎ ﻣﻌﻨﻲ( ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‪:‬‬
‫⊤]‪x := [1.24869565217391, 2.22608695652174, −2.13043478260869, 0.61913043478261‬‬
‫∗‬

‫ﻟﺬﺍ ﻧﺮﻡ ﺩﻭﻡ ﺧﻄﺎﻱ ﺗﻘﺮﻳﺐ )‪ x(15‬ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ‬

‫∗‪x(15) − x‬‬ ‫‪= 2.24 e − 05‬‬
‫‪2‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪ x(k) − x(k−1) 2 < ε = 1 e − 4‬ﻣﺎ ﺭﺍ ﺑﻪ ﺟﻮﺍﺑﻲ ﺑﺎ‬
‫ﻧﺮﻡ ﺧﻄﺎﻱ ‪ 2.24 e − 05‬ﺭﺳﺎﻧﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺍﻳﻦ ﻧﻜﺘﻪ ﻳﻚ ﻗﺎﻧﻮﻥ ﻛﻠﻲ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ﻫﻤﻮﺍﺭﻩ ﻧﻤﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ﻣﻌﻴﺎﺭ‬
‫ﺗﻮﻗﻒ ‪ x(k) − x(k−1) 2 < ε‬ﻭ ﻳﺎ ‪ Ax(k) − b < ε‬ﻣﺎ ﺭﺍ ﺑﻪ ﺟﻮﺍﺑﻲ ﺑﺎ ﺩﻗﺖ ‪ x(k) − x∗ < ε‬ﻣﻲﺭﺳﺎﻧﺪ‪.‬‬

‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ‬ ‫�‬

‫ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺟﻬﺖ ﺳﺮﻋﺖ ﺑﺨﺸﻴﺪﻥ ﺑﻪ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺍﺳﺖ ﻛﻪ ﺑﺎ ﻳﻚ ﺗﻐﻴﻴﺮ ﺟﺰﻳﻲ ﺩﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺣﺎﺻﻞ ﻣﻲ ﺷﻮﺩ‪.‬‬
‫ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺳﻌﻲ ﻣﻲﺷﻮﺩ ﺍﺯ ﺁﺧﺮﻳﻦ ﻣﻘﺎﺩﻳﺮ ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ )ﻳﺎ ﺟﺪﻳﺪﺗﺮﻳﻦ ﻣﻘﺪﺍﺭ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺑﺮﺍﻱ ﻣﺘﻐﻴﺮﻫﺎ( ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﮔﺮﺩﺩ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺩﻳﺪﻳﻢ‪ ،‬ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ x1 , . . . , xn‬ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ‪x1 , . . . , xn‬‬
‫∗‬ ‫∗‬ ‫)‪(k‬‬ ‫)‪(k‬‬

‫‪x1‬‬
‫)‪(k−1‬‬ ‫)‪(k−1‬‬
‫‪, . . . , xn‬‬ ‫)‪ x(k‬ﺍﺯ ﺗﻘﺮﻳﺐ ﺣﺎﺻﻞ ﺩﺭ ﺗﻜﺮﺍﺭ )‪(k − 1‬ﺍﻡ ﻳﻌﻨﻲ‬ ‫ﺣﺎﺻﻞ ﮔﺮﺩﻳﺪ ﻭ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ‪1 , . . . , xn‬‬
‫)‪(k‬‬

‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﺻﻮﻻ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺗﻘﺮﻳﺐ ﺣﺎﺻﻞ ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﺩﻗﻴﻖﺗﺮ ﺍﺯ ﺗﻘﺮﻳﺐ ﺣﺎﺻﻞ ﺍﺯ ﺗﻜﺮﺍﺭ )‪(k − 1‬ﺍﻡ ﺑﺎﺷﺪ‪ .‬ﺑﻪ‬
‫)‪ x(k−1‬ﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﺗﺎ ﺁﻧﺠﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺯ ﺗﻘﺮﻳﺐﻫﺎﻱ‬ ‫‪i‬‬ ‫)‪ x(k‬ﺩﻗﻴﻖﺗﺮ ﺍﺯ‬
‫‪i‬‬ ‫ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ‬
‫ﺟﺪﻳﺪ ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻧﻜﺘﻪ ﺍﻳﺪﻩ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺍﺳﺖ‪.‬‬
‫)‪ x(k‬ﺑﻪ ﺩﺳﺖ‬ ‫‪ x1‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻣﺘﻐﻴﺮﻫﺎﻱ ‪2 , . . . , xn‬‬
‫)‪(k‬‬ ‫)‪(k+1‬‬
‫ﺩﺭ ﺗﻜﺮﺍﺭ )‪(k + 1‬ﺍﻡ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺩﺭ ﺍﺑﺘﺪﺍ ﻣﻘﺪﺍﺭ‬
‫‪ x2‬ﺑﺮﺣﺴﺐ ‪ x1 , x3 , . . . , xn‬ﺑﻪ ﺩﺳﺖ‬
‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k+1‬‬
‫)‪ x(k+1‬ﻣﻲﺭﻭﻳﻢ‪ .‬ﺩﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ‪،‬‬ ‫‪2‬‬ ‫ﻣﻲﺁﻳﺪ‪ .‬ﺣﺎﻝ ﺑﻪ ﺳﺮﺍﻍ ﻣﺤﺎﺳﺒﻪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۲۱‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫)‪ x(k+1‬ﺍﺳﺘﻔﺎﺩﻩ‬
‫‪1‬‬ ‫)‪ x(k‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ ﺍﺯ ﺗﻘﺮﻳﺐ ﺑﻬﺘﺮ‬
‫‪ x1‬ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﻭ ﺑﻬﺘﺮ ﺍﺳﺖ ﺑﻪ ﺟﺎﻱ ﺍﻳﻦ ﻛﻪ ﺍﺯ ‪1‬‬
‫)‪(k+1‬‬
‫ﻣﻲﺁﻳﺪ‪ .‬ﺍﻣﺎ ﺗﺎ ﺍﻳﻨﺠﺎ‬
‫‪ x2‬ﺭﺍ ﺑﺮﺣﺴﺐ ‪ x1 , x3 , . . . , xn‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﻫﻤﻴﻨﻄﻮﺭ ﻣﻨﺎﺳﺐﺗﺮ ﺍﺳﺖ‬
‫)‪(k+1‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k+1‬‬
‫ﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺑﻬﺘﺮ ﺍﺳﺖ‬
‫)‪ x1 , x2 , x4 , . . . , x(k‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ ﻭ ﻫﻤﻴﻦﻃﻮﺭ ﻣﺘﻐﻴﺮﻫﺎﻱ ﺩﻳﮕﺮ ﺭﺍ ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ‬
‫)‪(k+1‬‬ ‫)‪(k+1‬‬ ‫)‪(k‬‬
‫‪ x3‬ﺑﺮﺣﺴﺐ ‪n‬‬
‫)‪(k+1‬‬
‫ﻛﻪ‬
‫ﺍﺳﺖ ﺑﺮ ﺣﺴﺐ ﻣﺘﻐﻴﺮﻫﺎﻱ ﺟﺪﻳﺪﺗﺮ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪.‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻮﺿﻴﺤﺎﺕ ﻗﺒﻞ‪ ،‬ﺩﺭ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ‪ ،۲۰.۴‬ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ‬
‫ﻣﻲﺷﻮﺩ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪b1 − a12 x2 − a13 x3 − · · · − a1n xn‬‬
‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬
‫‪a11‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k+1‬‬
‫‪b2 − a21 x1‬‬
‫)‪(k‬‬
‫‪− a23 x3 − · · · − a2n xn‬‬
‫)‪(k‬‬
‫)‪x(k+1‬‬
‫‪2‬‬ ‫=‬
‫‪a22‬‬ ‫)‪(۲۳.۴‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪ .‬‬
‫‪‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(k+1‬‬ ‫)‪(k+1‬‬ ‫)‪(k+1‬‬
‫)‪ (k+1‬‬
‫‪‬‬ ‫‪bn − an1 x1‬‬ ‫‪− an2 x2‬‬ ‫‪− · · · − an,n-1 xn-1‬‬
‫‪ xn‬‬ ‫=‬
‫‪ann‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭ ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ‪ ،‬ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺯ ﻣﺘﻐﻴﺮﻫﺎﻱ ﺟﺪﻳﺪﺗﺮ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ(‬ ‫ﻣﺜﺎﻝ ‪)۲۶−۴‬ﺭﻭﺵ ﮊﺍﮐﻮﺑﯽ ﺑﺮﺍﯼ ﺣﻞ ﯾ‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﻫﻤﺎﻥ ﺩﺳﺘﮕﺎﻩ ﻭ ﻫﻤﺎﻥ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۲۵-۴‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺍﻳﻦﺑﺎﺭ ﺭﻭﺵ‬
‫ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺭﺍ ﺑﺮﺍﻱ ﺣﻞ ﺁﻥ ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪ .‬ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪(k‬‬
‫‪6 + x2 − 2x3‬‬
‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬
‫‪10‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫)‪ (k+1‬‬ ‫)‪(k+1‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪25‬‬ ‫‪+‬‬ ‫‪x‬‬ ‫‪+ x3 − 3x4‬‬
‫‪ x2‬‬ ‫=‬ ‫‪1‬‬
‫‪10‬‬
‫‪‬‬
‫‪‬‬ ‫‪−11‬‬ ‫‪−‬‬
‫)‪(k+1‬‬ ‫)‪(k+1‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫)‪(k+1‬‬ ‫‪2x‬‬ ‫‪1‬‬ ‫‪+ x2‬‬ ‫‪+ x4‬‬
‫‪‬‬
‫‪‬‬ ‫‪x‬‬ ‫‪3‬‬ ‫=‬
‫‪‬‬
‫‪‬‬ ‫‪5‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪15‬‬ ‫‪−‬‬ ‫‪3x‬‬
‫)‪(k+1‬‬
‫‪+‬‬ ‫‪x3‬‬
‫)‪(k+1‬‬
‫)‪x(k+1‬‬
‫‪4‬‬ ‫=‬ ‫‪2‬‬
‫‪10‬‬
‫ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﻪ ﺍﺯﺍﻱ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ⊤]‪ [0, 0, 0, 0‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪k‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫)‪x(k) − x(k−1‬‬ ‫‪2‬‬
‫‪0‬‬ ‫‪0.6000000‬‬ ‫‪2.5000000‬‬ ‫‪-2.2000000‬‬ ‫‪1.5000000‬‬
‫‪1‬‬ ‫‪1.2900000‬‬ ‫‪1.9590000‬‬ ‫‪-2.0242000‬‬ ‫‪0.7098800‬‬ ‫‪1.19e+00‬‬
‫‪2‬‬ ‫‪1.2007400‬‬ ‫‪2.2046900‬‬ ‫‪-2.0973820‬‬ ‫‪0.6288548‬‬ ‫‪2.83e-01‬‬
‫‪3‬‬ ‫‪1.2399454‬‬ ‫‪2.2255999‬‬ ‫‪-2.1250872‬‬ ‫‪0.6198113‬‬ ‫‪5.31e-02‬‬
‫‪4‬‬ ‫‪1.2475774‬‬ ‫‪2.2263056‬‬ ‫‪-2.1298076‬‬ ‫‪0.6191276‬‬ ‫‪9.03e-03‬‬
‫‪5‬‬ ‫‪1.2485921‬‬ ‫‪2.2261402‬‬ ‫‪-2.1303833‬‬ ‫‪0.6191196‬‬ ‫‪1.18e-03‬‬
‫‪6‬‬ ‫‪1.2486907‬‬ ‫‪2.2260949‬‬ ‫‪-2.1304334‬‬ ‫‪0.6191282‬‬ ‫‪1.20e-04‬‬
‫‪7‬‬ ‫‪1.2486962‬‬ ‫‪2.2260878‬‬ ‫‪-2.1304353‬‬ ‫‪0.6191301‬‬ ‫‪9.32e-06‬‬

‫ﺑﻌﺪ ﺍﺯ ‪ ۷‬ﺗﻜﺮﺍﺭ ﺑﻪ ﺗﻘﺮﻳﺐ ⊤]‪ x(7) = [1.2486962, 2.2260878, −2.1304353, 0.6191301‬ﺑﺮﺍﻱ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲۲‬‬

‫∗‪ . x(15) − x‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﺎ ‪ ۸‬ﺗﻜﺮﺍﺭ ﻛﻤﺘﺮ ﺑﻪ‬ ‫‪2‬‬
‫ﺭﺳﻴﺪﻳﻢ ﻛﻪ ‪= 9.70 e − 06‬‬
‫ﺟﻮﺍﺑﻲ ﺑﻪ ﻫﻤﺎﻥ ﺩﻗﺖ ﺭﺳﻴﺪ‪.‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ‬ ‫�‬

‫ﺗﺎ ﺍﻳﻨﺠﺎ ﺭﻭﺵﻫﺎﻱ ﮊﺍﻛﻮﺑﻲ ﻭ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﺮﺍﻱ ﺣﻞ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﻄﻲ ﺍﺭﺍﻳﻪ ﺷﺪ‪ .‬ﺑﻪ ﻧﻈﺮ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺑﻪ ﺧﻮﺑﻲ ﻭ ﺳﺎﺩﮔﻲ‬
‫ﻗﺎﺩﺭﻧﺪ ﻛﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻛﻨﻨﺪ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺍﺭﺍﻱ ﻳﻚ ﻣﺸﻜﻞ ﺑﺰﺭﮒ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻋﺪﻡ‬
‫ﻫﻤﮕﺮﺍﻳﻲ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺭﻭﻱ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺍﻋﻤﺎﻝ ﺷﻮﻧﺪ ﻭ ﺩﻧﺒﺎﻟﻪﺍﻱ ﻭﺍﮔﺮﺍ ﺗﻮﻟﻴﺪ ﻛﻨﻨﺪ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺍﮔﺮ ﺩﻧﺒﺎﻟﻪ ﺑﺮﺩﺍﺭﻱ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﻪ ∗‪ x‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ∗‪ x‬ﻳﻚ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ‬
‫ﺍﺳﺖ‪ .‬ﺗﺒﻌﺎ ﺍﮔﺮ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﻫﺮ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﻫﻴﭻﮔﺎﻩ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﻣﻤﻜﻦ ﺍﺳﺖ‬
‫ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺟﻮﺍﺏ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ ﻭﻟﻲ ﺑﺎﺯ ﻫﻢ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﻫﻤﮕﺮﺍ ﻧﮕﺮﺩﺩ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ(‬ ‫ﻣﺜﺎﻝ ‪)۲۷−۴‬ﻋﺪﻡ ﻫﻤ ﺮﺍﯾﯽ ﺭﻭﺵ ﮊﺍﮐﻮﺑﯽ ﺑﺮﺍﯼ ﯾ‬

‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x + 9x2 − 2x3 = 13‬‬
‫‪ 1‬‬
‫‪− 5x1 + x2 + 2x3 = 2‬‬ ‫)‪(۲۴.۴‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x1 + 3x2 + 7x3 = 29‬‬
‫ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﻧﺎﻣﻨﻔﺮﺩ ﺍﺳﺖ ﻭ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ⊤]‪ x := [1, 2, 3‬ﺍﺳﺖ‪ .‬ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﺎ ﻧﻘﻄﻪ ﺷﺮﻭﻉ ]‪[0, 0, 0‬‬
‫⊤‬

‫ﺭﺍ ﺭﻭﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺗﻜﺮﺍﺭﻫﺎﻱ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫)‪(k‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪k‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬
‫‪0‬‬ ‫‪0.0000000 e+00‬‬ ‫‪0.0000000 e+00‬‬ ‫‪0.0000000 e+00‬‬
‫‪1‬‬ ‫‪1.3000000 e+01‬‬ ‫‪3.0000000 e+00‬‬ ‫‪4.1428571 e+00‬‬ ‫ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ‪ MATLAB‬ﻭ ﺑﺎ‬
‫‪2‬‬ ‫‪-5.7142857 e+00‬‬ ‫‪5.9714286 e+01‬‬ ‫‪-8.5714286 e-01‬‬
‫‪3‬‬ ‫‪-5.2614286 e+02‬‬ ‫‪-2.3857143 e+01‬‬ ‫‪-1.9816327 e+01‬‬ ‫ﻓﺮﻣﺖ ﺩﻗﺖ ﻣﻀﺎﻋﻒ ﺍﻧﺠﺎﻡ ﺷﺪﻩ‬
‫‪4‬‬ ‫‪1.8808163 e+02‬‬ ‫‪-2.5880816 e+03‬‬ ‫‪1.6469388 e+02‬‬ ‫ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ‬
‫‪5‬‬ ‫‪2.3635122 e+04‬‬ ‫‪6.1402041 e+02‬‬ ‫‪1.0595831 e+03‬‬
‫ﻣﻲﺷﻮﺩ‪ ،‬ﺑﺎ ﺍﻧﺠﺎﻡ ﺗﻜﺮﺍﺭﻫﺎ‪ ،‬ﺍﻧﺪﺍﺯﻩ‬
‫‪6‬‬ ‫‪-3.3940175 e+03‬‬ ‫‪1.1605945 e+05‬‬ ‫‪-7.0119009e+03‬‬
‫‪7‬‬ ‫‪-1.0585458 e+06‬‬ ‫‪-2.9432857 e+03‬‬ ‫‪-4.8765900 e+04‬‬ ‫)‪ x(k‬ﺍﻓﺰﺍﻳﺶ ﻣﻲﻳﺎﺑﺪ ﺗﺎ ﺍﻳﻦ ﻛﻪ‬ ‫‪i‬‬
‫‪8‬‬ ‫‪-7.1029229 e+04‬‬ ‫‪-5.1951943 e+06‬‬ ‫‪3.0370721 e+05‬‬ ‫ﺩﺭ ﺗﻜﺮﺍﺭ ‪ ۳۷۳‬ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ‬
‫‪9‬‬ ‫‪4.7364176 e+07‬‬ ‫‪-9.6255757 e+05‬‬ ‫‪2.2468100 e+06‬‬
‫‪..‬‬
‫‪.‬‬
‫‪..‬‬
‫‪.‬‬
‫‪..‬‬
‫‪.‬‬
‫‪..‬‬
‫‪.‬‬
‫ﻣﻲﺍﻓﺘﺪ ﻭ ‪ Inf‬ﺟﺎﻱ ﻣﻘﺎﺩﻳﺮ ﺭﺍ‬
‫‪372‬‬ ‫‪1.053230 e+307‬‬ ‫‪-1.283819 e+307‬‬ ‫‪1.245590 e+306‬‬ ‫ﻣﻲﮔﻴﺮﻳﺪ‪ .‬ﺳﭙﺲ ﻋﻤﻠﻴﺎﺕ ﺭﻭﻱ‬
‫‪373‬‬ ‫‪1.180349 e+308‬‬ ‫‪5.017033 e+307‬‬ ‫‪2.492851 e+306‬‬ ‫‪Inf‬ﻫﺎ ﺑﻪ ‪ NaN‬ﻣﻨﺠﺮ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫‪374‬‬ ‫‪-Inf‬‬ ‫‪Inf‬‬ ‫‪-Inf‬‬
‫‪375‬‬ ‫‪-Inf‬‬ ‫‪NaN‬‬ ‫‪NaN‬‬
‫‪376‬‬ ‫‪NaN‬‬ ‫‪NaN‬‬ ‫‪NaN‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻫﺮﭼﻨﺪ ﺩﺳﺘﮕﺎﻩ ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ⊤]‪ [0, 0, 0‬ﻗﺎﺩﺭ ﺑﻪ ﺣﻞ ﺍﻳﻦ‬
‫ﺩﺳﺘﮕﺎﻩ ﻧﮕﺮﺩﻳﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۲۳‬‬ ‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﯽ‬ ‫ﻓﺼﻞ ﭼﻬﺎﺭﻡ‪:‬‬

‫■ ﺗﻀﻤﻴﻦ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﻱ ﺑﺮﺧﻲ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﺎﺹ‬

‫ﺩﻳﺪﻳﻢ ﻛﻪ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﻭ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﻫﻤﻮﺍﺭﻩ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺮﺧﻲ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺧﺎﺹ‪ ،‬ﻫﻤﮕﺮﺍﻳﻲ ﺗﻀﻤﻴﻦ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﻫﻤ ﺮﺍﯾﯽ ﺭﻭﺵﻫﺎﯼ ﺗﮑﺮﺍﺭﯼ ﺑﺮﺍﯼ ﺩﺳﺘﮕﺎەﻫﺎﯼ ﻏﺎﻟﺐ ﻗﻄﺮﯼ ﺍﮐﯿﺪ ﺳﻄﺮﯼ ﻭ ﺩﺳﺘﮕﺎەﻫﺎﯼ ﻣﻌﯿﻦ ﻣﺜﺒﺖ‬ ‫ﻗﻀﯿﻪ ‪۱۳−۴‬‬
‫ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫• ﺍﮔﺮ ‪ A‬ﻣﺎﺗﺮﻳﺴﻲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺍﻛﻴﺪ ﺳﻄﺮﻱ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵﻫﺎﻱ ﮊﺍﻛﻮﺑﻲ ﻭ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‬
‫ﺩﻟﺨﻮﺍﻩ ﺑﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫• ﺍﮔﺮ ‪ A‬ﻣﺎﺗﺮﻳﺲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ ﺑﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ‬
‫ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫• ﻓﺮﺽ ﻛﻨﻴﺪ ‪ D‬ﻣﺎﺗﺮﻳﺲ ﻗﻄﺮﻱ ﺑﺎﺷﺪ ﻛﻪ ﺍﻋﻀﺎﻱ ﺭﻭﻱ ﻗﻄﺮ ﺁﻥ ﺑﺮﺍﺑﺮ ﺍﻋﻀﺎﻱ ﻗﻄﺮﻱ ‪ A‬ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ‪ A‬ﻣﺎﺗﺮﻳﺴﻲ ﻣﻌﻴﻦ‬
‫ﻣﺜﺒﺖ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ‪ 2D − A‬ﻧﻴﺰ ﻣﺎﺗﺮﻳﺴﻲ ﻣﻌﻴﻦ ﻣﺜﺒﺖ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﺎ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ‬
‫ﺑﻪ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺍﮔﺮ ﺩﺳﺘﮕﺎﻫﻲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻤﻜﻦ ﺍﺳﺖ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺭﻭﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻫﻤﮕﺮﺍ ﻳﺎ ﻭﺍﮔﺮﺍ ﺷﻮﻧﺪ‪.‬‬
‫ﻭﻟﻲ ﺍﮔﺮ ﺩﺳﺘﮕﺎﻫﻲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﻧﺒﺎﺷﺪ ﻭ ﺑﺘﻮﺍﻥ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﻣﻌﺎﺩﻻﺕ )ﻭ ﻳﺎ ﺩﻳﮕﺮ ﻋﻤﻠﻴﻠﺖ ﺳﻄﺮﻱ ﻣﻘﺪﻣﺎﺗﻲ( ﺁﻥ ﺭﺍ ﺑﻪ‬
‫ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﺗﺒﺪﻳﻞ ﻛﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺮﺍﻱ ﺗﻀﻤﻴﻦ ﻫﻤﮕﺮﺍﻳﻲ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺟﺎﺑﺠﺎﻳﻲ ﺳﻄﺮﻫﺎ ﺍﻧﺠﺎﻡ ﺷﻮﺩ‪ .‬ﻣﺜﺎﻝ‬
‫ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻣﻨﻈﻮﺭ ﺗﺒﺪﯾﻞ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻏﺎﻟﺐ ﻗﻄﺮﯼ ﺳﻄﺮﯼ(‬ ‫ﻣﺜﺎﻝ ‪)۲۸−۴‬ﺟﺎﺑﺠﺎﯾﯽ ﻣﻌﺎﺩﻻﺕ ﯾ‬
‫ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺩﺭ ﻣﺜﺎﻝ ‪ ۲۷-۴‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪x + 9x2 − 2x3 = 13‬‬
‫‪ 1‬‬
‫‪− 5x1 + x2 + 2x3 = 2‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x1 + 3x2 + 7x3 = 29‬‬
‫ﺩﻳﺪﻳﻢ ﻛﻪ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﺮﺍﻱ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻭﺍﮔﺮﺍ ﺷﺪ‪ .‬ﺍﻛﻨﻮﻥ ﺑﺎ ﺟﺎﺑﺠﺎﻳﻲ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ﻭ ﺩﻭﻡ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻝ ﺯﻳﺮ ﻣﻲﺭﺳﻴﻢ‪:‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪− 5x1 + x2 + 2x3 = 2‬‬
‫‪‬‬
‫‪x1 + 9x2 − 2x3 = 13‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪2x1 + 3x2 + 7x3 = 29‬‬
‫ﻛﻪ ﺩﺳﺘﮕﺎﻫﻲ ﻏﺎﻟﺐ ﻗﻄﺮﻱ ﺳﻄﺮﻱ ﺍﺳﺖ ﻭ ﻃﺒﻖ ﻗﻀﻴﻪ ‪ ۱۳-۴‬ﺭﻭﺵﻫﺎﻱ ﮊﺍﻛﻮﺑﻲ ﻭ ﮔﻮﺱ‪-‬ﺳﻴﺪﻝ ﺑﺎ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ ﺑﻪ‬
‫ﺟﻮﺍﺏ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲۴‬‬

‫‪k‬‬
‫)‪(k‬‬
‫‪x1‬‬ ‫‪x2‬‬
‫)‪(k‬‬
‫‪x3‬‬
‫)‪(k‬‬
‫ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪0‬‬ ‫‪0.0000000‬‬ ‫‪0.0000000‬‬ ‫‪0.0000000‬‬ ‫‪‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫‪‬‬ ‫‪2 − x2 − 2x3‬‬
‫‪1‬‬ ‫‪-0.6000000‬‬ ‫‪1.4444444‬‬ ‫‪4.1428571‬‬ ‫‪‬‬
‫‪‬‬ ‫‪x1‬‬
‫)‪(k+1‬‬
‫=‬
‫‪‬‬
‫‪‬‬ ‫‪−5‬‬
‫‪2‬‬ ‫‪1.3460317‬‬ ‫‪2.4317460‬‬ ‫‪3.6952381‬‬ ‫‪‬‬
‫‪‬‬ ‫)‪(k‬‬ ‫)‪(k‬‬
‫‪3‬‬ ‫‪1.3644444‬‬ ‫‪2.1160494‬‬ ‫‪2.7160998‬‬ ‫)‪(k+1‬‬ ‫‪13 − x1 + 2x3‬‬
‫‪‬‬ ‫‪x‬‬ ‫=‬
‫‪4‬‬ ‫‪0.9096498‬‬ ‫‪1.8964172‬‬ ‫‪2.8461376‬‬ ‫‪‬‬
‫‪‬‬
‫‪2‬‬
‫‪9‬‬
‫‪‬‬
‫‪‬‬
‫‪5‬‬ ‫‪0.9177385‬‬ ‫‪1.9758473‬‬ ‫‪3.0702070‬‬ ‫‪‬‬ ‫‪−‬‬
‫)‪(k‬‬ ‫)‪(k‬‬
‫‪‬‬
‫= )‪x(k+1‬‬ ‫‪29‬‬ ‫‪2x‬‬ ‫‪1 − 3x2‬‬
‫‪6‬‬ ‫‪1.0232522‬‬ ‫‪2.0247417‬‬ ‫‪3.0338545‬‬ ‫‪3‬‬
‫‪7‬‬
‫‪7‬‬ ‫‪1.0184901‬‬ ‫‪2.0049396‬‬ ‫ﺣﺎﻝ ﺑﺎ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺭﻭﺵ ﮊﺍﻛﻮﺑﻲ ﻭ ﺍﻧﺘﺨﺎﺏ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪2.9827529‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪25‬‬ ‫‪1.0000000‬‬ ‫‪2.0000000‬‬ ‫‪3.0000000‬‬ ‫⊤]‪ ،[0, 0, 0‬ﻧﺘﺎﻳﺞ ﺩﺭ ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ ﮔﺰﺍﺭﺵ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‬ ‫�‬

‫ﺩﺭ ﻧﮕﺎﻩ ﺍﻭﻝ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺩﺭ ﻣﻘﺎﻳﺴﻪ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﺣﺬﻓﻲ ﮔﻮﺱ ﭼﻨﺪﺍﻥ ﻣﻔﻴﺪ ﺑﻪ ﻧﻈﺮ ﻧﻤﻲﺭﺳﻨﺪ ﻭ ﻫﻴﭻ ﺑﺮﺗﺮﻱ ﻧﺴﺒﺖ ﺑﻪ‬
‫ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﻧﺪﺍﺭﻧﺪ‪ .‬ﻣﺨﺼﻮﺻﺎ ﺍﻳﻦ ﻛﻪ ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻣﺸﻜﻞ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﻧﻴﺰ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻭ ﻫﻤﭽﻨﻴﻦ ﺗﻌﺪﺍﺩ‬
‫ﺗﻜﺮﺍﺭﻫﺎﻱ ﻻﺯﻡ ﺍﺯ ﻗﺒﻞ ﻣﺸﺨﺺ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺑﺎ ﺍﻳﻦ ﺣﺎﻝ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺟﺎﻳﮕﺎﻩ ﺧﺎﺹ ﺧﻮﺩ ﺭﺍ ﺩﺍﺭﻧﺪ ﻭ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ‬
‫ﻣﺴﺎﻳﻞ ﻛﺎﺭﺑﺮﺩﻱ ﺑﺎ ﻣﻮﻓﻘﻴﺖ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﺮﺧﻲ ﺍﺯ ﻣﺰﺍﻳﺎﻱ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻧﺴﺒﺖ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‬
‫ﺑﺮﺷﻤﺮﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫• ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺩﺭ ﺣﻞ ﺑﺮﺧﻲ ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﺑﺰﺭﮒ )ﺑﺎ ﺩﻩﻫﺎ ﻫﺰﺍﺭ ﻣﺘﻐﻴﺮ( ﻧﺴﺒﺖ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﺩﺍﺭﺍﻱ ﻛﺎﺭﺍﻳﻲ‬
‫ﻭ ﺳﺮﻋﺖ ﺑﻴﺸﺘﺮﻱ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ‪ ،‬ﺍﮔﺮ ﺍﺯ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻣﻄﻤﻴﻦ ﺑﺎﺷﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ‬
‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺗﻮﺻﻴﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫• ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺗﻨﻚ ﺑﺎﺷﺪ‪ ،‬ﻳﻌﻨﻲ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺳﺘﮕﺎﻩ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺗﻨﻚ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺭ ﻫﺮ ﺗﻜﺮﺍﺭ ﺍﺯ ﺭﻭﺵﻫﺎﻱ‬
‫ﺗﻜﺮﺍﺭﻱ ﺑﻪ ﻋﻤﻠﻴﺎﺕ ﻛﻤﺘﺮﻱ ﻧﻴﺎﺯ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺑﺰﺭﮒ ﻭ ﺗﻨﻚ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ ﻟﺤﺎﻅ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ‪،‬‬
‫ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻧﺴﺒﺖ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ ﺗﺮﺟﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﻧﺪ )ﺍﻟﺒﺘﻪ ﻣﺸﺮﻭﻁ ﺑﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ(‪.‬‬

‫• ﺩﺭ ﺑﺮﺧﻲ ﻛﺎﺭﺑﺮﺩﻫﺎ‪ ،‬ﻣﺎﻧﻨﺪ ﺣﻞ ﻋﺪﺩﻱ ﻣﻌﺎﺩﻻﺕ ﺩﻳﻔﺮﺍﻧﺴﻴﻞ ﭘﺎﺭﻩﺍﻱ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﺣﻞ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺍﺯ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺩﺍﺭﻳﻢ ﻛﻪ‬
‫ﺩﺳﺘﮕﺎﻩﻫﺎﻱ ﻣﺘﻮﺍﻟﻲ ﺗﻔﺎﻭﺕﻫﺎﻱ ﻛﻮﭼﻜﻲ ﺑﺎ ﻫﻢ ﺩﺍﺭﻧﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺗﻮﺻﻴﻪ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﻪ‬
‫ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺑﻌﺪ ﺍﺯ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺍﻭﻝ‪ ،‬ﺟﻮﺍﺏ ﺁﻥ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﺭ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﺩﻭﻡ‬
‫ﺍﻧﺘﺨﺎﺏ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺍﻭﻝ ﻭ ﺩﻭﻡ ﺗﻔﺎﻭﺕ ﻛﻤﻲ ﺑﺎ ﻫﻢ ﺩﺍﺭﻧﺪ ﻟﺬﺍ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺟﻮﺍﺏﻫﺎﻱ‬
‫ﺁﻥﻫﺎ ﻧﻴﺰ ﺑﻪ ﻫﻢ ﻧﺰﺩﻳﻚ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻋﻠﺖ ﺟﻮﺍﺏ ﺩﺳﺘﮕﺎﻩ ﺍﻭﻝ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺧﻮﺏ ﺑﺮﺍﻱ ﺩﺳﺘﮕﺎﻩ ﺩﻭﻡ ﺍﺳﺖ ﻭ ﻟﺬﺍ‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺑﺎ ﺷﺮﻭﻉ ﺍﺯ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺧﻮﺏ ﻣﻲﺗﻮﺍﻧﺪ ﺑﺎ ﺗﻌﺪﺍﺩ ﺗﻜﺮﺍﺭ ﻛﻤﻲ ﺩﺳﺘﮕﺎﻩ ﺩﻭﻡ ﺭﺍ ﺣﻞ ﻛﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‬
‫ﺑﺎ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻛﻤﻲ ﻗﺎﺩﺭ ﺑﻪ ﺣﻞ ﺩﻧﺒﺎﻟﻪ ﺩﺳﺘﮕﺎﻩﻫﺎ ﺧﻮﺍﻫﻴﻢ ﺑﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۵‬‬
‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ‬
‫ﺑﺮﺍﯼ‬
‫ﻓﺼﻞ ﭘﻨﺠﻢ‬

‫ﺭﯾﺸەﯾﺎﺑﯽ‬
‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬
‫‪۱۲۶‬‬ ‫ﺭﻳﺸﻪ ﻭ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵‬‬
‫‪۱۳۱‬‬ ‫ﻣﺎﻫﻴﺖ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪. . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۵‬‬
‫‪۱۳۳‬‬ ‫ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۵‬‬
‫‪۱۳۶‬‬ ‫ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۵‬‬
‫‪۱۳۹‬‬ ‫ﺭﻭﺵﻫﺎﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۵‬‬
‫‪۱۴۱‬‬ ‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۵‬‬
‫‪۱۴۴‬‬ ‫ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۴.۵‬‬
‫‪۱۴۹‬‬ ‫ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۴.۵‬‬
‫‪۱۵۱‬‬ ‫ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۵‬‬
‫‪۱۵۳‬‬ ‫ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۵‬‬
‫‪۱۵۴‬‬ ‫ﺭﻭﺵ ﻭﺗﺮﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۶.۵‬‬
‫‪۱۵۶‬‬ ‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۷.۵‬‬
‫‪۱۵۹‬‬ ‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۷.۵‬‬

‫‪۱۲۵‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲۶‬‬

‫ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪﻫﺎﻱ ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﺗﻮﺿﻴﺢ ﺩﻫﻴﻢ‪.‬‬

‫ﺭﻳﺸﻪ ﻭ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ‬ ‫‪۱.۵‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ‪ f ،‬ﻳﻚ ﺗﺎﺑﻊ ﻳﻚ ﻣﺘﻐﻴﺮﻩ ﺣﻘﻴﻘﻲ ﺑﺎﺷﺪ‪ .‬ﻧﻘﻄﻪ ‪ α‬ﺭﺍ ﻳﻚ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ ،f ۱‬ﮔﻮﻳﻴﻢ ﻫﺮﮔﺎﻩ ‪ .f (α) = 0‬ﺩﺭ ﻭﺍﻗﻊ ﻣﻘﺪﺍﺭ‬
‫ﺗﺎﺑﻊ ﺩﺭ ﺭﻳﺸﻪﺍﺵ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﻣﻲﺷﻮﺩ‪ .‬ﻟﺬﺍ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f‬ﺑﻪ ﺻﻔﺮ ﺗﺎﺑﻊ ‪ f ۲‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻳﺸﻪ ﻋﺒﺎﺭﺕ ﺍﺳﺖ‬
‫ﺍﺯ ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺑﺎ ﻣﺤﻮﺭ ‪x‬ﻫﺎ‪ .‬ﻳﻚ ﺩﺳﺘﻪﺑﻨﺪﻱ ﺍﺯ ﺭﻳﺸﻪﻫﺎ ﺑﺮ ﺣﺴﺐ ﻣﺮﺗﺒﻪ ‪ ۳‬ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬

‫• ﺍﮔﺮ ‪ f (α) = 0‬ﻭ ‪ ،f ′ (α) ̸= 0‬ﺁﻥﮔﺎﻩ ‪ α‬ﺭﺍ ﻳﻚ ﺭﻳﺸﻪ ﺳﺎﺩﻩ ‪ ۴‬ﻭ ﻳﺎ ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ﻳﻚ ‪ ۵‬ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪ f‬ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫• ﺍﮔﺮ ‪ f (α) = f ′ (α) = 0‬ﻭ ‪ ،f ′′ (α) ̸= 0‬ﺁﻥﮔﺎﻩ ‪ α‬ﺭﺍ ﻳﻚ ﺭﻳﺸﻪ ﻣﻀﺎﻋﻒ ‪ ۶‬ﻭ ﻳﺎ ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ﺩﻭ ‪ ۷‬ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪f‬‬
‫ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫• ﺍﮔﺮ ‪ f (α) = f ′ (α) = · · · = f (k−1) (α) = 0‬ﻭ ‪ ،fk (α) ̸= 0‬ﺁﻥﮔﺎﻩ ‪ α‬ﺭﺍ ﻳﻚ ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ‪ ۸ k‬ﺑﺮﺍﻱ ﺗﺎﺑﻊ‬
‫‪ f‬ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫ﺍﮔﺮ ‪ α‬ﻳﻚ ﺭﻳﺸﻪ ﺳﺎﺩﻩ ﻭ ﻳﺎ ﻳﻚ ﺭﻳﺸﻪ ﺑﺎ ﻣﺮﺗﺒﻪ ﻓﺮﺩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺭﺍ ﻗﻄﻊ ﻣﻲﻛﻨﺪ ﺍﺯ ﺁﻥ‬
‫ﻋﺒﻮﺭ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ‪ α‬ﻳﻚ ﺭﻳﺸﻪ ﻣﻀﺎﻋﻒ ﻭ ﻳﺎ ﻳﻚ ﺭﻳﺸﻪ ﺑﺎ ﻣﺮﺗﺒﻪ ﺯﻭﺝ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺑﺮ ﻣﺤﻮﺭ‬
‫‪x‬ﻫﺎ ﻣﻤﺎﺱ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫‪ ،‬ﺩﻭ‪ ،‬ﺳﻪ ﻭ ﭼﻬﺎﺭ(‬ ‫ﺗﺎﺑﻊ ﺑﺎ ﺭﯾﺸەﻫﺎﯼ ﻣﺮﺗﺒﻪ ﯾ‬ ‫ﻣﺜﺎﻝ ‪)۱−۵‬ﯾ‬

‫‪ f (x) = e3(x−2) sin( x−1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﺯ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ﻣﻲﺗﻮﺍﻥ ﻓﻬﻤﻴﺪ‬‫ﺗﺎﺑﻊ )‪5 )(x − 2) (x − 3) (x − 4‬‬
‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬

‫ﻛﻪ ‪ α1 = 1‬ﺭﻳﺸﻪ ﺳﺎﺩﻩ‪ α2 = 2 ،‬ﺭﻳﺸﻪ ﻣﻀﺎﻋﻒ‪ α3 = 3 ،‬ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ﺳﻪ ﻭ ‪ α4 = 4‬ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ﭼﻬﺎﺭ ﺗﺎﺑﻊ ﺍﺳﺖ‪ .‬ﺭﻓﺘﺎﺭ‬
‫ﺗﺎﺑﻊ ﺩﺭ ﺍﻳﻦ ﺭﻳﺸﻪﻫﺎ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺑﻪ ﻧﻤﺎﻳﺶ ﺩﺭ ﺁﻣﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪1‬‬

‫ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ‪۱‬‬ ‫ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ‪۲‬‬

‫‪0.5‬‬

‫‪x‬‬
‫‪1‬‬ ‫‪1.5‬‬ ‫‪2‬‬ ‫‪2.5‬‬ ‫‪3‬‬ ‫‪3.5‬‬ ‫‪4‬‬ ‫‪4.5‬‬

‫‪−0.5‬‬
‫ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ‪۳‬‬ ‫ﺭﻳﺸﻪ ﻣﺮﺗﺒﻪ ‪۴‬‬

‫‪−1‬‬

‫‪1 Root‬‬ ‫‪of function‬‬ ‫‪5 Root‬‬ ‫‪of multiplicity one‬‬

‫‪2 Zero‬‬ ‫‪of function‬‬ ‫‪6 Double‬‬ ‫‪root‬‬
‫‪3 Multiplicity‬‬ ‫‪7 Root of multiplicity two‬‬
‫‪4 Simple root‬‬ ‫‪8 Root of multiplicity k‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۲۷‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﻳﻚ ﺗﺎﺑﻊ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺻﻼ ﺭﻳﺸﻪ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﻣﺎﻧﻨﺪ ﺗﺎﺑﻊ ‪ .f (x) = ex‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﻳﻚ ﺭﻳﺸﻪ ﺳﺎﺩﻩ‬
‫ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺑﺎﺷﺪ‪ .‬ﻣﺎﻧﻨﺪ ﺗﺎﺑﻊ )‪ .f (x) = ln(x‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺗﺎﺑﻊ ﺩﻭ ﻳﺎ ﭼﻨﺪ ﺭﻳﺸﻪ ﺳﺎﺩﻩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺮﺧﻲ ﺗﻮﺍﺑﻊ‬
‫ﺩﺍﺭﺍﻱ ﺑﻲﻧﻬﺎﻳﺖ ﺭﻳﺸﻪ ﺳﺎﺩﻩ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺑﻪ ﺗﺎﺑﻊ )‪ f (x) = sin(x‬ﻣﻲﺗﻮﺍﻥ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪.‬‬
‫ﻣﺘﺎﺳﻔﺎﻧﻪ ﺩﺭ ﺭﻳﺎﺿﻴﺎﺕ ﺍﺑﺰﺍﺭﻱ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ ﻛﻪ ﺩﺭ ﻣﻮﺭﺩ ﻭﺟﻮﺩ ﻭ ﺗﻌﺪﺍﺩ ﺭﻳﺸﻪﻫﺎﻱ ﻳﻚ ﺗﺎﺑﻊ ﺩﻟﺨﻮﺍﻩ ﺍﻃﻼﻋﺎﺕ ﻛﺎﻣﻠﻲ ﺑﻪ ﻣﺎ‬
‫ﺑﺪﻫﺪ‪ .‬ﺍﻣﺎ ﻗﻀﺎﻳﺎﻳﻲ ﻭﺟﻮﺩ ﺩﺍﺭﻧﺪ ﻛﻪ ﺩﺭﺣﺎﻻﺕ ﺧﺎﺹ ﺍﻃﻼﻋﺎﺗﻲ ﺩﺭ ﺭﻳﺸﻪﻫﺎ ﻣﻲﺩﻫﻨﺪ‪ .‬ﻗﻀﺎﻳﺎﻱ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫ﻗﻀﯿﻪ ‪۱−۵‬‬

‫• ﺍﮔﺮ ‪ f‬ﺗﺎﺑﻌﻲ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ ﻭ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ‪) f (x) < 0‬ﻳﺎ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ‬
‫‪′‬‬

‫]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ‪(f ′ (x) > 0‬‬

‫ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ‪ f‬ﺣﺪﺍﻛﺜﺮ ﺩﺍﺭﺍﻱ ﻳﻚ ﺭﻳﺸﻪ ﻣﺠﺰﺍ ﺩﺭ ﺑﺎﺯﻩ )‪ (a, b‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• ﺍﮔﺮ ‪ f‬ﺗﺎﺑﻌﻲ ﺩﻭ ﺑﺎﺭ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ ﻭ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ‪) f ′′ (x) < 0‬ﻳﺎ ﺑﻪ‬
‫ﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ‪ (f ′′ (x) > 00‬ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ‪ f‬ﺣﺪﺍﻛﺜﺮ ﺩﺍﺭﺍﻱ ﺩﻭ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (a, b‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻗﻀﻴﻪ ﻓﻮﻕ ﺍﺛﺒﺎﺕ ﻧﻤﻲﺷﻮﺩ‪ .‬ﻭﻟﻲ ﺑﺮﺍﻱ ﺩﺭﻙ ﺷﻬﻮﺩﻱ ﻗﺴﻤﺖ ﺩﻭﻡ ﻗﻀﻴﻪ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ ﺭﺍ ﻣﻼﺣﻈﻪ ﻛﻨﻴﺪ‪.‬‬

‫‪x‬‬ ‫‪x‬‬ ‫‪x‬‬

‫•‬ ‫•‬ ‫•‬

‫ﻗﻀﻴﻪ ﻓﻮﻕ ﺩﺭ ﻣﻮﺭﺩ ﺗﻌﺪﺍﺩ ﺭﻳﺸﻪﻫﺎ ﺍﻃﻼﻋﺎﺗﻲ ﺍﺭﺍﻳﻪ ﻣﻲﺩﻫﺪ ﻭ ﻗﻀﻴﻪ ﺯﻳﺮ ﻛﻪ ﺑﻪ ﻗﻀﻴﻪ ﺑﻮﻟﺰﺍﻧﻮ ‪ ۹‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ ،‬ﺷﺮﺍﻳﻂ ﻛﺎﻓﻲ‬
‫ﺑﺮﺍﻱ ﻭﺟﻮﺩ ﺭﻳﺸﻪ ﺭﺍ ﺍﺭﺍﻳﻪ ﻣﻲﺩﻫﺪ‪.‬‬
‫ﻗﻀﯿﻪ ﺑﻮﻟﺰﺍﻧﻮ‬ ‫ﻗﻀﯿﻪ ‪۲−۵‬‬

‫ﺍﮔﺮ ‪ f‬ﺗﺎﺑﻌﻲ ﭘﻴﻮﺳﺘﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ‪ f (a)f (b) < 0‬ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ‪ f‬ﺣﺪﺍﻗﻞ ﺩﺍﺭﺍﻱ ﻳﻚ ﺭﻳﺸﻪ‬
‫ﺩﺭ ﺑﺎﺯﻩ )‪ (a, b‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﻄﺎﺑﻖ ﺑﺎ ﻗﻀﻴﻪ ﺑﻮﻟﺰﺍﻧﻮ‪ ،‬ﻳﺎﻓﺘﻦ ﻳﻚ ﺑﺎﺯﻩ ﺣﺎﻭﻱ ﺭﻳﺸﻪ ﻣﻨﻮﻁ ﺑﻪ ﻳﺎﻓﺘﻦ ﺩﻭ ﻧﻘﻄﻪ ‪ a‬ﻭ ‪ b‬ﺍﺳﺖ ﻛﻪ ﻋﻼﻣﺖ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺩﻭ‬
‫ﻧﻘﻄﻪ ‪ a‬ﻭ ‪ b‬ﻣﺘﻔﺎﻭﺕ ﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﻗﻀﻴﻪ ﻭﺟﻮﺩ ﺣﺪﺍﻗﻞ ﻳﻚ ﺭﻳﺸﻪ ﺭﺍ ﺗﻀﻤﻴﻦ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻴﺶ ﺍﺯ ﻳﻚ ﺭﻳﺸﻪ ﻧﻴﺰ‬
‫ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪.‬‬
‫‪۱۰‬‬
‫ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﻳﺎ ﺭﻳﺸﻪﻫﺎﻱ ﻳﻚ ﺗﺎﺑﻊ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﻪ ﻣﺴﺎﻟﻪ ﺣﻞ ﻣﻌﺎﺩﻟﻪ‬
‫ﻳﻚ ﻣﺠﻬﻮﻟﻲ ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‪ ،‬ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪﻫﺎﻱ ﺗﺎﺑﻊ )‪ f (x‬ﻫﻤﺎﻥ ﺣﻞ ﻣﻌﺎﺩﻟﻪ ﻳﻚ ﻣﺠﻬﻮﻟﻲ ‪ f (x) = 0‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺑﺮﺍﻱ ﺣﻞ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ )ﻳﺎ ﺣﻞ ﻣﻌﺎﺩﻟﻪ ﻳﻚ ﻣﺠﻬﻮﻟﻲ( ﺭﻭﺵﻫﺎﻱ ﺯﻳﺮ ﻣﻤﻜﻦ ﺍﺳﺖ ﻣﻮﺭﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻗﺮﺍﺭ ﮔﻴﺮﺩ‬
‫‪9 Bolzano’s‬‬ ‫‪theorem‬‬ ‫‪10 Root-finding‬‬ ‫‪problem‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۲۸‬‬

‫ﺗﺤﻠﻴﻠﻲ ‪۱۱‬‬ ‫• ﺭﻭﺵﻫﺎﻱ‬

‫ﺧﻄﺎ ‪۱۲‬‬ ‫• ﺭﻭﺵﻫﺎﻱ ﺳﻌﻲ ﻭ‬

‫• ﺭﻭﺵﻫﺎﻱ ﺗﺮﺳﻴﻤﻲ‬

‫ﺗﻜﺮﺍﺭﻱ ‪۱۴‬‬ ‫• ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ‪ ۱۳‬ﻳﺎ ﺭﻭﺵﻫﺎﻱ‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﻃﻮﺭ ﻣﺠﻤﻞ‪ ،‬ﺳﻪ ﺭﻭﺵ ﺍﻭﻝ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﺯﻳﺮﺑﺨﺶ ﺑﻌﺪ‪،‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﻪ ﻃﻮﺭ ﻣﻔﺼﻞ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺭﻭﺵﻫﺎﻱ ﺗﺤﻠﻴﻠﻲ‬
‫ﺟﺒﺮﻱ ﺳﻌﻲ ﻣﻲﺷﻮﺩ ﺭﻳﺸﻪ ﻳﺎ ﺭﻳﺸﻪﻫﺎﻱ ﺗﺎﺑﻊ ﺍﺳﺘﺨﺮﺍﺝ ﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﻧﻤﻮﻧﻪ ﺍﺯ ﺭﻭﺵ‬ ‫√‬
‫ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﺤﻠﻴﻠﻲ ﺑﺎ ﻛﻤﻚ ﺍﺑﺰﺍﺭ‬
‫= ‪ x1,2‬ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺗﺎﺑﻊ ‪ f (x) = ax + bx + c‬ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ‬
‫‪2‬‬ ‫‪−b± b2 −4ac‬‬
‫‪2a‬‬ ‫ﺗﺤﻠﻴﻠﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻣﻮﻝ‬
‫ﻓﺮﻣﻮﻝ ﺑﺎ ﻛﻤﻚ ﺭﺍﺩﻳﻜﺎﻝ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﻣﻲﺗﻮﺍﻥ ﺭﻳﺸﻪﻫﺎﻱ ﺗﺎﺑﻊ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ‬
‫ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﺑﺮﺍﻱ ﻫﺮ ﺗﺎﺑﻊ ﺩﻟﺨﻮﺍﻩ ‪ f‬ﻣﻲﺗﻮﺍﻥ ﻳﻚ ﻓﺮﻣﻮﻝ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻫﺎ ﺍﺭﺍﻳﻪ ﻛﺮﺩ ﻭ ﻳﺎ ﺑﺎ ﺩﺳﺘﻜﺎﺭﻱﻫﺎﻱ ﺟﺒﺮﻱ ﺭﻳﺸﻪ ﺭﺍ‬
‫ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪ .‬ﺟﻮﺍﺏ ﺍﻳﻦ ﺳﻮﺍﻝ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺩﺳﺘﻪ ﺑﺴﻴﺎﺭ ﻛﻮﭼﻜﻲ ﺍﺯ ﺗﻮﺍﺑﻊ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﺗﺤﻠﻴﻠﻲ ﺭﻳﺸﻪﻳﺎﺑﻲ‬
‫ﻧﻤﻮﺩ‪ .‬ﺣﺘﻲ ﺩﺭ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )ﻛﻪ ﺍﺯ ﺳﺎﺩﻩﺗﺮﻳﻦ ﺗﻮﺍﺑﻊ ﻣﻲﺑﺎﺷﻨﺪ‪ (،‬ﺗﻨﻬﺎ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺣﺪﺍﻛﺜﺮ ﺍﺯ ﺩﺭﺟﻪ‬
‫ﭼﻬﺎﺭ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻫﺮﭼﻨﺪ ﻛﻪ ﺭﻭﺵﻫﺎﻱ ﺗﺤﻠﻴﻠﻲ ﻣﻄﻠﻮﺏ ﻭ ﻣﻨﺎﺳﺐ ﻫﺴﺘﻨﺪ‪ ،‬ﺍﻣﺎ ﻧﻤﻲﺗﻮﺍﻥ ﺁﻥﻫﺎ ﺭﺍ ﺑﺮﺍﻱ‬
‫ﻫﺮ ﺗﺎﺑﻊ ﺩﻟﺨﻮﺍﻩ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬

‫■ ﺭﻭﺵﻫﺎﻱ ﺳﻌﻲ ﻭ ﺧﻄﺎ‬

‫ﺭﻭﺵﻫﺎﻱ ﺳﻌﻲ ﻭ ﺧﻄﺎ ﺭﻭﺵﻫﺎﻳﻲ ﻫﺴﺘﻨﺪ ﻛﻪ ﺑﺮ ﺍﻳﺪﻩ ﺟﺴﺘﺠﻮﻱ ﺭﻳﺸﻪ ﺍﺳﺘﻮﺍﺭ ﻫﺴﺘﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﺟﺴﺘﺠﻮ ﺍﺯ ﻣﻨﻄﻖ ﺭﻳﺎﺿﻲ ﺧﺎﺻﻲ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻲﻛﻨﻨﺪ ﺑﻠﻜﻪ ﺑﻪ ﻃﻮﺭ ﺑﻲﺧﺮﺩﺍﻧﻪ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﺗﻌﺪﺍﺩ ﺯﻳﺎﺩﻱ ﺍﺯ ﻧﻘﺎﻁ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻨﺪ ﻭ ﺑﺮ ﺣﺴﺐ ﺍﻳﻦ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﺭﻳﺸﻪ ﺭﺍ ﺗﺨﻤﻴﻦ ﻣﻲﺯﻧﻨﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻧﻤﻮﻧﻪﺍﻱ ﺳﺎﺩﻩ ﺍﺯ ﺍﻳﻦ ﻧﻮﻉ ﺭﻭﺵﻫﺎ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﻳﺎ ﺭﻳﺸﻪﻫﺎﻱ‬
‫‪ f‬ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻫﺰﺍﺭﺍﻥ ﻧﻘﻄﻪ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ ﻭ ﺁﻥ ﻧﻘﻄﻪ ﻳﺎ ﻧﻘﺎﻃﻲ ﻛﻪ ﺑﻪ ﺻﻔﺮ ﻧﺰﺩﻳﻚﺍﻧﺪ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺭﻭﺵ ﭘﺮﻫﺰﻳﻨﻪ ﻣﻲﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺒﻲ ﺍﺯ ﺭﻳﺸﻪ ﺍﺭﺍﻳﻪ ﻧﻤﻲﻛﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎ ﺍﻋﻤﺎﻝ ﺑﻬﺒﻮﺩﻫﺎﻳﻲ‬
‫ﺭﻭﻱ ﺭﻭﺵ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﻗﺖ ﺭﺍ ﺍﻓﺰﺍﻳﺶ ﺩﺍﺩ ﻛﻪ ﺍﻟﺒﺘﻪ ﻣﺴﺘﻠﺰﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺑﻴﺸﺘﺮ ﺩﻳﮕﺮﻱ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪ .‬ﺩﺭ ﺗﻮﻟﻴﺪ ﻧﻘﺎﻁ ﻣﺬﻛﻮﺭ ﻣﻲﺗﻮﺍﻥ‬
‫ﺭﻭﻳﻜﺮﺩ ﻗﻄﻌﻲ ﻭ ﻳﺎ ﺗﺼﺎﺩﻓﻲ ﺭﺍ ﺩﻧﺒﺎﻝ ﻛﺮﺩ‪ .‬ﺩﺭ ﺭﻭﻳﻜﺮﺩ ﻗﻄﻌﻲ ﻣﻌﻤﻮﻻ ﺷﺒﻜﻪﺍﻱ ﺍﺯ ﻧﻘﺎﻁ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪،‬‬
‫ﺍﻣﺎ ﺩﺭ ﺭﻭﻳﻜﺮﺩ ﺗﺼﺎﺩﻓﻲ‪ ،‬ﻧﻘﺎﻁ ﺑﻪ ﺻﻮﺭﺕ ﺗﺼﺎﺩﻓﻲ ﺗﻮﻟﻴﺪ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫■ ﺭﻭﺵﻫﺎﻱ ﺗﺮﺳﻴﻤﻲ‬
‫ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﺮﺳﻴﻤﻲ ﺍﺯ ﺗﻮﺍﻧﺎﻳﻲ ﻛﺎﻣﭙﻴﻮﺗﺮﻫﺎ ﺩﺭ ﺭﺳﻢ ﺗﻮﺍﺑﻊ‪ ،‬ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺗﻘﺮﻳﺒﻲ ﺭﻳﺸﻪ ﻭ ﻳﺎ ﺣﺪﻭﺩ ﺭﻳﺸﻪ‪ ،‬ﻛﻤﻚ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‬
‫ﺩﺭ ﺳﺎﺩﻩﺗﺮﻳﻦ ﺣﺎﻟﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﺑﺎﺯﻩﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺭﺳﻢ ﻛﺮﺩ ﻭ ﺳﭙﺲ ﺑﻪ ﺟﺴﺘﺠﻮﻱ ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺑﺎ‬
‫ﻣﺤﻮﺭ ‪ x‬ﭘﺮﺩﺍﺧﺖ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬
‫‪11 َAnalytical‬‬ ‫‪methods‬‬ ‫‪13 Numerical‬‬ ‫‪methods‬‬
‫‪12 Try‬‬ ‫‪and error‬‬ ‫‪14 Iterative‬‬ ‫‪methods‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۲۹‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۲−۵‬ﺭﻭﺵ ﺗﺮﺳﯿﻤ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺗﻘﺮﯾﺒﯽ ﺭﯾﺸﻪ ﯾ‬

‫‬ ‫‪−x‬‬
‫‬
‫‪ f (x) = 668.0610‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﻳﺎ ﺭﻳﺸﻪﻫﺎﻱ ﺗﻘﺮﻳﺒﻲ ﺁﻥ ﺭﺍ ﺑﺎ ﺭﻭﺵ‬
‫‪x‬‬ ‫‪1‬‬ ‫‪−‬‬ ‫‪e‬‬ ‫‪6.81‬‬ ‫ﺗﺎﺑﻊ ‪− 40‬‬
‫ﺗﺮﺳﻴﻤﻲ ﺑﻴﺎﺑﻴﻢ‪.‬‬
‫‪150‬‬ ‫‬ ‫‪−x‬‬
‫‬
‫= )‪f (x‬‬ ‫‪668.0610‬‬
‫‪x‬‬ ‫‪1 − e 6.81 − 40‬‬
‫ﺍﺑﺘﺪﺍ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [−20, 200‬ﺭﺳﻢ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫‪100‬‬
‫ﺷﻜﻞ ﺭﻭﺑﺮﻭ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﺎﺑﻊ‬
‫ﻳﻚ ﺭﻳﺸﻪ ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ ﺩﺍﺭﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ‬
‫‪50‬‬ ‫ﻛﻪ ﺗﺎﺑﻊ ﻧﺰﻭﻟﻲ ﺍﺳﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﺭﻳﺎﻓﺖ ﻛﻪ ﺍﻳﻦ ﺗﺎﺑﻊ‬
‫ﺩﺭ ﺧﺎﺭﺝ ﺍﺯ ﺑﺎﺯﻩ ]‪ [−20, 200‬ﺭﻳﺸﻪﺍﻱ ﻧﺪﺍﺭﺩ‪ .‬ﺑﺎ‬
‫‪x‬‬ ‫ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﺷﻜﻞ‪ ،‬ﻧﻤﻲﺗﻮﺍﻥ ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺒﻲ ﺍﺯ‬
‫‪0‬‬ ‫•‬ ‫ﺭﻳﺸﻪ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺭﻳﺸﻪﻱ‬
‫ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ ]‪ [10, 20‬ﻭﺍﻗﻊ ﺍﺳﺖ‪.‬‬
‫‪−50‬‬
‫‪0 20 40 60‬‬ ‫‪100‬‬ ‫‪200‬‬

‫‪20‬‬

‫‬ ‫‪−x‬‬
‫‬
‫= )‪f (x‬‬ ‫‪668.0610‬‬
‫‪x‬‬ ‫‪1 − e 6.81 − 40‬‬
‫‪10‬‬
‫ﺣﺎﻝ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺭﻭﺑﺮﻭ‪ ،‬ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ ]‪ [10, 20‬ﺭﺳﻢ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ‬
‫‪x‬‬
‫‪0‬‬ ‫•‬ ‫]‪ [14.5, 15‬ﻭﺍﻗﻊ ﺍﺳﺖ‪.‬‬

‫‪−10‬‬
‫‪10 11 12 13 14 15 16 17 18 19 20‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۳۰‬‬

‫‪0.8‬‬ ‫‬ ‫‬

‫‪−x‬‬
‫= )‪f (x‬‬ ‫‪668.0610‬‬
‫‪x‬‬ ‫‪1 − e 6.81 − 40‬‬

‫‪0.4‬‬ ‫ﺩﺭ ﺍﺩﺍﻣﻪ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺭﻭﺑﺮﻭ‪ ،‬ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ ]‪[14.5, 15‬‬
‫ﺭﺳﻢ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻳﺸﻪ ﺣﺪﻭﺩﺍ‬
‫‪x‬‬ ‫ﺑﺮﺍﺑﺮ ‪ 14.8‬ﺍﺳﺖ‪ .‬ﺍﻟﺒﺘﻪ ﺑﺎﺯ ﻫﻢ ﻣﻲﺗﻮﺍﻥ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ‬
‫‪0‬‬ ‫•‬ ‫ﻳﻚ ﺑﺎﺯﻩ ﺍﻃﺮﺍﻑ ‪ 14.8‬ﻣﺜﻼ ﺑﺎﺯﻩ ]‪[14.78, 14.82‬‬
‫ﺭﺳﻢ ﻧﻤﻮﺩ ﻭ ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺐﺗﺮﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ‪.‬‬
‫‪−0.4‬‬ ‫ﻭﻟﻲ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﺑﻪ ﻫﻤﻴﻦ ﺗﻘﺮﻳﺐ ﺑﺴﻨﺪﻩ ﻣﻲﻛﻨﻴﻢ‬
‫ﻭ ﺗﻜﺮﺍﺭﻫﺎﻱ ﺑﻌﺪﻱ ﺭﺍ ﺍﺩﺍﻣﻪ ﻧﻤﻲﺩﻫﻴﻢ‪.‬‬

‫‪−0.8‬‬
‫‪14.6‬‬ ‫‪14.7‬‬ ‫‪14.8‬‬ ‫‪14.9‬‬

‫ﺩﻳﺪﻳﻢ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﺑﺎ ﺳﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﺗﺮﺳﻴﻤﻲ‪ ،‬ﺗﻘﺮﻳﺐ ‪ 14.8‬ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﺣﺎﺻﻞ ﺷﺪ‪ .‬ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﺍﻳﻦ ﺗﻘﺮﻳﺐ‬
‫ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ .f (14.8) = 0.0022‬ﻫﻤﭽﻨﻴﻦ ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺭﻳﺸﻪ ﺗﺎ ﺣﺪﺍﻗﻞ ‪ ۱۱‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪.14.801135945‬‬

‫ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ﺍﻣﻜﺎﻥ ﺭﺳﻢ ﺩﻗﻴﻖ ﺗﺎﺑﻊ ﻓﺮﺍﻫﻢ ﻧﺒﻮﺩ‪ ،‬ﻭﻟﻲ ﺑﺘﻮﺍﻥ ﺗﺎﺑﻊ ﺑﻪ ﺻﻮﺭﺕ ﺗﻔﺮﻳﻖ ﺩﻭ ﺗﺎﺑﻊ ﺳﺎﺩﻩﺗﺮ ﻧﻮﺷﺖ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ‬
‫ﺭﻭﻳﻜﺮﺩ ﺩﻳﮕﺮﻱ ﺭﺍ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺣﺪﻭﺩﻱ ﺍﺯ ﺭﻳﺸﻪ ﺍﺭﺍﻳﻪ ﻛﺮﺩ‪ .‬ﺍﻳﻦ ﺭﻭﻳﻜﺮﺩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻔﻜﻴﻚ ﺗﺎﺑﻊ ﺑﻪ ﺻﻮﺭﺕ ﺗﻔﺮﻳﻖ ﺩﻭ ﺗﺎﺑﻊ‬
‫ﺳﺎﺩﻩﺗﺮ ﻭ ﺭﺳﻢ ﺗﻮﺍﺑﻊ ﺳﺎﺩﻩﺗﺮ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺍﺑﺘﺪﺍ ﺗﺎﺑﻊ )‪ f (x‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻣﻲﻛﻨﻴﻢ‬
‫‪f (x) = f1 (x) − f2 (x).‬‬
‫ﺑﻪﻃﻮﺭﻱ ﻛﻪ ﺗﻮﺍﺑﻊ )‪ f1 (x‬ﻭ )‪ f2 (x‬ﺍﺯ ﻧﻈﺮ ﺗﺮﺳﻴﻤﻲ ﺩﺍﺭﺍﻱ ﺿﺎﺑﻄﻪﻱ ﺳﺎﺩﻩﺗﺮﻱ ﻧﺴﺒﺖ ﺑﻪ ﺗﺎﺑﻊ ﺍﺻﻠﻲ )‪ f (x‬ﺑﺎﺷﻨﺪ‪ .‬ﻓﺮﺽ‬
‫ﻛﻨﻴﻢ ﻧﻘﻄﻪ ‪ α‬ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﺩﻭ ﺗﺎﺑﻊ )‪ f1 (x‬ﻭ )‪ f2 (x‬ﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪∃ α : f1 (α) = f2 (α),‬‬
‫ﺩﺭ ﻧﺘﻴﺠﻪ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬
‫‪f (α) = f1 (α) − f2 (α) = 0.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﻘﻄﻪﻱ ‪ α‬ﻫﻤﺎﻥ ﺭﻳﺸﻪﻱ ﺗﺎﺑﻊ )‪ f (x‬ﺩﺭ ﺑﺎﺯﻩﻱ ﻣﻮﺭﺩ ﻧﻈﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪،‬ﻛﻪ ﺣﺪﻭﺩ ﺁﻥ ﺍﺯ ﻃﺮﻳﻖ ﺭﺳﻢ ﻣﺸﺨﺺ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۳−۵‬ﺗﺨﻤﯿﻦ ﺣﺪﻭﺩ ﺭﯾﺸﻪ(‬

‫ﺗﺎﺑﻊ ‪ f (x) = cos(x) − x‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺍﺯ ﺩﻭ ﺗﺎﺑﻊ ﺳﺎﺩﻩﺗﺮ )‪ cos(x‬ﻭ ‪ x‬ﺗﺸﻜﻴﻞ ﺷﺪﻩ ﺍﺳﺖ‪ ،‬ﻛﻪ ﺭﺳﻢ ﻧﻤﻮﺩﺍﺭ‬
‫ﺁﻥﻫﺎ ﺳﺎﺩﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳۱‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫‪f2 (x) = x‬‬

‫‪1‬‬
‫‪x‬‬
‫‪π‬‬
‫‪−1‬‬ ‫‪2‬‬
‫)‪f1 (x) = cos(x‬‬

‫ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﻧﻤﻮﺩﺍﺭ )‪ cos(x‬ﻭ ‪ x‬ﻫﻤﺎﻥ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f‬ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [0, π2‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻭﺳﻂ ﺍﻳﻦ‬
‫ﺑﺎﺯﻩ‪ ،‬ﻳﻌﻨﻲ ‪ ، π4‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﻣﻌﺮﻓﻲ ﻧﻤﻮﺩ‪.‬‬

‫ﻣﺎﻫﻴﺖ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ‬ ‫�‬

‫ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ‪ ،‬ﺗﻨﻬﺎ ﺍﺯ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ )ﻭ ﻣﻘﺪﺍﺭ ﻣﺸﺘﻖ ﺗﺎﺑﻊ( ﺩﺭ ﺑﺮﺧﻲ ﻧﻘﺎﻁ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ ﻭ ﻳﻚ ﺗﻘﺮﻳﺐ ﺑﺮﺍﻱ‬
‫ﺭﻳﺸﻪ ﺍﺭﺍﻳﻪ ﻣﻲﻧﻤﺎﻳﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺭﻭﻱ ﺗﻤﺎﻣﻲ ﺗﻮﺍﺑﻊ ﺍﻋﻤﺎﻝ ﻧﻤﻮﺩ ﻭ ﻣﺤﺪﻭﺩﻳﺘﻲ ﺩﺭ ﺧﺼﻮﺹ ﻧﻮﻉ‬
‫ﻭ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻟﺒﺘﻪ‪ ،‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭﺳﺖ ﺍﺳﺖ ﻛﻪ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺭﻭﻱ ﻫﺮ‬
‫ﺗﺎﺑﻌﻲ ﻗﺎﺑﻞ ﺍﻋﻤﺎﻝ ﻣﻲﺑﺎﺷﻨﺪ‪ ،‬ﺍﻣﺎ ﻣﻮﻓﻘﻴﺖ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺩﺭ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﻣﺴﺘﻠﺰﻡ ﺑﺮﺧﻲ ﺷﺮﺍﻳﻂ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﻪ ﺟﺎﻱ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺩﺳﺘﻜﺎﺭﻱﻫﺎﻱ ﺟﺒﺮﻱ ﻭ ﻳﺎ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺟﺒﺮﻱ ﺍﺯ ﺗﻜﺮﺍﺭ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﻛﻤﻚ‬
‫ﻣﻲﮔﻴﺮﻧﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵﻫﺎ‪ ،‬ﺍﺑﺘﺪﺍ ﻳﻚ )ﺩﺭ ﺑﺮﺧﻲ ﺭﻭﺵﻫﺎ ﺩﻭ ﻧﻘﻄﻪ( ﺑﻪ ﻋﻨﻮﺍﻥ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺳﭙﺲ ﺑﺮﻣﺒﻨﺎﻱ ﺣﺪﺱ ﻳﺎ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ‪ ،‬ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻳﻚ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ‬
‫ﺷﻮﺩ‪ .‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻣﺘﻔﺎﻭﺕ ﺑﻪ ﻃﻮﺭ ﻣﺘﻔﺎﻭﺗﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺗﻔﺎﻭﺕ ﺍﺻﻠﻲ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺩﺭ ﻧﺤﻮﻩ‬
‫ﺗﻮﻟﻴﺪ ﺩﻧﺒﺎﻟﻪ ﻣﺬﻛﻮﺭ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻏﺎﻟﺒﺎ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺩﺭ ﺑﻲﻧﻬﺎﻳﺖ ﺑﻪ ﺭﻳﺸﻪ ﻣﻲﺭﺳﺪ ﻭ ﺩﺭ ﻋﻤﻞ ﺍﻣﻜﺎﻥ ﺗﻮﻟﻴﺪ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻤﻠﻪﻱ ﺩﻧﺒﺎﻟﻪ‬
‫ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲﺗﺎ ﺟﻤﻠﻪ ﺍﺯ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺳﺎﺧﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﺁﺧﺮﻳﻦ ﺟﻤﻠﻪ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﻣﻌﻴﺎﺭﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪ ۱۵‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻫﺮﮔﺎﻩ ﺍﻳﻦ ﻣﻌﻴﺎﺭ ﺑﺮﻗﺮﺍﺭ ﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻮﻟﻴﺪ‬
‫ﺟﻤﻠﻪ ﺟﺪﻳﺪ ﺍﺯ ﺩﻧﺒﺎﻟﻪ ﻣﺘﻮﻗﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺮ ﻣﺒﻨﺎﻱ ﺍﻳﻦ ﺗﻮﺿﻴﺤﺎﺕ ﻣﺨﺘﺼﺮ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺳﻪ ﻣﻮﻟﻔﻪ‬
‫ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪.‬‬
‫‪ .۱‬ﺣﺪﺱ ﻳﺎ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ‬
‫‪ .۲‬ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭ‬
‫‪ .۳‬ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‬

‫■ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺍﻣﺎ ﺟﻤﻠﻪ ﺍﻭﻝ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﺎﻳﺪ ﺑﻪ ﺭﻭﺵ ﺩﺍﺩﻩ ﺷﻮﺩ ﻭ ﺳﭙﺲ‬
‫ﺭﻭﺵ ﺑﺮﺣﺴﺐ ﺍﻳﻦ ﺟﻤﻠﻪ ﺍﻭﻝ )ﻛﻪ ﺁﻥ ﺭﺍ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻧﺎﻣﻴﺪﻳﻢ( ﺑﻪ ﺗﻮﻟﻴﺪ ﺩﻧﺒﺎﻟﻪ ﻣﻲﭘﺮﺩﺍﺯﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ‬
‫ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺟﻤﻠﻪ ﺍﻭﻝ ﭼﮕﻮﻧﻪ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﺩ‪ .‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪﻱ ﺩﻟﺨﻮﺍﻫﻲ ﺭﺍ ﺍﻧﺘﺨﺎﺏ ﻧﻤﻮﺩ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ‬
‫ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺼﻞ ﺧﻮﺍﻫﻴﻢ ﺩﻳﺪ‪ ،‬ﺍﻧﺘﺨﺎﺏ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺴﻴﺎﺭ ﻣﻬﻢ ﺍﺳﺖ ﻭ ﺗﺎﺛﻴﺮ ﻣﻬﻤﻲ ﺩﺭ ﻣﻮﻓﻘﻴﺖ ﺭﻭﺵ ﻋﺪﺩﻱ ﺩﺭ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ‬
‫ﺩﺍﺭﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﻧﺘﺨﺎﺏ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻫﻤﻮﺍﺭﻩ ﻧﻤﻲﺗﻮﺍﻧﺪ ﺑﻪ ﺻﻮﺭﺕ ﺗﺼﺎﺩﻓﻲ ﺍﻧﺠﺎﻡ ﺷﻮﺩ ﻭ ﻧﻴﺎﺯﻣﻨﺪ ﺑﺮﺭﺳﻲ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﻭﺍﻗﻊ ﻳﺎﻓﺘﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻫﻤﺎﻥ ﻳﺎﻓﺘﻦ ﺣﺪﻭﺩ ﺭﻳﺸﻪ ﺍﺳﺖ‪ .‬ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﻓﻴﺰﻳﻚ ﻣﺴﺎﻟﻪ ﻛﻤﻚ ﻣﻲﻛﻨﺪ ﺗﺎ ﺣﺪﺳﻲ ﺑﺮﺍﻱ‬
‫ﺣﺪﻭﺩ ﺭﻳﺸﻪ ﺍﺭﺍﻳﻪ ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺑﺮﺧﻲ ﺗﻮﺍﺑﻊ ﻧﻴﺰ ﺑﺎ ﺗﺤﻠﻴﻞ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺣﺪﻭﺩ ﺭﻳﺸﻪ ﺭﺍ ﺗﺨﻤﻴﻦ ﺯﺩ ﻭ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻳﺎﻓﺖ‪.‬‬
‫‪15 Stopping‬‬ ‫‪criterion‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۳۲‬‬

‫ﻫﻤﭽﻨﻴﻦ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺭﻭﺵﻫﺎﻱ ﺗﺮﺳﻴﻤﻲ ﻭ ﺳﻌﻲ ﻭ ﺧﻄﺎ ﻧﻴﺰ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻳﻚ ﺗﻘﺮﻳﺐ ﻧﻪﭼﻨﺪﺍﻥ ﺩﻗﻴﻖ ﺍﺯ ﺭﻳﺸﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ ﻭ‬
‫ﺳﭙﺲ ﺁﻥ ﺗﻘﺮﻳﺐ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬

‫■ ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭ‬
‫ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ ﻛﻪ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﺑﻪ ﺭﻳﺸﻪ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ .‬ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﺻﻮﺭﺕ ﭘﻲﺩﺭﭘﻲ )ﻭ ﻧﻪ ﻳﻜﺠﺎ(‬
‫ﺗﻮﻟﻴﺪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﺑﺘﺪﺍ ﺣﺪﺱ ﻳﺎ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺑﺎ ﻛﻤﻚ ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭ ﺟﻤﻼﺕ ﺩﻧﺒﺎﻟﻪ‬
‫ﺑﻪ ﺻﻮﺭﺕ ﻣﺘﻮﺍﻟﻲ ﺗﻮﻟﻴﺪ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭ ﺑﺎ ﺷﺮﻭﻉ ﺍﺯ ﺣﺪﺱ ﻳﺎ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ‬
‫∞} ‪ {xk‬ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﺪ ﻛﻪ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻳﻚ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪.‬‬
‫‪k=0‬‬

‫ﺗﻔﺎﻭﺕ ﻋﻤﺪﻩ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ‪ ،‬ﺩﺭ ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭ ﺁﻥﻫﺎ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺼﻞ ﻧﺸﺎﻥ‬
‫ﺧﻮﺍﻫﻴﻢ ﺩﺍﺩ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻣﺘﻔﺎﻭﺕ‪ ،‬ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ‪.‬‬

‫■ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‬
‫∞‬
‫ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺣﺪ ﺁﻥ ﻫﻤﺎﻥ‬ ‫‪{xk }k=0‬‬ ‫ﻫﺮ ﺭﻭﺵ ﻋﺪﺩﻱ‪ ،‬ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺩﺭ ﻋﻤﻞ ﻧﻤﻲﺗﻮﺍﻥ ﺣﺪ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ ،‬ﭼﺮﺍ ﻛﻪ ﺍﻧﺠﺎﻡ ﺑﻲﻧﻬﺎﻳﺖ ﺑﺎﺭ ﺗﻜﺮﺍﺭ ﻋﻤﻠﻲ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ‬
‫ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻓﻘﻂ ﺗﻌﺪﺍﺩﻱ ﻣﺘﻨﺎﻫﻲ ﺟﻤﻠﻪ ﺍﺯ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻴﻢ ﻭ ﺁﺧﺮﻳﻦ ﺟﻤﻠﻪ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺣﺪ ﺩﻧﺒﺎﻟﻪ )ﻛﻪ ﻫﻤﺎﻥ‬
‫ﺭﻳﺸﻪ ﺍﺳﺖ( ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻫﺮ ﭼﻪ ﺗﻌﺪﺍﺩ ﺟﻤﻼﺕ ﺑﻴﺸﺘﺮﻱ ﺗﻮﻟﻴﺪ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺩﻗﺖ ﺑﻴﺸﺘﺮﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪ‬
‫ﺑﻪ ﺩﺳﺖ ﺁﻳﺪ‪ .‬ﺍﻣﺎ ﺳﻮﺍﻝ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﭼﻪ ﺗﻌﺪﺍﺩ ﺍﺯ ﺟﻤﻼﺕ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺑﺎﻳﺪ ﺗﻮﻟﻴﺪ ﻧﻤﻮﺩ‪ .‬ﭘﺎﺳﺦ ﺑﻪ ﺍﻧﺘﻈﺎﺭ ﻣﺎ ﺍﺯ ﺩﻗﺖ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪.‬‬
‫ﺩﺭ ﻭﺍﻗﻊ ﺑﺎﻳﺪ ﺑﺮﻣﺒﻨﺎﻱ ﺍﻧﺘﻈﺎﺭﻱ ﻛﻪ ﺍﺯ ﺩﻗﺖ ﺩﺍﺭﻳﻢ‪ ،‬ﻳﻚ ﻣﺤﻚ ﻳﺎ ﻣﻌﻴﺎﺭ ﻣﻮﺳﻮﻡ ﺑﻪ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻣﺸﺨﺺ ﻧﻤﺎﻳﻴﻢ‪ .‬ﺳﭙﺲ ﻫﺮﮔﺎﻩ‬
‫ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﺮﻗﺮﺍﺭ ﮔﺮﺩﻳﺪ‪ ،‬ﺗﻮﻟﻴﺪ ﺟﻤﻠﻪ ﺟﺪﻳﺪ ﺍﺯ ﺩﻧﺒﺎﻟﻪ ﻣﺘﻮﻗﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻫﺮﮔﺎﻩ ﺟﻤﻠﻪ ﺟﺪﻳﺪﻱ ﺍﺯ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ‬
‫ﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺭﺍ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﮔﺮ ﺍﻳﻦ ﻣﻌﻴﺎﺭ ﺑﺮﻗﺮﺍﺭ ﺑﻮﺩ‪ ،‬ﻛﻪ ﺗﻮﻗﻒ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺻﻮﺭﺕ ﺑﺮﻗﺮﺍﺭ ﻧﺒﻮﺩﻥ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‪،‬‬
‫ﺑﻪ ﺗﻮﻟﻴﺪ ﺩﻧﺒﺎﻟﻪ ﺍﺩﺍﻣﻪ ﻣﻲﺩﻫﻴﻢ‪.‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻳﻚ ﺭﻭﺵ ﻋﺪﺩﻱ ﺩﻧﺒﺎﻟﻪ } ‪ {xk‬ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﭘﺮﻛﺎﺭﺑﺮﺩ ﺍﺷﺎﺭﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﻣﻌﻴﺎﺭﻫﺎ‪ ،‬ﻳﻚ ﻋﺪﺩ ﻛﻮﭼﻚ ﻭ ﻣﺜﺒﺖ ‪ ε‬ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺎ ﺍﻳﻦ ﻋﺪﺩ ﺍﻧﺘﻈﺎﺭ ﺧﻮﺩ ﺭﺍ ﺍﺯ ﺧﻄﺎ ﻣﺸﺨﺺ ﻣﻲﻧﻤﺎﻳﻴﻢ‪ .‬ﻫﺮﭼﻪ ﻋﺪﺩ‬
‫ﻛﻮﭼﻚﺗﺮ ﺍﻧﺘﺨﺎﺏ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﻮﻗﻊ ﺩﻗﺖ ﺑﻴﺸﺘﺮﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬

‫• ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ‪ ،‬ﻫﺮ ﮔﺎﻩ‬

‫‪|f (xk )| < ε.‬‬ ‫)‪(۱.۵‬‬

‫• ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ‪ ،‬ﻫﺮ ﮔﺎﻩ‬

‫‪|xk − α| < ε,‬‬ ‫)‪(۲.۵‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ α‬ﺭﻳﺸﻪ ﺍﺳﺖ‪ .‬ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺍﻳﻦ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻫﻤﻮﺍﺭﻩ ﻣﻤﻜﻦ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﻣﻘﺪﺍﺭ ﺭﻳﺸﻪ ‪ α‬ﺩﺭ ﺩﺳﺖ‬
‫ﻧﻴﺴﺖ‪ .‬ﺩﺭ ﻣﻮﺍﺭﺩﻱ ﻛﻪ ﺑﺘﻮﺍﻥ ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ |‪ |xk − α‬ﻳﺎﻓﺖ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺍﻳﻦ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻛﺮﺩ‪.‬‬

‫• ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ‪ ،‬ﻫﺮ ﮔﺎﻩ‬

‫‪|xk − xk−1 | < ε.‬‬ ‫)‪(۳.۵‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳۳‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫• ﺩﺭ ﺗﻜﺮﺍﺭ ‪k‬ﺍﻡ ﻣﺘﻮﻗﻒ ﺷﻮ‪ ،‬ﻫﺮ ﮔﺎﻩ‬

‫| ‪|xk − xk−1‬‬
‫‪< ε.‬‬ ‫)‪(۴.۵‬‬
‫| ‪|xk‬‬
‫ﺩﺭ ﻋﻤﻞ ﻣﻤﻜﻦ ﺍﺳﺖ ﺗﺮﻛﻴﺒﻲ ﺍﺯ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﻓﻮﻕ ﺑﻪ ﻛﺎﺭ ﺭﻭﺩ‪.‬‬

‫ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‬ ‫‪۲.۵‬‬

‫ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ‪ ۱۶‬ﺍﺯ ﺍﺑﺘﺪﺍﻳﻲﺗﺮﻳﻦ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺭﻭﺵ ﻳﻚ ﺑﺎﺯﻩ ]‪ [a, b‬ﻛﻪ ‪f (a)f (b) < 0‬‬
‫ﺭﺍ ﻣﻲﮔﻴﺮﺩ ﻭ ﺑﺎ ﻛﻤﻚ ﻗﻀﻴﻪ ﺑﻮﻟﺰﺍﻧﻮ )ﻗﻀﻴﻪ ‪ ۲-۵‬ﺹ ؟(‪ ،‬ﺍﻳﻦ ﺑﺎﺯﻩ ﺭﺍ ﺑﻪ ﺑﺎﺯﻩﺍﻱ ﻛﻮﭼﻚﺗﺮ ﻭ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺑﺎ ﺗﻜﺮﺍﺭ‬
‫ﺍﻳﻦ ﻓﺮﺍﻳﻨﺪ‪ ،‬ﺑﺎﺯﻩﺍﻱ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻛﺎﻓﻲ ﻛﻮﭼﻚ ﻭ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﻳﺎﻓﺖ ﻣﻲﺷﻮﺩ ﻭ ﻧﻘﻄﻪ ﻭﺳﻂ ﺍﻳﻦ ﺑﺎﺯﻩ ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﻪ ﺭﻭﺵ ﺗﻨﺼﻴﻒ ﺑﺎﺯﻩ ‪ ۱۷‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﺎﻡﻫﺎﻱ ﺧﺮﺩﻛﺮﺩﻥ ﺩﻭﺩﻭﻳﻲ ‪ ۱۸‬ﻭ ﺭﻭﺵ ﺑﻮﻟﺰﺍﻧﻮ ﻧﻴﺰ ﺑﺮﺍﻱ‬
‫ﺍﺷﺎﺭﻩ ﺑﻪ ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﻧﺪ‪.‬‬
‫ﺑﺮﺍﻱ ﺗﻮﺿﻴﺢ ﺭﻭﺵ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺑﺎﺯﻩ ]‪ [a, b‬ﻛﻪ ‪ f (a)f (b) < 0‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻭﺭﻭﺩﻱ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ ﻭ ﻋﻼﻭﻩ ﺑﺮ ﺁﻥ‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ ﺩﻗﻴﻘﺎ ﻳﻚ ﺭﻳﺸﻪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﻪ ﺩﻭ ﺑﺎﺯﻩ ]‪ [a, c‬ﻭ ]‪[c, b‬‬
‫ﺗﻘﺴﻴﻢ ﻣﻲﻛﻨﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ c‬ﻧﻘﻄﻪ ﻭﺳﻂ ﺑﺎﺯﻩ ]‪ [a, b‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻳﻌﻨﻲ‬
‫‪a+b‬‬
‫=‪c :‬‬ ‫‪.‬‬ ‫)‪(۵.۵‬‬
‫‪2‬‬
‫ﺣﺎﻝ ﭼﻮﻥ ﺗﺎﺑﻊ ‪ f‬ﺩﺍﺭﺍﻱ ﻳﻚ ﺭﻳﺸﻪ ﺩﺭ ]‪ [a, b‬ﺩﺍﺭﺩ‪ ،‬ﻟﺬﺍ ﺍﮔﺮ ‪ c‬ﺭﻳﺸﻪ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺭﻳﺸﻪ ﺩﺭ ﻳﻜﻲ ﺍﺯ ﺑﺎﺯﻩﻫﺎﻱ )‪ (a, c‬ﻭ‬
‫)‪ (c, b‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﺍﺳﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺭﻳﺸﻪ ﺩﺭ ﻛﺪﺍﻡ ﺑﺎﺯﻩ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﭘﺎﺳﺦ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻝ ﻣﻲﺗﻮﺍﻥ ﺍﺯ‬
‫ﻋﻼﻣﺖ )‪ f (c‬ﻛﻤﻚ ﮔﺮﻓﺖ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ﻋﻼﻣﺖ )‪ f (c‬ﻭ )‪ f (a‬ﻳﻜﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (c, b‬ﻗﺮﺍﺭ ﺩﺍﺭﺩ ﻭ‬
‫ﺍﮔﺮ )‪ f (c‬ﺑﺎ )‪ f (b‬ﻫﻢﻋﻼﻣﺖ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (a, c‬ﻭﺍﻗﻊ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺍﮔﺮ ‪ f (a)f (c) > 0‬ﺁﻥﮔﺎﻩ‬
‫ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (c, b‬ﻭﺍﻗﻊ ﺍﺳﺖ ﻭ ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ‪ f (c)f (b) > 0‬ﺁﻥﮔﺎﻩ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (a, c‬ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺑﺎﺯﻩ‬
‫ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﺑﻪ ﺑﺎﺯﻩ ﺩﻳﮕﺮﻱ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ ﻛﻪ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﺍﺳﺖ ﻭ ﻃﻮﻝ ﺁﻥ ﻧﺼﻒ ﺑﺎﺯﻩ ﺍﻭﻟﻴﻪ ﺍﺳﺖ‪ .‬ﺑﺎ ﺗﻜﺮﺍﺭ ﺍﻳﻦ ﻓﺮﺍﻳﻨﺪ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﺎﺯﻩﺍﻱ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻛﺎﻓﻲ ﻛﻮﭼﻚ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﻳﺎﻓﺖ )ﺍﻟﺒﺘﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺧﻮﺩ ﺭﻳﺸﻪ ﻧﻴﺰ ﻳﺎﻓﺘﻪ ﺷﻮﺩ( ﻭ ﻫﺮ ﻧﻘﻄﻪ ﺍﺯ ﺍﻳﻦ ﺑﺎﺯﻩ‪ ،‬ﻣﺜﻼ‬
‫ﻭﺳﻂ ﺑﺎﺯﻩ‪ ،‬ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺭﻳﺸﻪ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺗﺎﻛﻨﻮﻥ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺮﺍﻱ ﺣﺎﻟﺘﻲ ﺗﺸﺮﻳﺢ ﺷﺪ ﻛﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺗﻨﻬﺎ ﻳﻚ ﺭﻳﺸﻪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ‬
‫ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺍﺭﺍﻱ ﺑﻴﺶ ﺍﺯ ﻳﻚ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﭼﮕﻮﻧﻪ ﻋﻤﻞ ﻣﻲﻛﻨﺪ‪.‬‬
‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ‪ ،‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻳﻜﻲ ﺍﺯ ﺭﻳﺸﻪ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻧﻤﺎﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺣﺎﻟﺖ ﻫﺮ ﮔﺎﻩ ﺑﺎﺯﻩ ﺑﻪ‬
‫ﺩﻭ ﻗﺴﻤﺖ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻳﻜﻲ ﺍﺯ ﺑﺎﺯﻩﻫﺎ ﺭﺍ ﺣﻔﻆ ﻭ ﺑﺎﺯﻩ ﺩﻳﮕﺮ ﺭﺍ ﺣﺬﻑ ﻣﻲﻛﻨﺪ‪ .‬ﺁﻥ ﺑﺎﺯﻩﺍﻱ ﻛﻪ ﺣﻔﻆ‬
‫ﻣﻲﺷﻮﺩ ﺣﺘﻤﺎ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺎﺯﻩﺍﻱ ﻛﻪ ﺣﺬﻑ ﻣﻲﺷﻮﺩ ﻧﻴﺰ ﻣﻤﻜﻦ ﺍﺳﺖ ﺷﺎﻣﻞ ﺭﻳﺸﻪ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺭﻭﺵ‬
‫ﺩﻭﺑﺨﺸﻲ ﻓﻘﻂ ﻳﻚ ﺭﻳﺸﻪ ﺭﺍ ﻣﻲﻳﺎﺑﺪ ﻭ ﺑﻘﻴﻪ ﺭﻳﺸﻪﻫﺎ ﺭﺍ ﺣﺬﻑ ﻣﻲﻛﻨﺪ‪ .‬ﻣﻮﻗﻌﻴﺖ ﺭﻳﺸﻪﻫﺎ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺩﺭ ﻳﺎﻓﺘﻪ ﺷﺪﻧﺸﺎﻥ ﺗﻮﺳﻂ‬
‫ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻣﻮﺛﺮ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ ۱‬ﻣﺮﺍﺣﻞ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺍﻟﮕﻮﺭﻳﺘﻢ ﺩﻭﺑﺨﺸﻲ ﺩﻧﺒﺎﻟﻪ } ‪{xk‬‬
‫ﺭﺍ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﺪ‪ .‬ﻣﻮﻟﻔﻪﻫﺎﻱ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻫﻤﺎﻥ ﻧﻘﺎﻁ ﻭﺳﻂ ﺑﺎﺯﻩﻫﺎ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻛﻠﻴﻪ ﻣﻌﻴﺎﺭﻫﺎﻱ ﺗﻮﻗﻒ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺩﺭ ؟ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺩﺭ‬
‫ﺍﻳﻦ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﻧﻜﺘﻪﺍﻱ ﻛﻪ ﺩﺭ ﻣﻮﺭﺩ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﺁﻥ ﺍﺳﺖ ﻛﻪ‪ ،‬ﺑﺎ ﺷﺮﻭﻉ ﺍﺯ ﺑﺎﺯﻩ ]‪ ،[a, b‬ﺑﻌﺪ ﺍﺯ ‪k‬‬
‫‪ b−a‬ﺧﻮﺍﻫﻴﻢ ﺭﺳﻴﺪ ﻭ ﻭﺳﻂ ﺍﻳﻦ ﺑﺎﺯﻩ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪ .‬ﭼﻮﻥ ﺭﻳﺸﻪ ﻧﻴﺰ ﺩﺭ ﺍﻳﻦ‬ ‫‪2k‬‬
‫ﺗﻜﺮﺍﺭ‪ ،‬ﺑﻪ ﺑﺎﺯﻩﺍﻱ ﺑﺎ ﻃﻮﻝ‬
‫ﺑﺎﺯﻩ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ ،‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﻓﺎﺻﻠﻪ ﺭﻳﺸﻪ ﺍﺯ ﻭﺳﻂ ﺑﺎﺯﻩ ﻛﻤﺘﺮ ﺍﺯ ﻧﺼﻒ ﻃﻮﻝ ﺑﺎﺯﻩ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ‬
‫‪16 Bisection‬‬ ‫‪18 Binary‬‬ ‫‪chopping‬‬
‫‪17 Interval‬‬ ‫‪halving‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۳۴‬‬

‫ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ ۱‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ‪.‬‬

‫ﻭﺭﻭﺩﻱ‪ :‬ﺍﻟﻒ( ﺗﺎﺑﻊ ‪ f‬ﻭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﻃﻮﺭﻱﻛﻪ ‪ f‬ﺭﻭﻱ ﺑﺎﺯﻩ ﺩﺍﺩﻩ ﺷﺪﻩ ﭘﻴﻮﺳﺘﻪ ﺍﺳﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ‪،f (a)f (b) < 0‬‬
‫ﺏ( ﻳﻚ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‪.‬‬
‫ﺧﺮﻭﺟﻲ‪ :‬ﻳﻚ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ )‪.(a, b‬‬
‫‪ :۱‬ﻗﺮﺍﺭ ﺑﺪﻩ ‪ xu := b ،xl := a‬ﻭ ‪.k = 1‬‬
‫‪.xk := xl +x‬‬‫‪2‬‬
‫‪u‬‬
‫‪ :۲‬ﻗﺮﺍﺭ ﺑﺪﻩ‬
‫‪ :۳‬ﺍﮔﺮ ‪ f (xk ) = 0‬ﻭ ﻳﺎ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺮﻗﺮﺍﺭ ﺑﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ‪ xk‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﻣﻌﺮﻓﻲ ﻛﻦ ﻭ ﻣﺘﻮﻗﻒ ﺷﻮ‪.‬‬

‫ﺍﮔﺮ ‪ f (xl )f (xk ) > 0‬ﺁﻥﮔﺎﻩ ﻗﺮﺍﺭ ﺑﺪﻩ ‪ ،xl = xk‬ﺩﺭ ﻏﻴﺮ ﺍﻳﻦ ﺻﻮﺭﺕ ﻗﺮﺍﺭ ﺑﺪﻩ ‪.xu = xk‬‬ ‫‪:۴‬‬
‫ﻗﺮﺍﺭ ﺑﺪﻩ ‪ k = k + 1‬ﻭ ﺑﻪ ﻗﺪﻡ ‪ ۲‬ﺑﺮﻭ‪.‬‬ ‫‪:۵‬‬

‫ﮔﺮﻓﺖ ﻛﻪ‬
‫‪b−a‬‬
‫‪| xk − α |< k .‬‬
‫‪2‬‬
‫ﺑﺎ ﻛﻤﻚ ﺍﺯ ﺭﺍﺑﻄﻪ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪ |xk − α| < ε‬ﺭﺍ ﻧﻴﺰ ﺩﺭ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺑﺮﺍﻱ‬
‫ﺑﺮﻗﺮﺍﺭﻱ ﺍﻳﻦ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻛﺎﻓﻲ ﺍﺳﺖ ﺑﻪ ﺗﻌﺪﺍﺩ ‪ k = m‬ﺗﻜﺮﺍﺭ ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫‪b−a‬‬
‫‪< ε.‬‬
‫‪2m‬‬
‫ﺑﻪ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻣﻲﺗﻮﺍﻥ ‪ m‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻧﺘﺨﺎﺏ ﻧﻤﻮﺩ‬
‫‬ ‫‬ ‫‬
‫‪b−a‬‬
‫‪m = log2‬‬ ‫‪.‬‬
‫‪ε‬‬
‫ﻫﻨﮕﺎﻡ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺩﻭﺑﺨﺸﻲ‪ ،‬ﻻﺯﻡ ﻧﻴﺴﺖ ﻛﻪ ﻛﻠﻴﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ } ‪ {xk‬ﺭﺍ ﻧﮕﻪﺩﺍﺭﻱ ﻛﺮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﺍ‬
‫ﺑﺎ ﺗﻌﺪﺍﺩ ﺳﻪ ﻣﺘﻐﻴﺮ ﻣﻲﺗﻮﺍﻥ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ ﻧﻤﻮﺩ‪ .‬ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ ۲‬ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﺍﻟﮕﻮﺭﻳﺘﻢ ‪ ۲‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ)ﻧﮕﺎﺭﺵ ﺩﻭﻡ(‪.‬‬

‫ﻭﺭﻭﺩﻱ‪ :‬ﺍﻟﻒ( ﺗﺎﺑﻊ ‪ f‬ﻭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﻃﻮﺭﻱﻛﻪ ‪ f‬ﺭﻭﻱ ﺑﺎﺯﻩ ﺩﺍﺩﻩ ﺷﺪﻩ ﭘﻴﻮﺳﺘﻪ ﺍﺳﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ‪،f (a)f (b) < 0‬‬
‫ﺏ( ﻳﻚ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ‪.‬‬
‫ﺧﺮﻭﺟﻲ‪ :‬ﻳﻚ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ )‪.(a, b‬‬
‫‪.c = a+b‬‬‫‪ :۱‬ﻗﺮﺍﺭ ﺑﺪﻩ ‪2‬‬
‫‪ :۲‬ﺍﮔﺮ ‪ f (c) = 0‬ﻭ ﻳﺎ ‪ c‬ﺩﺭ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺍﺩﻩ ﺷﺪﻩ ﺻﺪﻕ ﻛﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ‪ c‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﻣﻌﺮﻓﻲ ﻛﻦ ﻭ ﻣﺘﻮﻗﻒ‬
‫ﺷﻮ‪.‬‬
‫‪ :۳‬ﺍﮔﺮ ‪ f (a)f (c) > 0‬ﺁﻥﮔﺎﻩ ﻗﺮﺍﺭ ﺑﺪﻩ ‪ ،a = c‬ﺩﺭ ﻏﻴﺮ ﺍﻳﻦ ﺻﻮﺭﺕ ﻗﺮﺍﺭ ﺑﺪﻩ ‪.b = c‬‬
‫‪ :۴‬ﺑﻪ ﻗﺪﻡ ‪ ۱‬ﺑﺮﻭ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳۵‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۴−۵‬ﺭﻭﺵ ﺩﻭﺑﺨﺸ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺭﯾﺸﻪ ﺗﻘﺮﯾﺒﯽ ﯾ‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﺗﺎﺑﻊ‬
‫‪f (x) = x sin(x) − 1‬‬
‫ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 2‬ﺭﺍ ﺑﺎ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻭ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ‪ | f (c) |< 10−5‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﻣﺮﺍﺣﻞ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‪ ،‬ﺗﺎ ﺭﺳﻴﺪﻥ ﺑﻪ‬
‫ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺍﺩﻩ ﺷﺪﻩ‪ ،‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ‪ ،‬ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ﺑﻪ ﻫﻤﺮﺍﻩ ﺑﺎﺯﻩﻫﺎﻱ ﺣﺎﺻﻞ ﺩﺭ ﻫﻔﺖ‬
‫ﺗﻜﺮﺍﺭ ﺍﻭﻝ ﺍﺯ ﺭﻭﺵ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪k‬‬ ‫‪a‬‬ ‫‪b‬‬ ‫‪c‬‬ ‫)‪f (c‬‬
‫‪0 0.000000 2.000000 1.000000 -0.158529‬‬
‫‪1 1.000000 2.000000 1.500000 +0.496242‬‬
‫‪2 1.000000 1.500000 1.250000 +0.186231‬‬ ‫‪a=0‬‬
‫•‬
‫‪3 1.000000 1.250000 1.125000 +0.015051‬‬ ‫‪b=2‬‬
‫‪4 1.000000 1.125000 1.062500 -0.071827‬‬
‫‪5 1.062500 1.125000 1.093750 -0.028362‬‬
‫‪6 1.093750 1.125000 1.109375 -0.006643‬‬
‫‪7 1.109375 1.125000 1.117188 +0.004208‬‬
‫‪8 1.109375 1.117188 1.113281 -0.001216‬‬
‫‪9 1.113281 1.117188 1.115234 +0.001496‬‬
‫[‪0‬‬ ‫•‬ ‫‪]2‬‬
‫‪10 1.113281 1.115234 1.114258 +0.000140‬‬
‫• [‪1‬‬ ‫‪]2‬‬
‫‪11 1.113281 1.114258 1.113770 -0.000538‬‬
‫• [‪1‬‬ ‫‪] 1.5‬‬
‫‪12 1.113770 1.114258 1.114014 -0.000199‬‬
‫‪13 1.114014 1.114258 1.114136 -0.000030‬‬ ‫‪1 [ • ] 1.25‬‬
‫‪14 1.114136 1.114258 1.114197 +0.000055‬‬ ‫‪1 [ •] 1.125‬‬
‫‪15 1.114136 1.114197 1.114166 +0.000013‬‬ ‫‪1.0625 [ •] 1.125‬‬
‫‪16 1.114136 1.114166 1.114151 -0.000009‬‬ ‫‪1.09375 []• 1.125‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺎ ‪ ۱۶‬ﺗﻜﺮﺍﺭ‪ c = 1.114151 ،‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﻣﻌﺮﻓﻲ ﻣﻲﻛﻨﺪ‪ .‬ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﺍﮔﺮ ﻣﻌﻴﺎﺭ‬
‫ﺗﻮﻗﻒ ﺑﻪ ﺻﻮﺭﺕ ‪ | ck − ck−1 |< 10−3‬ﺑﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎﻳﺪ ﺩﺭ ﺗﻜﺮﺍﺭ ‪۱۱‬ﺍﻡ ﻣﺘﻮﻗﻒ ﻣﻲﺷﺪﻳﻢ‪.‬‬

‫■ ∗ ﺩﺭﺧﺼﻮﺹ ﻳﺎﻓﺘﻦ ﺑﺎﺯﻩ ﺍﻭﻟﻴﻪ ﺩﺭ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‬

‫ﺩﺭ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﺍﺑﺘﺪﺍ ﺑﺎﺯﻩﺍﻱ ﻣﺎﻧﻨﺪ ]‪ [a, b‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻭﺭﻭﺩﻱ ﺑﻪ ﺭﻭﺵ ﺩﺍﺩﻩ ﺷﻮﺩ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪.f (a)f (b) < 0‬‬
‫ﺳﭙﺲ‪ ،‬ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﻣﻲﺗﻮﺍﻧﺪ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺭﻳﺸﻪ ﺭﺍ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﻴﺎﺑﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻭﻗﺘﻲ ﺗﺎﺑﻌﻲ ﺩﺍﺭﺍﻱ ﭼﻨﺪﻳﻦ‬
‫ﺭﻳﺸﻪ ﺍﺳﺖ ﻭ ﻣﺎ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﻤﺎﻣﻲ ﺭﻳﺸﻪﻫﺎ ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎﻳﺪ ﺩﺭ ﺍﺑﺘﺪﺍ ﺑﻪ ﺗﻌﺪﺍﺩ ﺭﻳﺸﻪﻫﺎ ﺑﺎﺯﻩ ﺑﻴﺎﺑﻴﻢ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺩﺭ ﻫﺮ ﺑﺎﺯﻩ‬
‫ﺩﻗﻴﻘﺎ ﻳﻚ ﺭﻳﺸﻪ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺳﻮﺍﻝ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﺎﺯﻩ ﺑﺎﺯﻩﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺭﺍ ﭼﮕﻮﻧﻪ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺁﻳﺎ ﺭﻭﺷﻲ ﻛﺎﻣﻞ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ‬
‫ﺗﺎﺑﻊ ﺩﺍﺩﻩ ﺷﻮﺩ ﻭ ﭼﻨﻴﻦ ﺑﺎﺯﻩ ﻳﺎ ﺑﺎﺯﻩﻫﺎﻳﻲ ﺑﺮﮔﺮﺩﺍﻧﺪﻩ ﺷﻮﺩ؟ ﺩﺭ ﭘﺎﺳﺦ ﺑﺎﻳﺪ ﮔﻔﺖ ﻛﻪ ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﺭﻭﺷﻲ ﺗﺤﻠﻴﻠﻲ ﺑﺮﺍﻱ ﭼﻨﻴﻦ ﻛﺎﺭﻱ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۳۶‬‬

‫ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ ،‬ﻣﮕﺮ ﺩﺭ ﺗﻮﺍﺑﻊ ﺧﺎﺹ‪ ،‬ﻣﺎﻧﻨﺪ ﺧﺎﻧﻮﺍﺩﻩ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‪ .‬ﻟﺬﺍ ﺩﺭ ﻋﻤﻞ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺑﺎﺯﻩ ﻳﺎ ﺑﺎﺯﻩﻫﺎﻱ ﺣﺎﻭﻱ ﺟﻮﺍﺏ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﻓﻮﻕ‪ ،‬ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﺗﺼﻮﻳﺮﻱ ﻭ ﺭﻭﺵﻫﺎﻱ ﺳﻌﻲ ﻭ ﺧﻄﺎ ﺭﺟﻮﻉ ﻣﻲﻧﻤﺎﻳﻴﻢ‪.‬‬

‫ﺩﺭ ﺭﻭﺵﻫﺎﻱ ﺗﺼﻮﻳﺮﻱ‪ ،‬ﺍﺑﺘﺪﺍ ﺗﺎﺑﻊ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻛﺎﻣﭙﻴﻮﺗﺮ ﺩﺭ ﺩﺍﻣﻨﻪ ﻣﻮﺭﺩ ﻧﻈﺮ ﺭﺳﻢ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺑﻪ ﺗﺤﻠﻴﻞ ﺷﻜﻞ ﻭ‬
‫ﻳﺎﻓﺘﻦ ﺑﺎﺯﻩ ﻭ ﻳﺎ ﺑﺎﺯﻩﻫﺎ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺣﻀﻮﺭ ﺷﺨﺺ ﺗﺤﻠﻴﻞﮔﺮ ﻻﺯﻡ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺭﻭﺷﻲ ﺍﺗﻮﻣﺎﺗﻴﻚ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ‬
‫ﺭﻭﺵﻫﺎﻱ ﺳﻌﻲ ﻭ ﺧﻄﺎ‪ ،‬ﻫﺪﻑ ﺍﺭﺍﻳﻪ ﺍﻟﮕﻮﺭﻳﺘﻤﻲ ﺍﺳﺖ ﻛﻪ ﺗﺎﺑﻊ ﺭﺍ ﻣﻲﮔﻴﺮﺩ ﻭ ﻳﻚ ﺑﺎﺯﻩ ﺣﺎﻭﻱ ﺟﻮﺍﺏ ﺍﺭﺍﻳﻪ ﻣﻲﺩﻫﺪ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫ﺗﺎﺑﻊ ﺩﺭ ﺩﻭ ﺳﺮ ﺑﺎﺯﻩ ﺗﻐﻴﻴﺮ ﻋﻼﻣﺖ ﻣﻲﺩﻫﺪ‪ .‬ﺍﺯ ﺟﻤﻠﻪ ﻳﻚ ﻧﻤﻮﻧﻪ ﺍﺯ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺭﻭﺵ ﺟﺴﺘﺠﻮﻱ ﺟﻠﻮﺭﻭﻧﺪﻩ ‪ ۱۹‬ﺍﺷﺎﺭﻩ‬
‫ﻛﺮﺩ‪ ،‬ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺭ ﺭﻭﺵ ﺟﺴﺘﺠﻮﻱ ﺟﻠﻮﺭﻭﻧﺪﻩ‪ ،‬ﺍﺑﺘﺪﺍ ﺩﺍﻣﻨﻪﻱ ]‪ [l, u‬ﻛﻪ ﻗﺮﺍﺭ ﺍﺳﺖ ﺩﺭ ﺁﻥ ﺭﻳﺸﻪﻫﺎ ﻳﺎﻓﺘﻪ ﺷﻮﻧﺪ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ ﻭ ﺳﭙﺲ‬
‫ﺍﻳﻦ ﺩﺍﻣﻨﻪ ﺭﺍ ﺑﻪ ‪ m‬ﺯﻳﺮﺑﺎﺯﻩ ﺯﻳﺮ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‬
‫] ‪[x0 , x1 ], [x1 , x2 ], . . . , [xm−1 , xm‬‬ ‫)‪(۶.۵‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫‪u−l‬‬
‫=‪∆x :‬‬ ‫‪, xi := l + i∆x, i = 0, 1, . . . , m.‬‬
‫‪m‬‬
‫ﺣﺎﻝ ﻭﺟﻮﺩ ﺭﻳﺸﻪ ﺩﺭ ﺯﻳﺮﺑﺎﺯﻩ ﺍﻭﻝ ﺑﺮﺭﺳﻲ ﻣﻲﻛﻨﻴﻢ ﻭ ﺑﻪ ﺟﻠﻮ ﻣﻲﺭﻭﻳﻢ ﺗﺎ ﺗﻤﺎﻣﻲ ﺯﻳﺮﺑﺎﺯﻩﻫﺎ ﻛﻨﺘﺮﻝ ﺷﻮﻧﺪ‪ .‬ﺑﺮﺍﻱ ﺑﺮﺭﺳﻲ ﻭﺟﻮﺩ ﺭﻳﺸﻪ‬
‫ﺩﺭ ﻳﻚ ﺯﻳﺮﺑﺎﺯﻩ ] ‪ [xi−1 , xi‬ﺍﺯ ﻋﻼﻣﺖ ﺗﺎﺑﻊ ﺩﺭ ﺩﻭ ﺳﺮ ﺯﻳﺮﺑﺎﺯﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﮔﺮ ‪f (xi−1 )f (xi ) < 0‬‬
‫ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [xi−1 , xi‬ﺭﻳﺸﻪ ﺩﺍﺭﺩ‪ .‬ﺍﻟﺒﺘﻪ ﻧﻤﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺗﺎﺑﻊ ﺩﺭ ﺍﻳﻦ ﺯﻳﺮﺑﺎﺯﻩ ﺗﻨﻬﺎ ﻳﻚ ﺭﻳﺸﻪ ﺩﺍﺭﺩ ﻭ ﻣﻤﻜﻦ ﺍﺳﺖ‬
‫ﺑﻴﺶ ﺍﺯ ﻳﻚ ﺭﻳﺸﻪ ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺗﻨﻬﺎ ﻳﻜﻲ ﺍﺯ ﺭﻳﺸﻪﻫﺎ ﺭﺍ ﻣﻲﻳﺎﺑﺪ ﻭ ﺑﻘﻴﻪ ﺭﻳﺸﻪﻫﺎ‬
‫ﺍﺯ ﺩﺳﺖ ﻣﻲﺭﻭﻧﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﻋﻼﻣﺖ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﺩﻭ ﺳﺮ ﺯﻳﺮﺑﺎﺯﻩ ﻳﻜﺴﺎﻥ ﺑﺎﺷﺪ‪ ،‬ﻭﻟﻲ ﻳﻚ ﻳﺎ ﭼﻨﺪ ﺭﻳﺸﻪ ﺩﺭ ﺑﺎﺯﻩ‬
‫ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺭﻭﺵ ﺟﺴﺘﺠﻮﻱ ﭘﻴﺶﺭﻭﻧﺪﻩ ﺍﻳﻦ ﺭﻳﺸﻪﻫﺎ ﺭﺍ ﻧﻴﺰ ﺍﺯ ﺩﺳﺖ ﻣﻲﺩﻫﺪ‪ .‬ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ‬
‫ﻛﻪ ﻫﺮﭼﻪ ﻣﻘﺪﺍﺭ ‪ m‬ﺑﺰﺭﮒﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﺣﺘﻤﺎﻝ ﺍﺯ ﺩﺳﺖ ﺩﺍﺩﻥ ﺭﻳﺸﻪﻫﺎ ﻛﻢﺗﺮ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺎﺯﻩﻫﺎﻳﻲ ﻛﻮﭼﻚﺗﺮ )ﺩﻗﻴﻖﺗﺮ(‬
‫ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺍﻣﺎ ﺑﺰﺭﮔﻲ ‪ m‬ﺑﺎﻋﺚ ﺍﻓﺰﺍﻳﺶ ﺣﺠﻢ ﻣﺤﺎﺳﺒﺎﺕ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ‬ ‫‪۳.۵‬‬

‫ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ‪ ۲۰‬ﻣﺸﺎﺑﻪ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺩﺭ ﻫﺮ ﺗﻜﺮﺍﺭ ﺑﺎﺯﻩ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ﺑﺎﺯﻩﺍﻱ ﻛﻮﭼﻚﺗﺮ ﻭ ﺷﺎﻣﻞ‬
‫ﺭﻳﺸﻪ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‪ ،‬ﺗﻨﻬﺎ ﺍﺯ ﻋﻼﻣﺖ )‪ f (a‬ﻭ )‪ f (b‬ﺑﺮﺍﻱ ﺟﺴﺘﺠﻮﻱ ﺭﻳﺸﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﻣﻘﺎﺩﻳﺮ‬
‫)‪ f (a‬ﻭ )‪ f (b‬ﻧﻴﺰ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻃﻼﻋﺎﺗﻲ ﻓﺮﺍﻫﻢ ﻛﻨﺪ ﻛﻪ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﻧﺤﻮ ﻣﻨﺎﺳﺐﺗﺮﻱ ﺑﻪ ﺩﻭ ﺯﻳﺮﺑﺎﺯﻩ ﺗﻘﺴﻴﻢ ﺷﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪،‬‬
‫ﺍﮔﺮ ﺍﻧﺪﺍﺯﻩ )‪ f (a‬ﺍﺯ ﺍﻧﺪﺍﺯﻩ )‪ f (b‬ﻛﻮﭼﻚﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺭﻳﺸﻪ ﺑﻪ ‪ a‬ﻧﺰﺩﻳﻚﺗﺮ ﺑﺎﺷﺪ‪ .‬ﺑﺮ ﺍﻳﻦ ﺍﺳﺎﺱ ﺩﺭ ﺭﻭﺵ‬
‫ﻧﺎﺑﺠﺎﻳﻲ ﺑﺎﺯﻩ ]‪ [a, b‬ﻣﺘﻨﺎﺳﺐ ﺑﺎ ﻧﺴﺒﺖ )‪ f (a)/f (b‬ﺑﻪ ﺩﻭ ﻗﺴﻤﺖ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺯﺑﺎﻥ ﺭﻳﺎﺿﻲ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﺯ ﻧﻘﻄﻪ ‪ c‬ﺑﻪ‬
‫ﺩﻭ ﻗﺴﻤﺖ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ c‬ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﺧﻂ ﻭﺍﺻﻞ ﺑﻴﻦ ﻧﻘﺎﻁ ))‪ (a, f (a‬ﻭ ))‪ (b, f (b‬ﺑﺎ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫‪19 Incremental‬‬ ‫‪search‬‬ ‫‪20 Method‬‬ ‫)‪of False position (“Regula falsi” in Latin‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳۷‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫)‪f (a‬‬
‫•‬

‫‪b‬‬ ‫‪x‬‬
‫‪a‬‬ ‫‪+‬‬ ‫• •‬

‫‪a+ 2‬‬

‫‪c‬‬
‫=‬
‫‪b‬‬
‫)‪• f (b‬‬

‫‪a‬‬
‫‪−‬‬
‫)‪f ( (b‬‬
‫‪a) − f‬‬
‫‪f‬‬

‫‪( b (a‬‬
‫) ‪−‬‬
‫)‪a‬‬
‫ﻓﺮﻣﻮﻝ ﺗﺤﻠﻴﻠﻲ ﺧﻂ ﻭﺍﺻﻞ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫)‪f (b) − f (a‬‬
‫= )‪f (x) − f (a‬‬ ‫)‪(x − a‬‬
‫‪b−a‬‬
‫ﺍﮔﺮ ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﺧﻂ ﻓﻮﻕ ﺑﺎ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺭﺍ ‪ c‬ﻧﺎﻣﮕﺬﺍﺭﻱ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺧﻂ ﺍﺯ ﻧﻘﻄﻪ )‪ (c, 0‬ﻋﺒﻮﺭ ﻣﻲﻛﻨﺪ‪ .‬ﻟﺬﺍ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫)‪f (b) − f (a‬‬
‫= )‪0 − f (a‬‬ ‫)‪(c − a‬‬
‫‪b−a‬‬
‫ﻭ ﺑﺎ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺟﺒﺮﻱ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪b−a‬‬
‫‪c=a−‬‬ ‫‪f (a).‬‬ ‫)‪(۷.۵‬‬
‫)‪f (b) − f (a‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ﺩﻗﻴﻘﺎ ﻣﺸﺎﺑﻪ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺍﺳﺖ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﺩﺭ ﻗﺪﻡ ‪ ،۱‬ﻓﺮﻣﻮﻝ )‪ (۷.۵‬ﺑﺮﺍﻱ‬
‫ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﻣﻘﺪﺍﺭ ‪ c‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۵−۵‬ﺭﻭﺵ ﻧﺎﺑﺠﺎﯾﯽ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺭﯾﺸﻪ ﺗﻘﺮﯾﺒﯽ ﯾ‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻫﻤﺎﻥ ﺗﺎﺑﻊ ﻭ ﺷﺮﻁ ﺗﻮﻗﻒ ﻣﺜﺎﻝ ‪ ۴-۵‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺗﻲ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺎ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ﺭﻳﺸﻪ ﺗﺎﺑﻊ‬
‫‪ f (x) = x sin(x) − 1‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 2‬ﺑﻴﺎﺑﻴﻢ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﻪ ﺻﻮﺭﺕ ‪ | f (c) |< 10−5‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﻧﺘﺎﻳﺞ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪k‬‬ ‫‪a‬‬ ‫‪b‬‬ ‫‪c‬‬ ‫)‪f (c‬‬
‫‪0‬‬ ‫‪0.000000‬‬ ‫‪2.000000‬‬ ‫‪1.099750‬‬ ‫‪-0.0200190‬‬
‫‪1‬‬ ‫‪1.099750‬‬ ‫‪2.000000‬‬ ‫‪1.121241‬‬ ‫‪+0.009835‬‬
‫‪2‬‬ ‫‪1.099750‬‬ ‫‪1.121241‬‬ ‫‪1.114161‬‬ ‫‪+0.000006‬‬

‫ﺍﻣﺎ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺮﺍﻱ ﺭﺳﻴﺪﻥ ﺑﻪ ﺩﻗﺖ ‪ | f (c) |< 10−5‬ﺑﻪ ‪ ۳۴‬ﺗﻜﺮﺍﺭ ﻧﻴﺎﺯ ﺩﺍﺭﺩ‪.‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻓﺖ‪ ،‬ﺩﺭ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ﺳﺮﻳﻊﺗﺮ ﺍﺯ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﺪ‪.‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﺁﻳﺎ ﻫﻤﻮﺍﺭﻩ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ﺳﺮﻳﻊﺗﺮ ﻋﻤﻞ ﻣﻲﻛﻨﺪ؟ ﺑﺎﺳﺦ ﺳﻮﺍﻝ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ‬
‫ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۳۸‬‬

‫ﻣﻘﺎﯾﺴﻪ ﺑﯿﻦ ﺭﻭﺵ ﺩﻭﺑﺨﺸ ﻭ ﻧﺎﺑﺠﺎﯾﯽ(‬ ‫ﻣﺜﺎﻝ ‪)۶−۵‬ﯾ‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f (x) = x10 − 1‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 1.3‬ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﺩﻭﺑﺨﺸﻲ ﻭ ﻧﺎﺑﺠﺎﻳﻲ ﺑﻴﺎﺑﻴﻢ‪ .‬ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪ | f (c) |< 10−2‬ﺩﺭﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻧﺘﺎﻳﺞ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ ﺑﺎ ‪ ۷‬ﺗﻜﺮﺍﺭ ﻭ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ‬
‫ﺑﺎ ‪ ۳۴‬ﺗﻜﺮﺍﺭ ﺭﻳﺸﻪ ﺗﻘﺮﻳﺒﻲ ﺭﺍ )ﻣﻄﺎﺑﻖ ﺑﺎ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺩﺍﺩﻩ ﺷﺪﻩ( ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﺪ‪.‬‬

‫ﺗﻜﺮﺍﺭﻫﺎﻱ ﺭﻭﺵ ﺩﻭﺑﺨﺸﻲ‬ ‫ﺗﻜﺮﺍﺭﻫﺎﻱ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ‬

‫‪k‬‬ ‫‪a‬‬ ‫‪b‬‬ ‫‪c‬‬ ‫)‪f (c‬‬ ‫‪k‬‬ ‫‪a‬‬ ‫‪b‬‬ ‫‪c‬‬ ‫)‪f (c‬‬
‫‪0 0.000000 1.300000 0.650000 -0.986537‬‬ ‫‪0 0.000000 1.300000 0.094300 -1.000000‬‬
‫‪1 0.650000 1.300000 0.975000 -0.223670‬‬ ‫‪1 0.094300 1.300000 0.181759 -1.000000‬‬
‫‪2 0.975000 1.300000 1.137500 2.626720‬‬ ‫‪2 0.181759 1.300000 0.262874 -0.999998‬‬
‫‪3 0.975000 1.137500 1.056250 0.728491‬‬ ‫‪3 0.262874 1.300000 0.338105 -0.999980‬‬
‫‪4 0.975000 1.056250 1.015625 0.167707‬‬ ‫‪4 0.338105 1.300000 0.407878 -0.999873‬‬
‫‪5 0.975000 1.015625 0.995313 -0.045898‬‬ ‫‪5 0.407878 1.300000 0.472583 -0.999444‬‬
‫‪6 0.995313 1.015625 1.005469 0.056053‬‬ ‫‪6 0.472583 1.300000 0.532572 -0.998164‬‬
‫‪7 0.995313 1.005469 1.000391 0.003913‬‬ ‫‪7 0.532572 1.300000 0.588145 -0.995047‬‬
‫‪..‬‬ ‫‪8 0.588145 1.300000 0.639544 -0.988553‬‬
‫‪.‬‬
‫‪9 0.639544 1.300000 0.686943 -0.976600‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪28 0.994184 1.300000 0.995534 -0.043777‬‬
‫‪29 0.995534 1.300000 0.996573 -0.033751‬‬
‫‪30 0.996573 1.300000 0.997371 -0.025978‬‬
‫‪31 0.997371 1.300000 0.997985 -0.019968‬‬
‫‪32 0.997985 1.300000 0.998456 -0.015334‬‬
‫‪33 0.998456 1.300000 0.998817 -0.011766‬‬
‫‪34 0.998817 1.300000 0.999094 -0.009023‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻭﺵ ﻧﺎﺑﺠﺎﻳﻲ ﺑﺎ ﺗﻜﺮﺍﺭﻫﺎﻱ ﺑﺒﻴﺸﺘﺮﻱ ﺭﻳﺸﻪ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﺪ‪ .‬ﻋﻠﺖ ﺍﻳﻦ ﺍﺗﻔﺎﻕ ﺭﻓﺘﺎﺭ ﺗﺎﺑﻊ ‪ f‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺷﻜﻞ ﺍﻳﻦ ﺗﺎﺑﻊ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۳۹‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺭﻭﺵﻫﺎﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬ ‫‪۴.۵‬‬

‫ﺭﻭﺵﻫﺎﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﺒﺘﻨﻲ ﺑﺮ ﻣﻔﻬﻮﻡ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ ۲۱‬ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۵‬‬

‫ﻧﻘﻄﻪ ‪ p‬ﺭﺍ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺗﺎﺑﻊ )‪ g(x‬ﮔﻮﻳﻴﻢ‪ ،‬ﻫﺮﮔﺎﻩ‬
‫)‪p = g(p‬‬

‫ﺍﺯ ﻧﻈﺮ ﻫﻨﺪﺳﻲ‪ ،‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﺤﻞ ﺗﺎﺑﻊ ﻣﺤﻞ ﺑﺮﺧﻮﺭﺩ ﺗﺎﺑﻊ ﺑﺎ ﻧﻴﻤﺴﺎﺯ ﻧﺎﺣﻴﻪﻫﺎﻱ ﺍﻭﻝ ﻭ ﺳﻮﻡ ﺩﺭ ﻣﺨﺘﺼﺎﺕ ﺩﻛﺎﺭﺗﻲ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺑﺮﺍﻱ ﻳﻚ ﺗﺎﺑﻊ ﺩﻟﺨﻮﺍﻩ ‪ ،g‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺻﻼ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ ﻭ ﻳﺎ ﻣﻤﻜﻦ ﺍﺳﺖ ﻳﻚ ﻳﺎ ﭼﻨﺪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۷−۵‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﭼﻨﺪ ﺗﺎﺑﻊ(‬

‫◀ ﻧﻘﻄﻪ ‪ p = 0‬ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪ g1 (x) = x2‬ﺍﺳﺖ‪ .‬ﻫﻤﻴﻦﻃﻮﺭ ﻧﻘﻄﻪ ‪ p = 1‬ﻧﻴﺰ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ‬
‫ﺗﺎﺑﻊ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﻳﮕﺮﻱ ﻧﺪﺍﺭﺩ‪.‬‬
‫◀ﺩﺭ ﺗﺎﺑﻊ ‪ g2 (x) = x2 − 2‬ﻧﻘﺎﻁ ‪ p = −1‬ﻭ ‪ p = 2‬ﺩﻭ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬
‫◀ﺗﺎﺑﻊ ‪ g3 (x) = x3‬ﺩﺍﺭﺍﻱ ﺳﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ −1, 0, +1‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫‪21 Fixed‬‬ ‫‪Point‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴۰‬‬

‫‪g1 (x) = x2‬‬ ‫‪g2 (x) = x2 − 2‬‬ ‫‪g3 (x) = x3‬‬

‫‪2‬‬ ‫•‬

‫‪1‬‬ ‫•‬ ‫‪1‬‬ ‫•‬

‫‪x‬‬ ‫‪x‬‬ ‫‪x‬‬
‫•‬ ‫•‬
‫‪1‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪−1‬‬ ‫‪1‬‬
‫‪•−1‬‬ ‫‪•−1‬‬

‫‪ g4 (x) = x2x+3‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ‬

‫‪−2‬‬
‫◀ﺗﺎﺑﻊ‬
‫‪2‬‬

‫‪x‬‬ ‫ﺛﺎﺑﺖ ﺍﻳﻦ ﺗﺎﺑﻊ‪ ،‬ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ ‪ g(x) = x‬ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪−5 −4 −3 −2 −1‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪x2 − 2‬‬
‫•‬ ‫⇒= ‪= x =⇒ x2 − 2 = 2x2 + 3x‬‬
‫‪2x + 3‬‬
‫‪2‬‬
‫‪−2‬‬
‫‪g4 (x) = x2x+3‬‬ ‫‪‬‬
‫•‬ ‫‪−2‬‬ ‫‪x = −2,‬‬
‫⇒= ‪x2 + 3x + 2 = 0‬‬
‫‪x = −1.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺗﺎﺑﻊ ‪ g‬ﺩﺍﺭﺍﻱ ﺩﻭ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = −2‬ﻭ ‪p = −1‬‬

‫‪−4‬‬
‫ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪3‬‬

‫‪2‬‬ ‫◀ﺗﺎﺑﻊ )‪ g5 (x) = x − ln(x‬ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪p = 1‬‬

‫•‬
‫ﺍﺳﺖ‪ .‬ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 1‬ﻣﻨﺤﺼﺮ ﺑﻪ‬
‫‪1‬‬ ‫)‪g5 (x) = x − ln(x‬‬
‫‪x‬‬
‫ﻓﺮﺩ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺗﺎﺑﻊ ﺑﻪ ﺟﺰ ‪ ،p = 1‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﻳﮕﺮﻱ ﻧﺪﺍﺭﺩ‪.‬‬
‫‪−2 −1‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬ ‫‪6‬‬

‫)‪g6 (x) = cos(x‬‬ ‫◀ﺗﺎﺑﻊ )‪ g6 (x) = cos(x‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺑﺎ ﺭﺳﻢ ﻧﻤﻮﺩﺍﺭ‬
‫ﺗﺎﺑﻊ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻓﻬﻤﻴﺪ ﻛﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻨﺤﺼﺮ ﺑﻪ ‪x‬‬
‫‪π‬‬ ‫ﻓﺮﺩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺑﺮﺧﻼﻑ ﺗﻮﺍﺑﻊ ﻗﺒﻞ‪ ،‬ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ‬
‫‪2‬‬
‫ﻧﻤﻲﺗﻮﺍﻥ ﺣﺪﺱ ﺯﺩ ﻭ ﻳﺎ ﺑﻪ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺁﻥ ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﺮﺩ‪.‬‬

‫ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻳﻚ ﺗﺎﺑﻊ ﻫﻴﭻ ﻧﻘﻄﻪ ﺛﺎﺑﺘﻲ ﻧﺪﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻣﻤﻜﻦ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﻳﻚ‪ ،‬ﺩﻭ ﻭ ﻳﺎ‬
‫ﭼﻨﺪ ﻭ ﻳﺎ ﺣﺘﻲ ﺑﻲﻧﻬﺎﻳﺖ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﺑﺰﺍﺭ ﺟﺎﻣﻌﻲ ﺑﺮﺍﻱ ﺗﺸﺨﻴﺺ ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺩﺭ‬
‫ﺍﺩﺍﻣﻪ ﺩﻭ ﻗﻀﻴﻪ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﻣﻮﺍﺭﺩ ﺧﺎﺹ ﻣﻲﺗﻮﺍﻥ ﻭﺟﻮﺩ ﻭ ﻳﻜﺘﺎﻳﻲ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ ﺗﻀﻤﻴﻦ ﻧﻤﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۴۱‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺑﺎﺯﻩ‬ ‫ﺷﺮﻁ ﮐﺎﻓ ﺑﺮﺍﯼ ﻭﺟﻮﺩ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﯾ‬ ‫ﻗﻀﯿﻪ ‪۴−۵‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ g(x) ∈ [a, b] :‬ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ‪g‬‬
‫ﺩﺍﺭﺍﻱ ﺣﺪﺍﻗﻞ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ‪ g‬ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﻪ ﺗﻮﻱ ﺧﻮﺩ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺒﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺗﺎﺑﻊ ﺣﺪﺍﻗﻞ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺩﺍﺭﺩ‪.‬‬

‫ﺑﺎﺯﻩ‬ ‫ﺷﺮﻁ ﮐﺎﻓ ﺑﺮﺍﯼ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺑﻮﺩﻥ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﯾ‬ ‫ﻗﻀﯿﻪ ‪۵−۵‬‬

‫ﺍﮔﺮ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﻭ ﺩﺍﺭﺍﻱ ﻣﺸﺘﻖ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪|g (x)| < 1 :‬‬
‫‪′‬‬

‫ﺁﻥﮔﺎﻩ‪ :‬ﺗﺎﺑﻊ ‪ g‬ﺣﺪﺍﻛﺜﺮ ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ‪ g‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬
‫ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ‪.‬‬

‫ﺑﻌﺪ ﺍﺯ ﻣﻘﺪﻣﺎﺕ ﻓﻮﻕ‪ ،‬ﺳﻮﺍﻝ ﺍﺻﻠﻲ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﭼﮕﻮﻧﻪ ﻣﻲﺗﻮﺍﻥ ﻧﻘﻄﻪ ﻳﺎ ﻧﻘﺎﻁ ﺛﺎﺑﺖ ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ؟ ﺩﺭ ﻣﻮﺍﺭﺩ‬
‫ﻧﺎﺩﺭ ﻣﻲﺗﻮﺍﻥ ﻧﻘﻄﻪ ﻭ ﻳﺎ ﻧﻘﺎﻁ ﺛﺎﺑﺖ ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﺑﻪ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﻏﺎﻟﺐ ﻣﻮﺍﺭﺩ‪ ،‬ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺍﺯ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ‬
‫ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﻋﺎﺟﺰ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﻳﺎ ﻧﻘﺎﻁ ﺛﺎﺑﺖ ﻳﻚ ﺗﺎﺑﻊ ﺑﻪ ﺳﺮﺍﻍ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻣﻲﺭﻭﻳﻢ‪ .‬ﻳﻚ ﺭﻭﺵ‬
‫ﻋﺪﺩﻱ ﺳﺎﺩﻩ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ ،۲۲‬ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺁﻥ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪ .‬ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ‬
‫ﺍﺧﺘﺼﺎﺭ ﺑﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻭ ﻳﺎ ﺭﻭﺵ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﻛﺘﺎﺏ ﺍﺯ ﻭﺍﮊﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫�‬

‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺭﻭﺷﻲ ﺗﻜﺮﺍﺭﻱ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ‪ g‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺍﻳﺠﺎﺩ ﻣﻲﻛﻨﺪ‬
‫ﻛﻪ‪ ،‬ﺩﺭ ﺻﻮﺭﺕ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﺑﻪ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ‪ ،‬ﺍﺑﺘﺪﺍ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0‬ﺑﺮﺍﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺯﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺷﺮﻭﻉ ﺩﻧﺒﺎﻟﻪ ﺩﺭ ﻧﻈﺮ‬
‫ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺳﭙﺲ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﻮﻟﻔﻪ ﺑﻌﺪﻱ ﺩﻧﺒﺎﻟﻪ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‬
‫) ‪ .x1 = g(x0‬ﻫﻤﻴﻦﻃﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ ) ‪ .x2 = g(x1‬ﺑﺎ ﺍﺩﺍﻣﻪ ﺭﻭﻧﺪ ﻓﻮﻕ‪ ،‬ﺩﻧﺒﺎﻟﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ‪ {xk }kk=0‬ﺳﺎﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺷﻤﺎﻱ ﺍﻳﺠﺎﺩ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫) ‪g(x0‬‬ ‫) ‪g(x1‬‬ ‫) ‪g(x2‬‬ ‫) ‪g(x3‬‬ ‫) ‪g(xk‬‬

‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫‪...‬‬ ‫‪xk‬‬ ‫‪xk+1‬‬ ‫‪...‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ‪ p‬ﻫﻤﮕﺮﺍ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺗﺎﺑﻊ ‪ g‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪p = lim xk+1‬‬
‫∞→‪k‬‬

‫) ‪= lim g(xk‬‬
‫∞→‪k‬‬

‫‪= g( lim xk ),‬‬ ‫)ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﭘﻴﻮﺳﺘﮕﻲ ﺗﺎﺑﻊ ‪(g‬‬

‫∞→‪k‬‬

‫‪= g(p).‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺧﻼﺻﻪ )‪ .p = g(p‬ﻳﻌﻨﻲ ﺣﺪ ﺩﻧﺒﺎﻟﻪ } ‪ {xk‬ﻫﻤﺎﻥ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ g‬ﺍﺳﺖ‪ .‬ﻣﻄﺎﻟﺐ ﻓﻮﻕ ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪22 Simple‬‬ ‫‪fixed-point iteration method‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴۲‬‬

‫ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﻮﺩﻥ ﺣﺪ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻗﻀﯿﻪ ‪۶−۵‬‬
‫∞‬
‫ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ ﺷﺮﻭﻉ ﺍﻭﻟﻴﻪ ‪ x0‬ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ‬
‫‪{xk }k=0‬‬ ‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ g‬ﻳﻚ ﺗﺎﺑﻊ ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ ﻭ‬
‫ﺑﻪ ﻧﻘﻄﻪﺍﻱ ﻣﺎﻧﻨﺪ ‪ p‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ‪ p‬ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪ g‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﻭ ﺗﺎﺑﻊ ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۸−۵‬ﻣﺜﺎﻟ ﺍﺯ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ(‬

‫◀ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻣﺜﺎﻝ ‪ ۷-۵‬ﺩﻳﺪﻳﻢ‪ ،‬ﺗﺎﺑﻊ )‪ g(x) := x − ln(x‬ﺩﺍﺭﺍﻱ‬
‫ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ‪ p = 1‬ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻨﺠﺎ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻭﺵ‬
‫‪k‬‬ ‫‪xk‬‬
‫ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﺮ ﺭﻭﻱ ﺍﻳﻦ ﺗﺎﺑﻊ ﺍﻋﻤﺎﻝ ﻛﻨﻴﻢ‪ .‬ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺭﺍ ﺑﺮﺍﺑﺮ ‪x0 = 5‬‬
‫‪0‬‬ ‫‪5.000000000000000‬‬
‫‪1‬‬ ‫‪3.390562087565899‬‬
‫ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﺎ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫‪2‬‬ ‫‪2.169566372376528‬‬ ‫ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪:‬‬
‫‪3‬‬ ‫‪1.395039053190428‬‬ ‫‪x1 =x0 − ln(x0 ) = 5 − ln(5) = 3.390562087565899,‬‬
‫‪4‬‬ ‫‪1.062116643190369‬‬ ‫‪x2 =x1 − ln(x1 ) = 2.169566372376528,‬‬
‫‪5‬‬ ‫‪1.001852892889292‬‬
‫‪6‬‬ ‫‪1.000001714488514‬‬ ‫‪x3 =x2 − ln(x2 ) = 1.395039053190428,‬‬
‫‪7‬‬ ‫‪1.000000000001470‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫ﺑﺎ ﺍﺩﺍﻣﻪ ﻓﺮﺍﻳﻨﺪ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﻧﺒﺎﻟﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺗﻮﻟﻴﺪ ﻧﻤﻮﺩ‪ .‬ﻧﺘﺎﻳﺞ‬
‫‪1‬‬
‫ﺩﺭ ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥ ﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲ ﮔﺮﺩﺩ‪ ،‬ﺍﻳﻦ‬
‫ﺩﻧﺒﺎﻟﻪ ﺑﻪ ‪ 1‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﻫﻤﺎﻥ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﺍﺳﺖ‪.‬‬
‫◀ ﺩﺭ ﻣﺜﺎﻝ ‪ ۷-۵‬ﺩﻳﺪﻳﻢ ﻛﻪ ﺗﺎﺑﻊ )‪ g(x) := cos(x‬ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬
‫‪k‬‬ ‫‪xk‬‬ ‫| ‪|xk − xk−1‬‬
‫‪0‬‬ ‫‪0.000000‬‬ ‫–‬ ‫ﺩﺭ ﺑﺎﺯﻩ ] ‪ [0, π2‬ﺍﺳﺖ ﻛﻪ ﻧﺘﻮﺍﻧﺴﺘﻴﻢ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﺗﺤﻠﻴﻠﻲ ﺁﻥ ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺩﺭ‬
‫‪1‬‬ ‫‪1.000000‬‬ ‫‪1.0 e+00‬‬ ‫ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﺭﺍ ﺑﺎ ﻛﻤﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬
‫‪2‬‬ ‫‪0.540302‬‬ ‫‪4.6 e-01‬‬
‫‪3‬‬ ‫‪0.857553‬‬ ‫‪3.2 e-01‬‬
‫ﺑﻴﺎﺑﻴﻢ‪ .‬ﻧﻘﻄﻪ ﺷﺮﻭﻉ ‪ x0 = 0‬ﻭ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪4‬‬ ‫‪0.654290‬‬ ‫‪2.0 e-01‬‬ ‫‪|xk − xk−1 | < 5.0 × 10−2‬‬
‫‪5‬‬ ‫‪0.793480‬‬ ‫‪1.4 e-01‬‬
‫‪6‬‬ ‫‪0.701369‬‬ ‫‪9.2 e-02‬‬ ‫ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺭ ﺟﺪﻭﻝ ﻣﻘﺎﺑﻞ ﮔﺰﺍﺭﺵ ﺷﺪﻩ‬
‫‪7‬‬ ‫‪0.763960‬‬ ‫‪6.3 e-02‬‬ ‫ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺑﺎ ﻣﻌﻴﺎﺭ ﺗﻮﻗﻒ ﻓﻮﻕ‪ ،‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬
‫‪8‬‬ ‫‪0.722102‬‬ ‫‪4.2 e-02‬‬
‫ﺗﻘﺮﻳﺒﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪.0.722102‬‬

‫■ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬

‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺍﺭﺍﻱ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺟﺎﻟﺒﻲ ﺍﺳﺖ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﺗﺎﺑﻊ )‪ g(x‬ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺎﺷﺪ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ‬
‫ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0‬ﺍﻧﺠﺎﻡ ﺩﻫﻴﻢ‪ .‬ﺗﺎﺑﻊ )‪ g(x‬ﺭﺍ ﺩﺭ ﻳﻚ ﻣﺨﺘﺼﺎﺕ ﺩﻛﺎﺭﺗﻲ ﺭﺳﻢ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻣﻮﻗﻌﻴﺖ ‪ x0‬ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﻣﺸﺨﺺ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﻃﺒﻖ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﻣﻮﻟﻔﻪ ﺩﻭﻡ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﺻﻮﺭﺕ ) ‪ x1 = g(x0‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺑﺎ ﻛﻤﻚ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ‪ g‬ﻣﻲﺗﻮﺍﻥ‬
‫ﻣﻮﻗﻌﻴﺖ ‪ x1‬ﺩﺭ ﺭﻭﻱ ﻣﺤﻮﺭ ‪y‬ﻫﺎ ﺭﺍ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ ﻛﻪ ﻣﻮﻗﻌﻴﺖ ‪ x1‬ﺭﺍ ﺩﺭ ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺑﺪﺍﻧﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‬
‫ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺳﺎﺩﮔﻲ ﺍﺯ ﻧﻴﻤﺴﺎﺯ ﺍﻭﻝ ﻭ ﺳﻮﻡ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﺯ ﻧﻘﻄﻪ ‪ x1‬ﺩﺭ ﻣﺤﻮﺭ ‪y‬ﻫﺎ ﺧﻄﻲ ﺍﻓﻘﻲ ﺭﺳﻢ ﻣﻲﻛﻨﻴﻢ‬
‫ﺗﺎ ﻧﻴﻤﺴﺎﺯ ﺍﻭﻝ ﻭ ﺳﻮﻡ ﺭﺍ ﻗﻄﻊ ﻛﻨﺪ‪ .‬ﺳﭙﺲ ﺍﺯ ﻣﺤﻞ ﻗﻄﻊ ﺧﻄﻲ ﻋﻤﻮﺩﻱ ﺭﺳﻢ ﻣﻲﻛﻨﻴﻢ ﺗﺎ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺭﺍ ﻗﻄﻊ ﻛﻨﺪ‪ .‬ﻣﺤﻞ ﺗﻘﺎﻃﻊ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۴۳‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﻫﻤﺎﻥ ﻧﻘﻄﻪ ‪ x1‬ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺑﺎ ﻣﺸﺨﺺ ﺑﻮﺩﻥ ﻣﻮﻗﻌﻴﺖ ‪ x0‬ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺗﻮﺍﻧﺴﺘﻴﻢ ﻣﻮﻗﻌﻴﺖ‬
‫) ‪ x1 = g(x0‬ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺭﺍ ﻧﻴﺰ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﺑﺎ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺮﺍﻳﻨﺪ ﻣﻮﻗﻌﻴﺖ ‪ x3 ،x2‬ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻳﮕﺮ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﺩﺳﺖ ﺁﻭﺭﺩ‪.‬‬

‫ﺩﺭ ﺷﻜﻞ ﻣﻘﺎﺑﻞ‪ ،‬ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬

‫) ‪x1 = g(x0‬‬
‫) ‪(x1 , x1‬‬ ‫ﺍﻋﻤﺎﻝ ﺷﺪﻩ ﺭﻭﻱ ﺗﺎﺑﻊ )‪ g(x) = cos(x‬ﺩﺭ ﻣﺜﺎﻝ‬
‫‪ ۷-۵‬ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎ‬
‫) ‪x3 = g(x2‬‬
‫ﺭﺳﻢ ﺧﻂﻫﺎﻱ ﺍﻓﻘﻲ ﻭ ﻋﻤﻮﺩﻱ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﻮﻗﻌﻴﺖ‬
‫ﻫﻨﺪﺳﻲ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﺷﺪﻩ ﺗﻮﺳﻂ ﺭﻭﺵ‬
‫ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﻪ ﺳﺎﺩﮔﻲ ﻳﺎﻓﺖ‪.‬‬
‫) ‪(x2 , x2‬‬ ‫) ‪x2 = g(x1‬‬
‫(‪g‬‬ ‫ﺍﻟﺒﺘﻪ ﻟﺰﻭﻣﻲ ﻧﺪﺍﺭﺩ ﻛﻪ ﺗﻤﺎﻣﻲ ﺟﺰﺋﻴﺎﺕ ﻣﻮﺟﻮﺩ ﺩﺭ‬
‫)‪x‬‬
‫=‬ ‫ﺷﻜﻞ ﺭﺳﻢ ﺷﻮﺩ‪ .‬ﺑﻠﻜﻪ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺳﺎﺩﮔﻲ ﻓﻘﻂ‬
‫‪co‬‬
‫(‪s‬‬
‫)‪x‬‬
‫ﻣﺴﻴﺮﻱ ﻛﻪ ﺑﻪ ﺻﻮﺭﺕ ﺧﻂﭼﻴﻦ ﺭﺳﻢ ﺷﺪﻩ ﺭﺍ ﺩﻧﺒﺎﻝ‬
‫ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﮔﻮﻳﺎﺗﺮﻱ ﺑﺮﺍﻱ‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺍﻳﻦ ‪x‬‬
‫•‬ ‫•‬ ‫•‬ ‫• •‬ ‫•‬
‫‪x0‬‬ ‫‪x2 x4 p x3‬‬ ‫‪x1‬‬ ‫ﻣﺴﻴﺮ ﺑﻪ ﺻﻮﺭﺕ ﻣﺠﺰﺍ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﺩﺭ ﺷﻜﻞ ﻣﻘﺎﺑﻞ‪ ،‬ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺧﻼﺻﻪ ﻭ ﮔﻮﻳﺎﺗﺮ‬

‫‪۲‬‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۷-۵‬ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪۶‬‬ ‫ﺑﺮﺍﻱ ﺗﻮﻟﻴﺪ ﺍﻳﻦ ﻣﺴﻴﺮ‪ ،‬ﺩﺳﺘﻮﺭﺍﻟﻌﻤﻞ ﺯﻳﺮ ﺭﺍ ﺑﻪ ﻛﺎﺭ‬
‫‪۷‬‬ ‫‪۳‬‬ ‫ﺑﺮﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫•‬
‫‪۵‬‬
‫• ﺍﺯ ‪ x0‬ﺭﻭﻱ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺑﻪ ﺻﻮﺭﺕ ﻋﻤﻮﺩﻱ‬
‫‪۱‬‬ ‫‪۴‬‬ ‫ﺣﺮﻛﺖ ﻛﻦ ﺗﺎ ﺑﻪ ﻧﻤﻮﺩﺍﺭ ﺗﺎﺑﻊ ‪ g‬ﺑﺮﺳﻲ‪.‬‬
‫• ﺑﻌﺪ ﺑﻪ ﺻﻮﺭﺕ ﺍﻓﻘﻲ ﺣﺮﻛﺖ ﻛﻦ ﺗﺎ ﻧﻴﻤﺴﺎﺯ‬
‫ﺍﻭﻝ ﻭ ﺳﻮﻡ ﺭﺍ ﻗﻄﻊ ﻛﻨﻲ‬
‫• ﺑﻌﺪ ﺑﻪ ﺻﻮﺭﺕ ﻋﻤﻮﺩﻱ ﺣﺮﻛﺖ ﻛﻦ ﺗﺎ‬
‫‪x‬‬ ‫ﻧﻤﻮﺩﺍﺭ ‪ g‬ﺭﺍ ﻗﻄﻊ ﻛﻨﻲ‬
‫‪x0‬‬ ‫• ﺭﻭﺍﻝ ﻓﻮﻕ ﺭﺍ ﺗﻜﺮﺍﺭ ﻛﻦ‬

‫ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ‪ ،‬ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﺑﻬﺘﺮ‪ ،‬ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﺩﻭ ﺗﺎﺑﻊ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴۴‬‬

‫‪g(x) = 2.8x − x2‬‬ ‫‪g(x) = − 92 x3 +‬‬ ‫‪23 2‬‬

‫‪2 x‬‬ ‫‪− 9x + 3‬‬
‫‪1.8‬‬
‫‪1‬‬

‫‪x‬‬ ‫‪x‬‬
‫‪x0‬‬ ‫‪1.8‬‬ ‫‪x0‬‬ ‫‪1‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫�‬

‫ﻣﻄﺎﺑﻖ ﺑﺎ ﻗﻀﻴﻪ ‪ ،۶-۵‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﻳﻚ ﺭﻭﺵ ﻣﻨﺎﺳﺐ ﻭ ﺳﺮﺭﺍﺳﺖ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﺮ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺩﺭ ﺍﻳﻦ ﻗﻀﻴﻪ ﻳﻚ‬
‫ﻓﺮﺽ ﺑﺰﺭﮒ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﻭ ﺁﻥ ﻓﺮﺽ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﺷﺪﻩ‪ .‬ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ ﺍﮔﺮ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ‬
‫ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﻮﻓﻖ ﺷﺪﻩ ﻛﻪ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ ﺻﻴﺪ ﻛﻨﺪ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ‪،‬‬
‫ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻫﻤﻮﺍﺭﻩ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ؟ ﭘﺎﺳﺦ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﻧﺘﻴﺠﻪ ﺯﻳﺮ‪ ،‬ﺑﺪﻳﻬﻲﺗﺮﻳﻦ ﺣﺎﻟﺘﻲ‬
‫ﻛﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺪ ﺭﺍ ﺑﻴﺎﻥ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺍﻳﻦ ﻧﺘﻴﺠﻪ ﺑﻼﻓﺎﺻﻠﻪ ﺍﺯ ﻗﻀﻴﻪ ‪ ۶-۵‬ﺍﺳﺘﻨﺘﺎﺝ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻧﺘﯿﺠﻪ ‪۷−۵‬‬
‫ﺍﮔﺮ ‪ g‬ﺗﺎﺑﻌﻲ ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ ﻛﻪ ﺩﺍﺭﺍﻱ ﻫﻴﭻ ﻧﻘﻄﻪ ﺛﺎﺑﺘﻲ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺑﺎ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‪،‬‬
‫ﻫﻤﮕﺮﺍ ﻧﺨﻮﺍﻫﺪ ﺷﺪ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻋﺪﻡ ﻭﺟﻮﺩ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﻣﻨﺠﺮ ﺑﻪ ﻭﺍﮔﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻝ ﺩﻳﮕﺮﻱ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﮔﺮﺩﺩ ﺍﻳﻦ‬
‫ﺍﺳﺖ ﻛﻪ‬

‫» ﺁﻳﺎ ﻭﺟﻮﺩ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﺩﻟﻴﻠﻲ ﺑﺮ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺧﻮﺍﻫﺪ ﺑﻮﺩ؟ «‬

‫ﭘﺎﺳﺦ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺗﺎﺑﻊ ‪ g(x) = 2x‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﺗﻨﻬﺎ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 0‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫∞} ‪ {x0 2n‬ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﻧﺒﺎﻟﻪﺍﻱ ﻭﺍﮔﺮﺍ‬
‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺷﺮﻭﻉ ﺍﻭﻟﻴﻪ ‪ x0 ̸= 0‬ﻣﻨﺠﺮ ﺑﻪ ﺩﻧﺒﺎﻟﻪ ‪n=0‬‬
‫ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻭﺟﻮﺩ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﺗﻀﻤﻴﻨﻲ ﺑﺮﺍﻱ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ‪ p‬ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ g‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = p‬ﻣﻨﺠﺮ ﺑﻪ ﺩﻧﺒﺎﻟﻪ‬
‫ﺛﺎﺑﺖ · · · ‪ p, p, p,‬ﻣﻲﺷﻮﺩ‪ ،‬ﻛﻪ ﺩﻧﺒﺎﻟﻪﺍﻱ ﻫﻤﮕﺮﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ ﻧﻘﻄﻪ ﺷﺮﻭﻉ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﻀﻤﻴﻦ ﺷﺪﻩ‬
‫ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻨﺠﺎ ﻣﻨﻈﻮﺭ ﺍﺯ ﻫﻤﮕﺮﺍﻳﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻛﻪ ﺁﻥ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‬
‫ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ‪:‬‬

‫» ﺁﻳﺎ ﻫﻤﮕﺮﺍﻳﻲ ﻭ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ؟ «‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۴۵‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺩﺭ ﭘﺎﺳﺦ ﺍﻳﻦ ﺳﻮﺍﻝ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻋﻼﻭﻩ ﺑﺮ ﺍﻧﺘﺨﺎﺏ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﻪ ﺗﺎﺑﻊ ‪ g‬ﻧﻴﺰ ﺑﺴﺘﮕﻲ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﺭﻭﺷﻦ ﺷﺪﻥ‬
‫ﻣﻮﺿﻮﻉ‪ ،‬ﺳﻪ ﺣﺎﻟﺖ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬

‫• ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﻴﭻ ﺣﺪﺱ ﺍﻭﻟﻴﻪﺍﻱ )ﻫﺮﭼﻨﺪ ﻧﺰﺩﻳﻚ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ( ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﻤﮕﺮﺍ‬
‫ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺭﻭﺵ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮ ﺭﻭﻱ ﺗﺎﺑﻊ ‪ g(x) = 2x‬ﺑﺎ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻏﻴﺮﺻﻔﺮ‪ ،‬ﻫﻴﭻﮔﺎﻩ ﻫﻤﮕﺮﺍ‬
‫ﻧﺨﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬
‫• ﺍﻣﺎ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﺩﻳﮕﺮ‪ ،‬ﺑﺎ ﺍﻧﺘﺨﺎﺏ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻧﺰﺩﻳﻚ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻫﻤﮕﺮﺍﻳﻲ ﺑﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬
‫ﺑﻪ ﺁﻥ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ ﺗﻀﻤﻴﻦ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺭﻭﺵ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮ ﺭﻭﻱ ﺗﺎﺑﻊ ‪ g(x) = x2‬ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ‬
‫ﺍﻭﻟﻴﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (−1, 1‬ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 0‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ ،‬ﺍﻣﺎ ﺑﻪ ﺍﺯﺍﻱ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺩﻳﮕﺮ ﻭﺍﮔﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫• ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫ﻧﻤﻮﻧﻪ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﻭﻱ ﺗﺎﺑﻊ )‪ g(x) = cos(x‬ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﻟﺤﺎﻅ ﻫﻤﮕﺮﺍﻳﻲ ﺩﺭ ﺭﻓﺘﺎﺭﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺍﺯ ﺧﻮﺩ ﻧﺸﺎﻥ ﻣﻲﺩﻫﺪ‪ .‬ﺩﺭ ﺑﺮﺧﻲ ﺣﺎﻻﺕ‪ ،‬ﻫﺮ ﺍﻧﺘﺨﺎﺏ‬
‫ﺑﺮﺍﻱ ﺣﺪﺱ ﺍﻭﻟﻴﻪ )ﻫﺮﭼﻨﺪ ﻧﺰﺩﻳﻚ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ( ﻣﻨﺠﺮ ﺑﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﺣﺎﻻﺕ ﺩﻳﮕﺮ ﺑﺎ ﺍﻧﺘﺨﺎﺏ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‬
‫ﻭ ﻳﺎ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻧﺰﺩﻳﻚ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺁﻥ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﺴﻠﻤﺎ ﺍﻳﻦ ﺣﺎﻻﺕ ﻣﻄﻠﻮﺏ‬
‫ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﻣﻔﺎﻫﻴﻢ ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﺍﺳﺮﻱ ‪ ۲۳‬ﻭ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ‪ ۲۴‬ﺗﻌﺮﻳﻒ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻫﻤ ﺮﺍﯾﯽ ﺳﺮﺍﺳﺮﯼ ﻭ ﻣﻮﺿﻌ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﺗﻌﺮﯾﻒ ‪۸−۵‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ p‬ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ‪ g‬ﺑﺎﺷﺪ‪.‬‬

‫ﺍﻟﻒ( ﮔﻮﻳﻢ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ‪ g‬ﺑﻪ ﻃﻮﺭ ﺳﺮﺍﺳﺮﻱ ﻫﻤﮕﺮﺍ ﺑﻪ ‪ p‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻫﺮﮔﺎﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﻧﻘﻄﻪ ﺷﺮﻭﻉ ﺩﻟﺨﻮﺍﻩ‬
‫ﺩﺭ ﺩﺍﻣﻨﻪ ﺗﺎﺑﻊ ‪ ،g‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ‪ p‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪.‬‬

‫ﺏ( ﮔﻮﻳﻢ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ‪ g‬ﺑﻪ ﻃﻮﺭ ﻣﻮﺿﻌﻲ ﻫﻤﮕﺮﺍ ﺑﻪ ‪ p‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﻫﺮﮔﺎﻩ ﺑﺎﺯﻩﺍﻱ ﻣﺎﻧﻨﺪ )‪ (a, b‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ p‬ﺩﺍﺧﻞ ﺍﻳﻦ ﺑﺎﺯﻩ ﺑﺎﺷﺪ ﻭ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﻧﻘﻄﻪ ﺷﺮﻭﻉ ﺍﺯ ﺩﺍﺧﻞ ﺍﻳﻦ ﺑﺎﺯﻩ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ‪ p‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪.‬‬

‫ﺑﻪ ﻃﻮﺭ ﺳﺮﺍﺳﺮﻱ ﻫﻤﮕﺮﺍ ﺷﺪﻥ ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﺍﻧﺘﺨﺎﺏ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‪ ،‬ﻣﺎ ﺭﺍ ﺑﻪ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻲﺭﺳﺎﻧﺪ‪ .‬ﺩﺭ‬
‫ﻣﻘﺎﺑﻞ‪ ،‬ﺑﻪ ﻃﻮﺭ ﻣﻮﺿﻌﻲ ﻫﻤﮕﺮﺍ ﺷﺪﻥ ﺑﻪ ﻣﻌﻨﻲ ﻫﻤﻮﺍﺭﻩ ﻫﻤﮕﺮﺍ ﺷﺪﻥ ﻧﻴﺴﺖ‪ ،‬ﺑﻠﻜﻪ ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﻧﻘﺎﻁ ﺍﻭﻟﻴﻪﺍﻱ ﻣﻲﺗﻮﺍﻥ‬
‫ﻳﺎﻓﺖ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﺁﻥ ﺭﻭﺵ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۹−۵‬ﻫﻤ ﺮﺍﯾﯽ ﺳﺮﺍﺳﺮﯼ ﻭ ﻣﻮﺿﻌ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ(‬

‫◀ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮ ﺭﻭﻱ ﺗﺎﺑﻊ )‪ g(x) = cos(x‬ﺩﺍﺭﺍﻱ ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﺍﺳﺮﻱ ﺑﻪ ﻧﻘﻄﻪ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫◀ ﺗﺎﺑﻊ ‪ g(x) = x2‬ﺩﺍﺭﺍﻱ ﺩﻭ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 0‬ﻭ ‪ p = 1‬ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ‬
‫ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪ ﺩﺭ ﺑﺎﺯﻩ )‪ (−1, 1‬ﺑﻪ ‪ p = 0‬ﻫﻤﮕﺮﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ‪ p = 0‬ﺑﻪ ﻃﻮﺭﻱ ﻣﻮﺿﻌﻲ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‬
‫ﻭ ﺍﻟﺒﺘﻪ ﺍﻳﻦ ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﺍﺳﺮﻱ ﻧﻴﺴﺖ‪ .‬ﺍﻣﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ )ﺑﻪ ﺟﺰ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ (1‬ﻫﻴﭻﮔﺎﻩ ﺑﻪ ‪ p = 1‬ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺍﺭﺍﻱ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺑﻪ ‪ p = 1‬ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬
‫◀ ﺗﺎﺑﻊ )‪ g(x) = 21 (x2 + 1‬ﺩﺍﺭﺍﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ‪ p = 1‬ﺍﺳﺖ‪ .‬ﻣﻲﺗﻮﺍﻥ ﺛﺎﺑﺖ ﻛﺮﺩ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ‬
‫‪23 Globally‬‬ ‫‪Convergent‬‬ ‫‪24 Locally‬‬ ‫‪Convergent‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴۶‬‬

‫ﺣﺪﺱ ﺍﻭﻟﻴﻪ )‪ ،x0 ∈ (0, 1‬ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ 1‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ ،‬ﺍﻣﺎ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 > 1‬ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪،‬‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﻃﻮﺭ ﻣﻮﺿﻌﻲ ﺑﻪ ‪ p = 1‬ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺷﻮﺩ ﻛﻪ ﻫﺮﭼﻨﺪ ﺑﻪ ﺍﺯﺍﻱ ‪ x0‬ﻧﺰﺩﻳﻚ ﻭ ﻛﻮﭼﻚﺗﺮ‬
‫ﺍﺯ ‪ 1‬ﻫﻤﮕﺮﺍﻳﻲ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ ،‬ﺍﻣﺎ ﻫﻤﮕﺮﺍﻳﻲ ﺑﻪ ﺍﺯﺍﻱ ‪ x0‬ﺑﺰﺭﮒﺗﺮ ﺍﺯ ‪ 1‬ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺪ‪ ،‬ﻫﺮ ﭼﻨﺪ ﻛﻪ ﺑﻪ ‪ 1‬ﻧﺰﺩﻳﻚ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ‬
‫ﺩﻳﮕﺮ ﺑﺎﺯﻩﺍﻱ ﺷﺎﻣﻞ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻳﺎﻓﺖ ﻧﻤﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﻧﻘﻄﻪ ﺩﺭ ﺁﻥ ﺑﺎﺯﻩ ﻫﻤﮕﺮﺍﻳﻲ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪.‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﺍﺳﺮﻱ ﺧﺎﺻﻴﺖ ﻣﻄﻠﻮﺑﻲ ﺍﺳﺖ ﻭﻟﻲ ﺩﺭ ﻣﻮﺍﺭﺩ ﻧﺎﺩﺭ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﻧﻴﺰ ﺧﺎﺻﻴﺖ ﻣﻨﺎﺳﺒﻲ ﺍﺳﺖ‬
‫ﻛﻪ ﺑﺎ ﻭﺟﻮﺩ ﺁﻥ ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ﺑﺎ ﺍﻧﺘﺨﺎﺏ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻧﺰﺩﻳﻚ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻣﻄﻤﻴﻦ ﺑﻮﺩ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﻗﻀﺎﻳﺎﻳﻲ ﺩﺭ ﺧﺼﻮﺹ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﺑﺎ ﻛﻤﻚ ﺁﻥﻫﺎ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﻪ ﺍﻧﺘﺨﺎﺏ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻣﻨﺎﺳﺐ ﭘﺮﺩﺍﺧﺖ‪.‬‬

‫ﺷﺮﻁ ﮐﺎﻓ ﺑﺮﺍﯼ ﻫﻤ ﺮﺍ ﺑﻮﺩﻥ ﻣﻮﺿﻌ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻗﻀﯿﻪ ‪۹−۵‬‬

‫ﺍﮔﺮ ‪ p‬ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﺑﻪ ﻃﻮﺭ ﭘﻴﻮﺳﺘﻪ ﻣﺸﺘﻖﭘﺬﻳﺮ ‪ g‬ﺑﺎﺷﺪ ﻭ ‪ ،|g ′ (p)| < 1‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﻃﻮﺭ‬
‫ﻣﻮﺿﻌﻲ ﺑﻪ ‪ p‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ ‪ ،|g (p)| < 1‬ﺁﻥﮔﺎﻩ ﺑﺎﺯﻩﺍﻱ ﺑﺎﺯ ﺷﺎﻣﻞ ‪ p‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ ﻫﺮ ﻧﻘﻄﻪ ﺷﺮﻭﻉ ﺍﺯ ﺍﻳﻦ‬
‫‪′‬‬

‫ﺑﺎﺯﻩ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺷﺮﻁ ‪ |g ′ (p)| < 1‬ﺗﻀﻤﻴﻦ ﻣﻲﻛﻨﺪ ﻛﻪ ﺣﺪﺱﻫﺎﻱ ﺍﻭﻟﻴﻪﺍﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﺁﻥﻫﺎ ﺑﻪ‬
‫ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪ .‬ﺍﺯ ﻗﻀﻴﻪ ﻓﻮﻕ ﻭ ﻫﻤﭽﻨﻴﻦ ﻗﻀﻴﻪ ‪ ۵-۵‬ﻣﻲﺗﻮﺍﻥ ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﻮﺩ‪.‬‬

‫ﺷﺮﻁ ﮐﺎﻓ ﺑﺮﺍﯼ ﻫﻤ ﺮﺍ ﺑﻮﺩﻥ ﻣﻮﺿﻌ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻗﻀﯿﻪ ‪۱۰−۵‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﻭ ﺩﺍﺭﺍﻱ ﻣﺸﺘﻖ ﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ‬
‫]‪ x ∈ [a, b‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ |g ′ (x)| ≤ l < 1 :‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ‬
‫ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ )ﻛﻪ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﻣﻲﺑﺎﺷﺪ( ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪.‬‬

‫ﺑﺎ ﺗﺮﻛﻴﺐ ﻗﻀﺎﻳﺎﻱ ﻓﻮﻕ ﻭ ﻫﻤﭽﻨﻴﻦ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﻀﻴﻪ ‪ ۴-۵‬ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺭﺍ ﺍﺳﺘﻨﺘﺎﺝ ﻛﺮﺩ‪:‬‬

‫ﻗﻀﯿﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ −‬ﻗﻀﯿﻪ ﺷﺮﺍﯾﻂ ﮐﺎﻓ ﺑﺮﺍﯼ ﻫﻤ ﺮﺍﯼ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻧﺘﯿﺠﻪ ‪۱۱−۵‬‬
‫ﺍﮔﺮ ‪ g‬ﺗﺎﺑﻌﻲ ﺑﺎ ﺷﺮﺍﻳﻂ ﺯﻳﺮ ﺑﺎﺷﺪ‬

‫)‪ (۱‬ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﻭ ‪ g ′‬ﺩﺭ )‪ (a, b‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ‬

‫)‪ (۲‬ﺗﺎﺑﻊ ‪ g‬ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﻪ ﺗﻮﻱ ﺧﻮﺩﺵ ﺑﺒﺮﺩ‪ .‬ﻳﻌﻨﻲ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ]‪ ،x ∈ [a, b‬ﺁﻥﮔﺎﻩ ]‪.g(x) ∈ [a, b‬‬

‫)‪ (۳‬ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ ρ < 1‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ ﻛﻪ ﺑﻪﺍﺯﺍﻱ ﻫﺮ )‪ x ∈ (a, b‬ﺁﻥﮔﺎﻩ ‪| g ′ (x) |≤ ρ < 1‬‬

‫ﺩﺭ ﺁﻥ ﺻﻮﺭﺕ‬

‫ﺍﻟﻒ( ﺗﺎﺑﻊ ‪ g‬ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ]‪ [a, b‬ﺍﺳﺖ‪ .‬ﻭ ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺍﻳﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ‪.‬‬

‫ﺏ( ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ )‪ x0 ∈ (a, b‬ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۴۷‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ‪ ،‬ﻛﺎﺭﺑﺮﺩ ﻧﺘﻴﺠﻪ ‪ ۱۱-۵‬ﺩﺭ ﺗﺤﻠﻴﻞ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺑﺎﺯﻩ(‬ ‫ﺗﺎﺑﻊ ﺩﺭ ﯾ‬ ‫ﻣﺜﺎﻝ ‪)۱۰−۵‬ﺑﺮﺭﺳ ﻭﺟﻮﺩ ﻭ ﯾ ﺘﺎﯾﯽ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻭ ﻫﻤ ﺮﺍﯾﯽ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﯼ ﯾ‬

‫ﺗﺎﺑﻊ )‪ g1 (x) = cos(x‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 1‬ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺗﺎﺑﻊ ‪ g1‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 1‬ﭘﻴﻮﺳﺘﻪ ﺍﺳﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ﻣﺸﺘﻖ ﺁﻥ ﻧﻴﺰ‬
‫ﺩﺭ ﺍﻳﻦ ﺑﺎﺯﻩ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬
‫ﺑﻪﺍﺯﺍﻱ ﻫﺮ ]‪ x ∈ [0, 1‬ﺩﺍﺭﻳﻢ ]‪ cos(x) ∈ [cos(1), 1‬ﻭ ﭼﻮﻥ ]‪ [cos(1), 1] ⊂ [0, 1‬ﻟﺬﺍ ﺗﺎﺑﻊ ‪ g1‬ﺑﺎﺯﻩ ]‪ [0, 1‬ﺭﺍ ﺑﻪ ﺩﺍﺧﻞ‬
‫ﺧﻮﺩ ﺑﺎﺯﻩ ﻣﻲﺑﺮﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻃﺒﻖ ﻗﻀﻴﻪ ‪ ،۴-۵‬ﺗﺎﺑﻊ ‪ g1‬ﺩﺍﺭﺍﻱ ﺣﺪﺍﻗﻞ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺩﺭ ]‪ [0, 1‬ﺍﺳﺖ‪.‬‬
‫ﺩﺍﺭﻳﻢ )‪ g1′ (x) = − sin(x‬ﻭ ﺑﻪ ﺍﺯﺍﻱ ]‪ x ∈ [0, 1‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪| g1′ (x) |≤ sin(1) < 1‬‬
‫ﻟﺬﺍ ﻃﺒﻖ ﻗﻀﻴﻪ ‪ ۵-۵‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ ﺷﺮﻭﻉ ﺍﺯ ﻳﻚ‬
‫ﻧﻘﻄﻪ ﺩﺍﺧﻞ ﺑﺎﺯﻩ ]‪ [0, 1‬ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫■ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬

‫ﮔﻔﺘﻴﻢ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﻤﻜﻦ ﺍﺳﺖ ﻫﻤﮕﺮﺍ ﻧﺸﻮﺩ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﭘﻴﺶ ﻣﻲﺁﻳﺪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻫﻨﮕﺎﻡ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﺭﻓﺘﺎﺭ‬
‫ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﭼﮕﻮﻧﻪ ﺍﺳﺖ‪ .‬ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺩﺭ ﺣﺎﻟﺖ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﭼﻬﺎﺭ ﺣﺎﻟﺖ ﺯﻳﺮ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪.‬‬

‫• ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻣﺜﺒﺖ ﻳﺎ ﻣﻨﻔﻲ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻛﻨﺪ‪ ،‬ﺑﻪ ﺍﺻﻄﻼﺡ ﺩﻧﺒﺎﻟﻪ ﻭﺍﮔﺮﺍ ﺷﻮﺩ‪.‬‬

‫• ﻣﻤﻜﻦ ﺍﺳﺖ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﺻﻮﺭﺕ ﺗﻨﺎﻭﺑﻲ ﺗﻜﺮﺍﺭ ﺷﻮﺩ ﻭ ﻟﺬﺍ ﻫﻴﭻﮔﺎﻩ ﻫﻤﮕﺮﺍ ﻧﺸﻮﺩ‪.‬‬

‫• ﻣﻤﻜﻦ ﺍﺳﺖ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﻧﻪ ﺑﻪ ﻃﻮﺭ ﺗﻨﺎﻭﺑﻲ ﺗﻜﺮﺍﺭ ﺷﻮﻧﺪ ﻭ ﻧﻪ ﺑﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻭﺍﮔﺮﺍ ﺷﺪ‪ .‬ﺑﻠﻜﻪ ﺑﻪ ﻃﻮﺭ ﺁﺷﻮﺑﻨﺎﻛﻲ ﺗﻮﻟﻴﺪ‬
‫ﮔﺮﺩﻧﺪ‪.‬‬

‫• ﻣﻤﻜﻦ ﺍﺳﺖ ﻫﻨﮕﺎﻡ ﺗﻮﻟﻴﺪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺑﺎ ﺣﺎﻟﺖﻫﺎﻱ ﺗﻘﺴﻴﻢ ﺑﺮ ﺻﻔﺮ ﻭ ﻳﺎ ﻣﻨﻔﻲ ﺯﻳﺮ ﺭﺍﺩﻳﻜﺎﻝ ﻣﻮﺍﺟﻪ ﺷﻮﻳﻢ ﻛﻪ ﺩﺭ‬
‫ﺍﻳﻦ ﺻﻮﺭﺕ ﺍﺩﺍﻣﻪ ﺗﻮﻟﻴﺪ ﺩﻧﺒﺎﻟﻪ ﻏﻴﺮﻣﻤﻜﻦ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ‪ ،‬ﻭﻗﻮﻉ ﺳﻪ ﺣﺎﻟﺖ ﺍﻭﻝ ﺍﺯ ﺣﺎﻟﺖﻫﺎﻱ ﻓﻮﻕ ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻭﻗﻮﻉ ﺣﺎﻟﺖ ﭼﻬﺎﺭﻡ ﺩﺭ ﻣﺜﺎﻝ ‪ ۱۴-۵‬ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ ‪ −‬ﺗﻮﻟﯿﺪ ﺩﻧﺒﺎﻟﻪ ﻭﺍﮔﺮﺍ(‬ ‫ﻣﺜﺎﻝ ‪)۱۱−۵‬ﻋﺪﻡ ﻫﻤ ﺮﺍﯾﯽ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﯼ ﯾ‬

‫ﺗﺎﺑﻊ ‪ g(x) = x3 + x − 1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 1‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬
‫ﺭﺍ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 0.95‬ﺩﻧﺒﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺴﻴﺎﺭ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 1‬ﻧﺰﺩﻳﻚ ﺍﺳﺖ‪ .‬ﺑﻪ‬
‫ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ x0 ،‬ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺴﻴﺎﺭ ﻣﻨﺎﺳﺒﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺭ ﺟﺪﻭﻝ ﻭ ﺷﻜﻞ ﺯﻳﺮ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۴۸‬‬

‫‪k‬‬ ‫‪xk‬‬
‫‪0‬‬ ‫‪9.500000000000000 e-01‬‬
‫‪1‬‬ ‫‪8.073750000000000 e-01‬‬
‫‪2‬‬ ‫‪3.336659386308591 e-01‬‬ ‫‪1‬‬
‫‪3‬‬ ‫‪-6.291860452365163 e-01‬‬
‫‪4‬‬ ‫‪-1.878265120927628 e+00‬‬
‫‪5‬‬ ‫‪-9.504558821228233 e+00‬‬ ‫‪x‬‬
‫‪6‬‬ ‫‪-8.691144520747710e+02‬‬ ‫‪1‬‬
‫‪7‬‬ ‫‪-6.564951024954759 e+08‬‬
‫‪8‬‬ ‫‪-2.829400798127567 e+26‬‬
‫‪9‬‬ ‫‪-2.265079320443056 e+79‬‬
‫‪10‬‬ ‫‪-1.162118046291276e+238‬‬
‫‪11‬‬ ‫‪-Inf‬‬ ‫‪g(x) = x3 + x − 1‬‬
‫‪12‬‬ ‫‪-Inf‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺍﻧﺪﺍﺯﻩ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺑﺰﺭﮒ ﻭ ﺑﺰﺭﮒﺗﺮ ﻣﻲﺷﻮﻧﺪ ﻭ ﺩﺭ ﻧﻬﺎﻳﺖ ﺑﻌﺪ ﺍﺯ ﺩﻩ ﺗﻜﺮﺍﺭ ﺳﺮﺭﻳﺰﻱ ﺍﺗﻔﺎﻕ‬
‫ﻣﻲﺍﻓﺘﺪ ﻭ ﺣﺎﺻﻞ ‪ −Inf‬ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ ‪ −‬ﺗﻮﻟﯿﺪ ﺁﺷﻮﺑﻨﺎﮎ ﺩﻧﺒﺎﻟﻪ(‬ ‫ﻣﺜﺎﻝ ‪)۱۲−۵‬ﻋﺪﻡ ﻫﻤ ﺮﺍﯾﯽ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﯼ ﯾ‬

‫ﺗﺎﺑﻊ ‪ g(x) = x2 − 2‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﺩﻭ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = −1‬ﻭ ‪ p = 2‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺭﻭﺵ‬
‫ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 1.9900000‬ﻛﻪ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 2‬ﺑﺴﻴﺎﺭ ﻧﺰﺩﻳﻚ ﺍﺳﺖ ﺭﺍ ﺩﻧﺒﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻧﺘﺎﻳﺞ‬
‫ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‪ ،‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺭﻓﺘﺎﺭ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﻧﻴﺰ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺑﻌﺪ ﺍﺯ ﺩﻩ ﻫﺰﺍﺭ ﺗﻜﺮﺍﺭ‪ ،‬ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻃﻮﺭ ﺁﺷﻮﺑﻨﺎﻛﻲ ﺩﺭ ﺑﺎﺯﻩ ]‪ [−2, +2‬ﺗﻮﻟﻴﺪ ﻣﻲﺷﻮﻧﺪ‬
‫ﻭ ﺍﺯ ﺍﻳﻦ ﺑﺎﺯﻩ ﺧﺎﺭﺝ ﻧﻤﻲﺷﻮﻧﺪ‪ ،‬ﺍﻣﺎ ﻫﻤﮕﺮﺍﻳﻲ ﻫﻢ ﺍﺗﻔﺎﻕ ﻧﻤﻲﺍﻓﺘﺪ‪ ،‬ﻫﺮﭼﻨﺪ ﻛﻪ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻣﻨﺎﺳﺐ ﺍﻧﺘﺨﺎﺏ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪k‬‬ ‫‪xk‬‬ ‫‪k‬‬ ‫‪xk‬‬ ‫‪k‬‬ ‫‪xk‬‬
‫‪0‬‬ ‫‪1.990000‬‬ ‫‪8‬‬ ‫‪1.934821‬‬ ‫‪1004‬‬ ‫‪-1.777913‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪1‬‬ ‫‪1.999996‬‬ ‫‪9‬‬ ‫‪1.743532‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪2‬‬ ‫‪1.999984‬‬ ‫‪10‬‬ ‫‪1.039906‬‬ ‫‪9995‬‬ ‫‪-1.321900‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪x‬‬
‫‪3‬‬ ‫‪1.999936‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪9996 -0.252580‬‬
‫‪−2‬‬ ‫‪2‬‬
‫‪4‬‬ ‫‪1.999744‬‬ ‫‪1000‬‬ ‫‪-1.949103‬‬ ‫‪9997 -1.936203‬‬
‫‪5‬‬ ‫‪1.998976‬‬ ‫‪1001‬‬ ‫‪1.799005‬‬ ‫‪9998 1.748882‬‬
‫‪6‬‬ ‫‪1.995905‬‬ ‫‪1002‬‬ ‫‪1.236421‬‬ ‫‪9999 1.058591‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪7‬‬ ‫‪1.983638‬‬ ‫‪1003‬‬ ‫‪-0.471260‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪g(x) = x2 − 2‬‬

‫ﺗﺎﺑﻊ ‪ −‬ﺗﻮﻟﯿﺪ ﺁﺷﻮﺑﻨﺎﮎ ﺗﻨﺎﻭﺑﯽ(‬ ‫ﻣﺜﺎﻝ ‪)۱۳−۵‬ﻋﺪﻡ ﻫﻤ ﺮﺍﯾﯽ ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﯼ ﯾ‬

‫ﺗﺎﺑﻊ ‪ g(x) = 1 − x3‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ · · · ‪ p = 0.6823278037‬ﺍﺳﺖ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ‬
‫ﻛﻪ ﺍﮔﺮ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 1‬ﺷﺮﻭﻉ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪:‬‬
‫· · · ‪1, 0, 1, 0, 1, 0,‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۴۹‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﻛﻪ ﺩﻧﺒﺎﻟﻪﺍﻱ ﺗﻨﺎﻭﺑﻲ ﺍﺳﺖ ﻭ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫ﺣﺎﻝ ﺍﮔﺮ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﺎ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 1‬ﺷﺮﻭﻉ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻌﺪ‬
‫‪k‬‬ ‫‪xk‬‬
‫‪0‬‬ ‫‪0.500000000000000‬‬ ‫ﺍﺯ ﺗﻌﺪﺍﺩﻱ ﺗﻜﺮﺍﺭ ﺩﻭﺑﺎﺭﻩ ﺩﻧﺒﺎﻟﻪ ﺗﻨﺎﻭﺑﻲ ﻣﻲﺷﻮﺩ ﻭ ﻣﻮﻟﻔﻪﻫﺎﻱ ‪ 0‬ﻭ ‪ 1‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺗﻨﺎﻭﺑﻲ‬
‫‪1‬‬ ‫‪0.875000000000000‬‬ ‫ﺍﻳﺠﺎﺩ ﻣﻲﻛﻨﺪ‪ .‬ﺑﺮﺍﻱ ﻧﻤﻮﻧﻪ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 0.5‬ﺩﻧﺒﺎﻟﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺭ ﺟﺪﻭﻝ‬
‫‪2‬‬ ‫‪0.330078125000000‬‬
‫‪3‬‬ ‫‪0.964037470519543‬‬ ‫ﺭﻭﺑﺮﻭ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﻓﺮﺍﻳﻨﺪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻣﺼﻮﺭ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪4‬‬ ‫‪0.104054188327677‬‬
‫‪5‬‬ ‫‪0.998873376780835‬‬ ‫‪1‬‬
‫‪6‬‬ ‫‪0.003376063247860‬‬
‫‪7‬‬ ‫‪0.999999961520296‬‬
‫‪8‬‬ ‫‪0.000000115439107‬‬
‫‪9‬‬ ‫‪1.000000000000000‬‬
‫‪10‬‬ ‫‪0‬‬
‫‪11‬‬ ‫‪1‬‬
‫‪12‬‬ ‫‪0‬‬
‫‪13‬‬ ‫‪1‬‬ ‫‪x‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪g(x) = 1 − x‬‬ ‫‪3‬‬ ‫‪1‬‬

‫ﺩﺭ ﺑﺎﻻ ﻣﺜﺎﻝﻫﺎﻳﻲ ﺍﺯ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝﻫﺎ ﺩﻳﺪﻡ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻧﻤﻲﺗﻮﺍﻧﺪ‬
‫ﻳﻚ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺭﺍ ﺻﻴﺪ ﻛﻨﺪ‪ ،‬ﻫﺮﭼﻨﺪ ﻛﻪ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺴﻴﺎﺭ ﻧﺰﺩﻳﻚ ﺍﻧﺘﺨﺎﺏ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﻳﻚ ﺧﺼﻮﺻﻴﺖ ﻣﻨﻔﻲ ﻭ‬
‫ﻧﺎﺍﻣﻴﺪﻛﻨﻨﺪﻩ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻭ ﻛﺎﺭﺍﻳﻲ ﻛﻪ ﻗﺎﺑﻠﻴﺖ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺯﻳﺮ ﺳﻮﺍﻝ ﻣﻲﺑﺮﺩ‪ .‬ﺍﻣﺎ ﺑﺎ ﻭﺟﻮﺩ ﺍﻳﻦ‪ ،‬ﺭﻭﺵﻫﺎﻱ‬
‫ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺟﺎﻳﮕﺎﻩ ﺧﻮﺩ ﺩﺭ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺭﺍ ﺩﺍﺭﻧﺪ ﻭ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻣﻮﺍﺭﺩ‪ ،‬ﺑﺎ ﻣﻮﻓﻘﻴﺖ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﻧﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﻫﻤﺎﻥﻃﻮﺭ ﻛﻪ‬
‫ﺫﻛﺮ ﺷﺪ‪ ،‬ﺩﺭ ﻣﻮﺍﺭﺩﻱ ﻣﻲﺗﻮﺍﻥ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺗﻀﻤﻴﻦ ﻧﻤﻮﺩ‪.‬‬

‫ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﺎﻁ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ﻛﺎﺭﺑﺮﺩ ﺩﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺍﺯ ﺍﻳﻦ ﺭﻭﺵ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ‬
‫ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪﻫﺎﻱ ﻳﻚ ﺗﺎﺑﻊ ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺑﺎﻳﺪ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪ f (x) = 0‬ﺭﺍ ﺑﻪ ﻳﻚ ﻣﺴﺎﻟﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‬
‫)‪ x = g(x‬ﺗﺒﺪﻳﻞ ﻛﺮﺩﻩ ﻭ ﺳﭙﺲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺍﻋﻤﺎﻝ ﻛﺮﺩﻩ ﺗﺎ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ g‬ﻭ ﻳﺎ ﻫﻤﺎﻥ ﺭﻳﺸﻪ ‪ f‬ﺍﺳﺘﺨﺮﺍﺝ ﮔﺮﺩﺩ‪.‬‬
‫ﺍﻟﺒﺘﻪ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻓﺮﻡ ﻣﺘﻔﺎﻭﺕ ﺑﻪ ﻣﺴﺎﻟﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ ﻭ ﺳﭙﺲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬
‫ﺭﺍ ﺭﻭﻱ ﻫﺮ ﻳﻚ ﺍﺯ ﻣﺴﺎﻳﻞ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪ f (x) = cos(x) − x‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﻣﺴﺎﻟﻪﻱ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﻮﺍﺑﻊ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪:‬‬
‫· · · ‪g1 (x) = cos(x), g2 (x) = 2x − cos(x), g3 (x) = Arccos(x),‬‬
‫ﻭ ﻣﻲﺗﻮﺍﻥ ﻫﺮ ﻳﻚ ﺍﺯ ﺗﻮﺍﺑﻊ ﻓﻮﻕ ﺭﺍ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ‪ f‬ﺑﻪ ﻛﺎﺭ ﮔﺮﻓﺖ‪ .‬ﺍﻣﺎ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﺎ ﺑﺮﺧﻲ ﺍﺯ ﺗﻮﺍﺑﻊ ﻓﻮﻕ‬
‫ﻫﻤﮕﺮﺍﻳﻲ ﻭ ﺑﺮﺧﻲ ﺩﻳﮕﺮ ﻭﺍﮔﺮﺍﻳﻲ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ‪ .‬ﺑﺮﺍﻱ ﺗﻮﺿﻴﺢ ﺑﻴﺸﺘﺮ ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۱۴−۵‬ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺭﯾﺸﻪ ﯾ‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f (x) = x3 + 4x2 − 10‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [1, 2‬ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺎ ﺭﺳﻢ ﺷﻜﻞ ﻣﺘﻮﺟﻪ ﻣﻲﺷﻮﻳﻢ‬
‫ﻛﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺭ ﺑﺎﺯﻩ ﺩﺍﺩﻩ ﺷﺪﻩ‪ ،‬ﺩﻗﻴﻘﺎ ﻳﻚ ﺭﻳﺸﻪ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﺣﻞ ﺍﻳﻦ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺍﻧﺘﺨﺎﺏﻫﺎﻱ ﺯﻳﺮ ﺭﺍ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۵۰‬‬

‫ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪.‬‬

‫‪g1 (x) = x − x − 4x + 10,‬‬
‫‪3‬‬ ‫‪2‬‬
‫)‪(۸.۵‬‬
‫‪r‬‬
‫‪10‬‬
‫= )‪g2 (x‬‬ ‫‪− 4x,‬‬ ‫)‪(۹.۵‬‬
‫‪x‬‬
‫‪1p‬‬
‫= )‪g3 (x‬‬ ‫‪10 − x3 ,‬‬ ‫)‪(۱۰.۵‬‬
‫‪2‬‬
‫‪r‬‬
‫‪10‬‬
‫= )‪g4 (x‬‬ ‫‪,‬‬ ‫)‪(۱۱.۵‬‬
‫‪4+x‬‬
‫‪x3 + 4x2 − 10‬‬
‫‪g5 (x) = x −‬‬ ‫‪.‬‬ ‫)‪(۱۲.۵‬‬
‫‪3x2 + 8x‬‬
‫ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻪﺍﺯﺍﻱ ﻫﺮﻛﺪﺍﻡ ﺍﺯ ﺗﻮﺍﺑﻊ ﻓﻮﻕ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪k‬‬ ‫‪g1‬‬ ‫‪g2‬‬ ‫‪g3‬‬ ‫‪g4‬‬ ‫‪g5‬‬
‫‪0‬‬ ‫‪1.5‬‬ ‫‪1.5‬‬ ‫‪1.5‬‬ ‫‪1.5‬‬ ‫‪1.5‬‬
‫‪1‬‬ ‫‪-0.875‬‬ ‫‪0.8165‬‬ ‫‪1.286953768‬‬ ‫‪1.348399725‬‬ ‫‪1.373333333‬‬
‫‪2‬‬ ‫‪6.732‬‬ ‫‪2.9969‬‬
‫√‬ ‫‪1.402540804‬‬ ‫‪1.367376372‬‬ ‫‪1.365262015‬‬
‫‪3‬‬ ‫‪-469.7‬‬ ‫‪−8.65‬‬ ‫‪1.345458374‬‬ ‫‪1.364957015‬‬ ‫‪1.365230014‬‬
‫‪4‬‬ ‫‪1.0 e108‬‬ ‫‪1.375170253‬‬ ‫‪1.365264748‬‬ ‫‪1.365230013‬‬
‫‪5‬‬ ‫‪-Inf‬‬ ‫‪1.360094193‬‬ ‫‪1.365225594‬‬
‫‪6‬‬ ‫‪1.367846968‬‬ ‫‪1.365230576‬‬
‫‪7‬‬ ‫‪1.363887004‬‬ ‫‪1.365229942‬‬
‫‪8‬‬ ‫‪1.365916734‬‬ ‫‪1.365230022‬‬
‫‪9‬‬ ‫‪1.364878217‬‬ ‫‪1.365230012‬‬
‫‪10‬‬ ‫‪1.365410062‬‬ ‫‪1.365230014‬‬
‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬

‫‪15‬‬ ‫‪1.365223680‬‬ ‫‪1.365230013‬‬

‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬

‫‪20‬‬ ‫‪1.365230236‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬

‫‪25‬‬ ‫‪1.365230006‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬

‫‪30‬‬ ‫‪1.365230013‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﺍﻧﺘﺨﺎﺏ ‪ g1‬ﻣﻨﺠﺮ ﺑﻪ ﻭﺍﮔﺮﺍﻳﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻧﺘﺨﺎﺏ ‪ g2‬ﻣﻨﺠﺮ ﺑﻪ ﺷﻜﺴﺖ ﻣﻲﺷﻮﺩ‪ .‬ﻭﻟﻲ ﺍﻧﺘﺨﺎﺏﻫﺎﻱ ‪g4 ،g3‬‬
‫ﻭ ‪ g5‬ﻣﻨﺠﺮ ﺑﻪ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻭ ﺍﻟﺒﺘﻪ ‪ g5‬ﺑﺎ ﺗﻜﺮﺍﺭﻫﺎﻱ ﻛﻤﺘﺮﻱ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﺍﺻﻄﻼﺡ ﻣﻲﮔﻮﻳﻴﻢ ﻛﻪ ﺳﺮﻋﺖ‬
‫ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺎ ‪ g5‬ﺑﻴﺸﺘﺮ ﺍﺳﺖ‪.‬‬

‫ﺗﺒﺪﻳﻞ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺑﻪ ﻣﺴﺎﻟﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ‪ ،‬ﻛﺎﺭ ﭘﻴﭽﻴﺪﻩﺍﻱ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺍﻣﺎ ﺁﻥ ﭼﻪ ﻣﻬﻢ ﺍﺳﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺗﺒﺪﻳﻞ ﺑﻪ‬
‫ﮔﻮﻧﻪﺍﻱ ﺑﺎﺷﺪ ﻛﻪ ﺑﺎ ﺍﻋﻤﺎﻝ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﻫﻤﮕﺮﺍﻳﻲ ﺍﺗﻔﺎﻕ ﺑﻴﺎﻓﺘﺪ ﻭ ﺑﺎ ﺗﻌﺪﺍﺩ ﺗﻜﺮﺍﺭ ﻛﻤﻲ ﺑﻪ ﺟﻮﺍﺏ ﻣﻄﻠﻮﺏ ﺑﺮﺳﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ‬
‫ﺭﺍﺳﺘﺎ ﺑﺎﻳﺪ ﺳﻌﻲ ﻛﻨﻴﻢ‪ ،‬ﺁﻥ ﻣﺴﺎﻟﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺘﻲ ﺭﺍ ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ ﻛﻪ ﺷﺮﺍﻳﻂ ﻧﺘﻴﺠﻪ ‪ ۱۱-۵‬ﺭﺍ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﻛﻪ ﺍﻟﺒﺘﻪ ﻛﺎﺭ ﭼﻨﺪﺍﻥ ﺳﺎﺩﻩ‬
‫ﻭ ﺳﺮﺭﺍﺳﺘﻲ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵۱‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬ ‫‪۵.۵‬‬

‫ﺭﻭﺵ ﻧﻴﻮﺗﻮﻥ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ )ﻋﺪﺩﻱ( ﺑﺮﺍﻱ ﺣﻞ ﻣﺴﺎﻟﻪ ﺭﻳﺸﻪﻳﺎﺑﻲ ‪ f (x) = 0‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0‬ﺭﺍ ﻣﻲﮔﻴﺮﺩ‬
‫ﻭ ﺑﺎ ﺭﺍﺑﻄﻪ ﺑﺎﺯﮔﺸﺘﻲ ﺯﻳﺮ‬
‫) ‪f (xk‬‬
‫‪xk+1 = xk −‬‬ ‫)‪(۱۳.۵‬‬
‫) ‪f ′ (xk‬‬
‫ﺩﻧﺒﺎﻟﻪﺍﻱ ﺗﻮﻟﻴﺪ ﻛﺮﺩ ﻭ ﺩﺭﺻﻮﺭﺗﻲ ﻛﻪ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻫﻤﮕﺮﺍ ﮔﺮﺩﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ ﻳﻚ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺭﺍ ﺍﺯ ﻳﻚ ﺩﻳﺪﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ ﻛﻪ ﺑﺮﺍﻱ ﺣﻞ ﻣﺴﺎﻟﻪ ‪ ،f (x) = 0‬ﺗﺎﺑﻊ )‪g(x‬‬
‫ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻌﺮﻓﻲ ﻣﻲﻛﻨﺪ‬
‫)‪f (x‬‬
‫‪g(x) = x −‬‬ ‫‪.‬‬
‫)‪f ′ (x‬‬

‫ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ‪ ،‬ﻓﺮﻣﻮﻝ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺍﺯ ﻛﺠﺎ ﺁﻣﺪﻩ ﺍﺳﺖ ﻭ ﭼﺮﺍ ﺗﺎﺑﻊ ‪ g‬ﺩﺭ ﻣﺴﺎﻟﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﻓﻮﻕ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ؟ ﺩﺭ ﻭﺍﻗﻊ ﺍﺯ ﺩﻳﺪﮔﺎﻩﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ ﺭﺳﻴﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﻳﻚ ﺩﻳﺪﮔﺎﻩ‬
‫ﺍﺷﺎﺭﻩ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺑﻪ ﻧﻮﻋﻲ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻧﻴﺰ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺍﺑﺘﺪﺍ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﺍﺧﺘﻴﺎﺭ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺧﻂ ﻣﻤﺎﺱ ﺑﺮ ﺗﺎﺑﻊ ﺩﺭ ﺍﻳﻦ ﻧﻘﻄﻪ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺩﺭ ﻧﻈﺮ‬
‫ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﺭﻳﺸﻪﻱ ﺍﻳﻦ ﺧﻂ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺑﻬﺘﺮ ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻓﺮﺍﻳﻨﺪ ﺭﺍ ﺍﺩﺍﻣﻪ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﺗﺎ‬
‫ﺗﻘﺮﻳﺐﻫﺎﻱ ﺑﻬﺘﺮﻱ ﺑﺮﺍﻱ ﺭﻳﺸﻪ ﺣﺎﺻﻞ ﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺍﻳﺪﻩ ﺩﻗﻴﻖﺗﺮ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫)‪f (x‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ‪ x0‬ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﺮﺍﻱ ﺭﻳﺸﻪ )‪ f (x‬ﺑﺎﺷﺪ‪ .‬ﺧﻂ‬

‫) ‪• f (x0‬‬ ‫ﻣﻤﺎﺱ ﺑﺮ ﺗﺎﺑﻊ )‪ f (x‬ﺩﺭ ﻧﻘﻄﻪ ‪ x0‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪y = f (x0 ) + f ′ (x0 )(x − x0 ).‬‬
‫ﺍﻳﻦ ﺧﻂ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ ﻭ ﺭﻳﺸﻪ‬
‫ﺍﻳﻦ ﺧﻂ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺩﻳﮕﺮﻱ ﺍﺯ ﺭﻳﺸﻪ )‪ f (x‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪.‬‬
‫‪ x0 − ff′(x‬ﺑﻪ ﺩﺳﺖ‬ ‫ﺭﻳﺸﻪ ﺍﻳﻦ ﺧﻂ ﺑﻪ ﻃﻮﺭ ﺗﺤﻠﻴﻠﻲ ﻭ ﺑﻪ ﺻﻮﺭﺕ ) ‪(x0‬‬
‫)‪0‬‬

‫‪x‬‬
‫•‬ ‫•‬ ‫ﻣﻲﺁﻳﺪ‪ .‬ﺣﺎﻝ ﺍﻳﻦ ﻣﻘﺪﺍﺭ ﺭﺍ ‪ x1‬ﻧﺎﻣﮕﺬﺍﺭﻱ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺷﻜﻞ ﺭﻭﺑﺮﻭ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫‪x0‬‬
‫‪x1 =x0 − ff′(x‬‬ ‫)‪0‬‬
‫) ‪(x0‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۵۲‬‬

‫)‪f (x‬‬

‫ﻣﺠﺪﺩﺍ ﻓﺮﺍﻳﻨﺪ ﻓﻮﻕ ﺭﺍ ﺗﻜﺮﺍﺭ ﻣﻲﻛﻨﻴﻢ‪ ،‬ﻣﻨﺘﻬﻲ ﺍﻳﻦ ﺑﺎﺭ ﺧﻂ ﻣﻤﺎﺱ ﺑﺮ ﺗﺎﺑﻊ‬
‫ﺩﺭ ﻧﻘﻄﻪ ‪ x1‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ ﻭ ﺭﻳﺸﻪ ﺍﻳﻦ‬
‫ﺧﻂ ﺭﺍ ‪ x2‬ﻣﻲﻧﺎﻣﻴﻢ ﻭ ﺁﻥ ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺩﻳﮕﺮﻱ ﺍﺯ ﺭﻳﺸﻪ ﻣﻌﺮﻓﻲ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﻫﻤﺎﻧﻨﺪ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ‬
‫) ‪f (x1‬‬
‫‪x2 = x1 −‬‬ ‫‪.‬‬
‫•‬ ‫) ‪f ′ (x1‬‬

‫ﺑﻪ ﻫﻤﻴﻦ ﺗﺮﺗﻴﺐ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺑﺎ ﻓﺮﻣﻮﻝ ) ‪(xk‬‬

‫‪xk+1 = xk − ff′(x‬‬ ‫)‪k‬‬
‫‪x‬‬
‫•‬ ‫•‬ ‫•‬ ‫ﻣﻲﺗﻮﺍﻥ ﺗﻮﻟﻴﺪ ﻧﻤﻮﺩ‪.‬‬
‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ‪ ،‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﻳﻚ ﺗﺎﺑﻊ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﺩ‪.‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۱۵−۵‬ﺭﻭﺵ ﻧﯿﻮﺗﻦ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺭﯾﺸﻪ ﯾ‬

‫‪ f (x) := x − 0.8 sin(x) − 2π‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺭﺍ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﻴﺎﺑﻴﻢ‪.‬‬
‫ﺗﺎﺑﻊ ‪10‬‬
‫ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪xk − 0.8 sin(xk ) − 2π‬‬
‫‪xk+1 = xk −‬‬ ‫‪10‬‬
‫) ‪1 − 0.8 cos(xk‬‬
‫ﺩﺭ ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ‪ ،‬ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 10‬ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﻴﻦﻃﻮﺭ ﺩﺭ ﻫﺮ ﺗﻜﺮﺍﺭ ﻛﻤﻴﺖﻫﺎﻱ‬
‫|) ‪ |xk − xk−1 | ،|f (xk‬ﻭ |‪ |xk − α‬ﻧﻴﺰ ﮔﺰﺍﺭﺵ ﺷﺪﻩ ﺍﺳﺖ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ α‬ﺭﻳﺸﻪ ﺗﺎﺑﻊ ﺍﺳﺖ ﻛﻪ ﻣﻘﺪﺍﺭ ﺁﻥ ﺗﺎ ‪ ۱۶‬ﺭﻗﻢ ﺑﺎ‬
‫ﻣﻌﻨﺎﻱ ﺩﺭﺳﺖ ﺑﺮﺍﺑﺮ ‪ 1.419135783830583‬ﺍﺳﺖ‪ .‬ﻫﻤﭽﻨﻴﻦ‪ ،‬ﺩﺭ ﺳﺘﻮﻥ ﺳﻮﻡ ﺟﺪﻭﻝ‪ ،‬ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ‪ xk‬ﮔﺰﺍﺭﺵ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪k‬‬ ‫‪xk‬‬ ‫) ‪NCSD(xk‬‬ ‫| ‪f (xk ) |xk − xk−1‬‬ ‫|‪|xk − α‬‬
‫‪0‬‬ ‫‪10.00000000000000‬‬ ‫‪-1‬‬ ‫‪9.8 e+00‬‬ ‫‪-‬‬ ‫‪8.6 e+00‬‬
‫‪1‬‬ ‫‪4.132023352541080‬‬ ‫‪0‬‬ ‫‪4.2 e+00 5.9 e+00‬‬ ‫‪2.7 e+00‬‬
‫‪2‬‬ ‫‪1.231613517061376‬‬ ‫‪1‬‬ ‫‪1.5 e-01 2.9 e+00‬‬ ‫‪1.9 e-01‬‬
‫‪3‬‬ ‫‪1.437556559531969‬‬ ‫‪2‬‬ ‫‪1.6 e-02 2.1 e-01‬‬ ‫‪1.8 e-02‬‬
‫‪4‬‬ ‫‪1.419286179886509‬‬ ‫‪4‬‬ ‫‪1.3 e-04 1.8 e-02‬‬ ‫‪1.5 e-04‬‬
‫‪5‬‬ ‫‪1.419135794002687‬‬ ‫‪8‬‬ ‫‪8.9 e-09 1.5 e-04‬‬ ‫‪1.0 e-08‬‬
‫‪5‬‬ ‫‪1.419135783830583‬‬ ‫‪16‬‬ ‫‪6.7 e-17 6.4 e-11‬‬ ‫‪7.6 e-17‬‬
‫‪6‬‬ ‫‪1.419135783830583‬‬ ‫‪16‬‬ ‫‪6.7 e-17 0.0 e-00‬‬ ‫‪7.6 e-17‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﻫﺮ ﺗﻜﺮﺍﺭ ﺣﺪﻭﺩﺍ ﺧﻄﺎﻱ ﻣﻄﻠﻖ ﺑﻪ ﺗﻮﺍﻥ ﺩﻭ ﻣﻲﺭﺳﺪ )ﻭ ﻳﺎ ﺗﻌﺪﺍﺩ ﺍﺭﻗﺎﻡ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺩﺭ ‪ xk‬ﺣﺪﻭﺩﺍ ﺩﻭ‬
‫ﺑﺮﺍﺑﺮ ﻣﻲﺷﻮﺩ(‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺗﻨﻬﺎ ﺩﺭ ‪ ۵‬ﺗﻜﺮﺍﺭ ﺑﻪ ﺗﻘﺮﻳﺒﻲ ﻣﻲﺭﺳﻴﻢ ﻛﻪ ﺩﺍﺭﺍﻱ ‪ ۱۶‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺩﺭﺳﺖ ﺍﺳﺖ‪ .‬ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ ﻛﻪ‬
‫ﺍﻳﻦ ﻣﺤﺎﺳﺒﺎﺕ ﺩﺭ ﺩﻗﺖ ﻣﻀﺎﻋﻒ ﺍﻧﺠﺎﻡ ﺷﺪﻩ ﻭ ﺣﺪﺍﻛﺜﺮ ﺩﻗﺖ ﻣﻤﻜﻦ ﺑﺮﺍﺑﺮ ‪ ۱۶‬ﺭﻗﻢ ﺑﺎﻣﻌﻨﻲ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺍﺩﺍﻣﻪ ﺗﻜﺮﺍﺭﻫﺎ ﺩﺭ ﺩﻗﺖ‬
‫ﻣﻀﺎﻋﻒ ﻣﻨﺠﺮ ﺑﻪ ﺟﻮﺍﺑﻲ ﺩﻗﻴﻖﺗﺮ ﻧﺨﻮﺍﻫﺪ ﺷﺪ ﻭ ﻫﻤﺎﻥ ﺟﻮﺍﺏ ﻗﺒﻠﻲ ﺗﻜﺮﺍﺭ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﻧﻜﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺍﺭﺍﻳﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵۳‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫• ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﻣﺜﺎﻝ ﻓﻮﻕ ﺩﻳﺪﻳﻢ‪ ،‬ﺳﺮﻋﺖ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺴﻴﺎﺭ ﺑﺎﻻ ﺍﺳﺖ ﻭ ﺑﺎ ﺗﻌﺪﺍﺩ ﺗﻜﺮﺍﺭﻫﺎﻱ ﻧﺎﭼﻴﺰﻱ ﺑﻪ ﺟﻮﺍﺏ ﻣﻲﺭﺳﺪ‪.‬‬
‫ﺩﺭ ﻣﻮﺭﺩ ﺍﻳﻦ ﺧﺼﻮﺻﻴﺖ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺩﺭ ﺑﺨﺶ ﺑﻌﺪ ﺑﻴﺸﺘﺮ ﺻﺤﺒﺖ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﻫﻤﻴﻦ ﺟﺎ ﺑﻪ ﺍﻳﻦ ﻧﻜﺘﻪ ﺍﺷﺎﺭﻩ ﻣﻲﺷﻮﺩ‬
‫ﻛﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺳﺮﻳﻊ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺧﺎﺹ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ‪ ،‬ﻋﻼﻭﻩ ﺑﺮ ﺍﻃﻼﻋﺎﺕ ﺧﻮﺩ‬
‫ﺗﺎﺑﻊ ﺍﺯ ﺍﻃﻼﻋﺎﺕ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ ﺍﻭﻝ ﻧﻴﺰ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﻧﻜﺘﻪ ﺩﻟﻴﻞ ﺳﺮﻋﺖ ﺑﺎﻻﻱ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺍﺳﺖ‪ .‬ﺑﻊ ﻋﺒﺎﺭﺕ‬
‫ﺩﻳﮕﺮ‪ ،‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﻳﻊ ﺧﻮﺩ ﺭﺍ ﻣﺪﻳﻮﻥ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺍﻃﻼﻋﺎﺕ ﻣﺸﺘﻖ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﻪ‬
‫ﻛﺎﺭﮔﻴﺮﻱ ﻣﺸﺘﻖ ﺗﺎﺑﻊ ﻣﺴﺘﻠﺰﻡ ﺍﺳﺘﺨﺮﺍﺝ ﻣﺸﺘﻖ ﺍﺳﺖ ﻛﻪ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻊ ﺑﺎ ﺿﺎﺑﻄﻪ ﺑﺰﺭﮒ ﻭ ﭘﻴﭽﻴﺪﻩ ﺩﺭﺩﺳﺮﺳﺎﺯ ﺍﺳﺖ‪.‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬ ‫�‬

‫ﺩﻳﺪﻳﻢ ﻛﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﻳﻚ ﺧﺎﺻﻴﺖ ﻣﻄﻠﻮﺏ ﺑﺮﺍﻱ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﻮﺩ ﻛﻪ ﻫﻤﻮﺍﺭﻩ ﻭﺟﻮﺩ ﻧﺪﺍﺷﺖ ﻭ ﺑﺮﻗﺮﺍﺭﻱ ﺁﻥ ﻣﺴﺘﻠﺰﻡ‬
‫ﺷﺮﺍﻳﻂ ﺧﺎﺻﻲ ﺑﻮﺩ‪ .‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻧﻴﺰ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ‬

‫» ﺁﻳﺎ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺩﺍﺭﺍﻱ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺍﺳﺖ ﻳﺎ ﺧﻴﺮ؟ «‬

‫ﺟﻮﺍﺏ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻫﺎﻱ ﺳﺎﺩﻩ ﻣﺜﺒﺖ ﺍﺳﺖ‪ .‬ﻳﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﺑﺎﺯﻩﺍﻱ ﺣﻮﻝ ﻫﺮ ﺭﻳﺸﻪ ﺳﺎﺩﻩ ﺗﺎﺑﻊ ‪ f‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﮔﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ‬
‫ﺭﺍ ﺍﺯ ﺁﻥ ﺑﺎﺯﻩ ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﻪ ﺁﻥ ﺭﻳﺸﻪ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺑﺮﺍﻱ ﺍﺛﺒﺎﺕ ﻫﻤﮕﺮﺍﻱ ﻣﻮﺿﻌﻲ ﺑﻮﺩﻥ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ α‬ﻳﻚ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪f‬ﺑﺎﺷﺪ‪ .‬ﺗﻮﺟﻪ ﺩﺍﺭﻳﻢ ﻛﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬
‫ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪﻫﺎﻱ ﺗﺎﺑﻊ ‪ ،f‬ﻫﻤﺎﻥ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ‪ g‬ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫)‪f (x‬‬
‫‪g(x) := x −‬‬
‫)‪f ′ (x‬‬
‫ﺣﺎﻝ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪′‬‬ ‫‪2‬‬ ‫‪′′‬‬ ‫‪′′‬‬
‫)‪(f (x)) − f (x)f (x‬‬ ‫)‪f (x)f (x‬‬
‫‪g ′ (x) = 1 −‬‬ ‫‪2‬‬ ‫=‬ ‫‪2‬‬
‫))‪(f ′ (x‬‬ ‫))‪(f ′ (x‬‬
‫ﻭ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ‪ g ′ (α) = 0‬ﻭ ﻟﺬﺍ ‪ .|g (α)| < 1‬ﺣﺎﻝ ﻃﺒﻖ ﻗﻀﻴﻪ ‪ ۱۰-۵‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺩﺍﺭﺍﻱ‬
‫‪′‬‬

‫ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺩﺭ ‪ α‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻋﻼﻭﻩ ﺑﺮ ﺧﺎﺻﻴﺖ ﻫﻤﮕﺮﺍﻳﻲ ﺳﺮﻳﻊ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺧﺎﺻﻴﺖ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻮﺿﻌﻲ ﺭﺍ ﺟﺰﻭ ﻣﺰﺍﻳﺎﻱ‬
‫ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺮﺷﻤﺮﺩ‪.‬‬
‫ﺩﺭﺳﺖ ﺍﺳﺖ ﻛﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﻪ ﺻﻮﺭﺕ ﻣﻮﺿﻌﻲ ﺍﺳﺖ ﻭ ﻳﻚ ﺑﺎﺯﻩ ﺍﻃﺮﺍﻑ ﺭﻳﺸﻪ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺑﺎ ﻫﺮ‬
‫ﻧﻘﻄﻪ ﺷﺮﻭﻉ ﺍﺯ ﺁﻥ ﺑﺎﺯﻩ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ .‬ﺍﻣﺎ ﻋﻤﻮﻣﺎ ﺍﻳﻦ ﺑﺎﺯﻩ ﻛﻮﭼﻚ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺍﮔﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻛﺎﻓﻲ ﻧﺰﺩﻳﻚ ﺑﻪ ﺭﻳﺸﻪ‬
‫ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﺣﺘﻤﺎﻝ ﻭﺍﮔﺮﺍﻳﻲ ﺭﻭﺵ ﺑﺴﻴﺎﺭ ﺑﺎﻻ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﺻﻄﻼﺣ ًﺎ ﮔﻔﺘﻪ ﻣﻲﺷﻮﺩ ﻛﻪ‪ :‬ﺭﻭﺵ ﻧﻴﻮﺗﻮﻥ ﺑﻪ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺣﺴﺎﺱ ﺍﺳﺖ‪.‬‬
‫ﺍﻳﻦ ﻧﻜﺘﻪ ﺍﻳﺮﺍﺩ ﺍﺳﺎﺳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺍﻣﺎ ﺭﻭﺵ ﻧﻴﻮﺗﻮﻥ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﻫﻤﮕﺮﺍﻳﻲ ﺗﻀﻤﻴﻦ ﺷﺪﻩ ﺩﺍﺭﺩ ﻭ ﻳﺎ ﺑﺎﺯﻩ ﻫﻤﮕﺮﺍﻳﻲ ﺁﻥ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ﻛﺎﻓﻲ ﺑﺰﺭﮒ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ‬
‫ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ ﺩﻏﺪﻏﻪ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ ﻭ ﻣﻲﺗﻮﺍﻧﻴﻢ ﺑﺎ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﺩﻟﺨﻮﺍﻩ‪ ،‬ﺑﺎ ﺳﺮﻋﺖ ﻣﻨﺎﺳﺒﻲ ﻳﻚ ﺗﻘﺮﻳﺐ ﺩﻗﻴﻖ ﺍﺯ‬
‫ﺭﻳﺸﻪ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻤﺎﻳﻴﻢ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺩﻭ ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﻋﺪﺩ(‬ ‫ﻣﺜﺎﻝ ‪)۱۶−۵‬ﺭﻭﺵ ﻧﯿﻮﺗﻦ ﺑﺮﺍﯼ ﺗﻘﺮﯾﺐ ﺟﺬﺭ ﯾ‬

‫√‬
‫ﺭﻳﺸﻪ ﻣﺜﺒﺖ ﺗﺎﺑﻊ ‪f (x) = x2 − A‬‬ ‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺟﺬﺭ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﻭ ﻣﺜﺒﺖ ‪ A‬ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ‪A‬‬
‫√‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺗﻘﺮﻳﺐ ‪ A‬ﻣﻌﺎﺩﻝ ﻳﺎﻓﺘﻦ ﺗﻘﺮﻳﺒﻲ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺍﻳﻦ ﺭﻳﺸﻪ ﺍﺯ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪،‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۵۴‬‬

‫ﺑﻪ ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺯﻳﺮ ﻣﻲﺭﺳﻴﻢ‬

‫‪2‬‬
‫‪(xk ) − A‬‬
‫‪xk+1 = xk −‬‬
‫‪2xk‬‬
‫‬ ‫‬
‫‪1‬‬ ‫‪A‬‬
‫=‬ ‫‪xk +‬‬
‫‪2‬‬ ‫‪xk‬‬
‫√‬
‫ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻣﺜﺒﺖ‪ ،‬ﺩﻧﺒﺎﻟﻪ ﻓﻮﻕ ﺑﺎ ﺳﺮﻋﺖ ﺑﻪ ‪ A‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ‬
‫√‬
‫ﺍﻭﻟﻴﻪ ﻣﻨﻔﻲ‪ ،‬ﺩﻧﺒﺎﻟﻪ ﺑﻪ ‪ − A‬ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬
‫√‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﺭﻭﺵ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ A‬ﺍﺭﺍﻳﻪ ﻣﻲﻛﻨﺪ‪.‬‬

‫ﻋﺪﺩ(‬ ‫ﻣﺜﺎﻝ ‪)۱۷−۵‬ﺭﻭﺵ ﻧﯿﻮﺗﻦ ﺑﺮﺍﯼ ﺗﻘﺮﯾﺐ ﻋﮑﺲ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ A ̸= 0‬ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻋﻜﺲ ‪ A‬ﻳﻌﻨﻲ ‪ A1‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ ﻭ ﺍﻟﺒﺘﻪ ﻧﻤﻲﺧﻮﺍﻫﻴﻢ ﺍﺯ ﻋﻤﻞ ﺗﻘﺴﻴﻢ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪.‬‬
‫ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ‪ A1‬ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f (x) = x1 − A‬ﻣﻲﺑﺎﺷﺪ ﻟﺬﺍ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ ، A1‬ﻣﻲﺗﻮﺍﻥ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺯﺩ‪ .‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬
‫ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺍﻳﻦ ﺭﻳﺸﻪ ﺑﻪ ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺯﻳﺮ ﻣﻨﺘﻬﻲ ﻣﻲﺷﻮﺩ‬
‫‪1‬‬
‫‪xk‬‬ ‫‪−A‬‬
‫‪xk+1 = xk −‬‬ ‫‪−1‬‬
‫‪2xk‬‬
‫‪2‬‬
‫) ‪= 2xk − A (xk‬‬
‫ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪﺍﺯﺍﻱ ﻫﺮ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻣﺜﺒﺖ‪ ،‬ﺩﻧﺒﺎﻟﻪ ﻓﻮﻕ ﺑﺎ ﺳﺮﻋﺖ ﺑﻪ ‪ A1‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﺭﻭﺵ‬
‫ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ A1‬ﺍﺭﺍﻳﻪ ﻣﻲﻛﻨﺪ ﻛﻪ ﺩﺭ ﺁﻥ ﺍﺯ ﻋﻤﻞ ﺗﻘﺴﻴﻢ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻲﮔﺮﺩﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻲﺗﻮﺍﻥ ﻋﻤﻞ ﺗﻘﺴﻴﻢ ﺭﺍ ﺑﺎ ﻋﻤﻠﻴﺎﺕ ﺟﻤﻊ‪،‬‬
‫ﺗﻔﺮﻳﻖ ﻭ ﺿﺮﺏ ﺗﻘﺮﻳﺐ ﺯﺩ‪.‬‬

‫ﺭﻭﺵ ﻭﺗﺮﻱ‬ ‫‪۶.۵‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺎ ﺑﻬﺮﻩ ﮔﺮﻓﺘﻦ ﺍﺯ ﺍﻃﻼﻋﺎﺕ ﻣﺸﺘﻖ ﺗﺎﺑﻊ‪ ،‬ﺗﻮﺍﻧﺴﺖ ﺑﺎ ﺳﺮﻋﺖ ﺑﻴﺸﺘﺮﻱ ﺭﻳﺸﻪﻫﺎﻱ‬
‫ﺳﺎﺩﻩ ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﺑﻪ ﻣﺸﺘﻖ ﺗﺎﺑﻊ ﺩﺳﺘﺮﺳﻲ ﻧﺪﺍﺭﻳﻢ ﻭ ﻳﺎ ﺍﻳﻦ ﻛﻪ ﻓﺮﺍﻫﻢ ﻛﺮﺩﻥ ﻣﺸﺘﻖ ﺗﺎﺑﻊ‬
‫ﻣﺸﻜﻞ ﻭ ﭘﺮﻫﺰﻳﻨﻪ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﮔﻮﻧﻪ ﻣﻮﺍﺭﺩ ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺐ ﺑﺮﺍﻱ ﻣﺸﺘﻖ ﺗﺎﺑﻊ ﻳﺎﻓﺘﻪ ﺷﻮﺩ ﻭ ﺩﺭ ﻓﺮﻣﻮﻝ‬
‫ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺟﺎﻳﮕﺰﺍﺭﻱ ﺷﻮﺩ‪ .‬ﺭﻭﺵ ﻭﺗﺮﻱ ‪ ۲۵‬ﺭﻭﺷﻲ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺍﻳﻦ ﺷﻴﻮﻩ ﺍﺑﺪﺍﻉ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺭﻭﺵ ﻭﺗﺮﻱ‪ ،‬ﻣﺸﺘﻖ‬
‫)ﺧﻂ ﻣﻤﺎﺱ( ﺩﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺭﺍ ﺑﺎ ﺧﻄﻲ ﻛﻪ ﺍﺯ ﺩﻭ ﻧﻘﻄﻪ ﻗﺒﻠﻲ ﻣﻲﮔﺬﺭﺩ‪ ،‬ﺟﺎﻳﮕﺬﻳﻦ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ‪ ،‬ﺭﻭﺵ ﻭﺗﺮﻱ‬
‫) ‪ f ′ (xk‬ﺩﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺭﺍ ﺑﺎ‬
‫) ‪f (xk ) − f (xk−1‬‬
‫‪xk − xk−1‬‬
‫ﺟﺎﻳﮕﺰﻳﻦ ﻣﻲﻛﻨﺪ ﻭ ﻓﺮﻣﻮﻝ ﺗﻜﺮﺍﺭﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‬
‫
‬
‫‪f (xk ) xk − xk−1‬‬
‫‪xk+1‬‬ ‫‪= xk −‬‬ ‫)‪(۱۴.۵‬‬
‫‪f (xk ) − f (xk−1‬‬
‫ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺩﺭ ﺭﻭﺵ ﻭﺗﺮﻱ )ﺑﺮﺧﻼﻑ ﺭﻭﺵﻫﺎﻳﻲ ﻛﻪ ﺗﺎ ﻛﻨﻮﻥ ﻣﻄﺮﺡ ﺷﺪﻩﺍﻧﺪ( ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ xk+1‬ﺍﺯ ﺩﻭ‬
‫ﻣﻘﺪﺍﺭ ﻗﺒﻞ ﻳﻌﻨﻲ ‪ xk‬ﻭ ‪ xk−1‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺗﻲ‪ ،‬ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﻭﺗﺮﻱ‪ ،‬ﻳﻚ ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺩﻭ ﺑﺎﺯﮔﺸﺘﻲ ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺭﻭﺵ ﻭﺗﺮﻱ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪25 Secant‬‬ ‫‪mrthod‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵۵‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۱۸−۵‬ﺭﻭﺵ ﻭﺗﺮﯼ ﺑﺮﺍﯼ ﯾﺎﻓﺘﻦ ﺭﯾﺸﻪ ﯾ‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﺭﻭﺵ ﻭﺗﺮﻱ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪ f (x) := x3 +x−1‬ﺑﺎ ﺣﺪﺱﻫﺎﻱ‬
‫‪k‬‬ ‫‪xk‬‬ ‫ﺍﻭﻟﻴﻪ ‪ x0 = 0‬ﻭ ‪ x1 = 1‬ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻣﻌﺎﺩﻟﻪ ﺗﻜﺮﺍﺭ ﺭﻭﺵ ﻭﺗﺮﻱ ﺑﺮﺍﻱ ﺍﻳﻦ‬
‫‪-1‬‬ ‫‪0.00000000000000‬‬ ‫ﻣﺜﺎﻝ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪0‬‬ ‫‪1.00000000000000‬‬ ‫) ‪(x3k + xk − 1)(xk − xk−1‬‬
‫‪xk+1 = xk −‬‬
‫‪1‬‬ ‫‪0.50000000000000‬‬ ‫) ‪x3k + xk − (x3k−1 + xk−1‬‬
‫‪2‬‬ ‫‪0.63636363636364‬‬ ‫ﺣﺎﻝ ﺑﺎ ﺷﺮﻭﻉ ﺍﺯ ﻧﻘﺎﻁ ﺍﻭﻟﻴﻪ ‪ x0 = 0‬ﻭ ‪ ،x1 = 1‬ﺗﻘﺮﻳﺐﻫﺎﻱ ‪ x2‬ﻭ ‪ x3‬ﺑﻪ ﺻﻮﺭﺕ‬
‫‪3‬‬ ‫‪0.69005235602094‬‬ ‫ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪4‬‬ ‫‪0.68202041964819‬‬ ‫)‪(1 + 1 − 1)(1 − 0‬‬ ‫‪1‬‬
‫‪x2 = 1 −‬‬ ‫‪= = 0.5,‬‬
‫‪5‬‬ ‫‪0.68232578140989‬‬ ‫)‪1 + 1 − (0 + 0‬‬ ‫‪2‬‬
‫
‬
‫‪6‬‬ ‫‪0.68232780435903‬‬ ‫)‪0.53 + 0.5 − 1 (0.5 − 1‬‬
‫‪x3 = 0.5 −‬‬ ‫‪= 0.63636363636364.‬‬
‫‪7‬‬ ‫‪0.68232780382802‬‬ ‫)‪0.53 + 0.5 − (1 + 1‬‬
‫ﺑﻪ ﻫﻤﻴﻦ ﺗﺮﺗﻴﺐ ﻣﻲﺗﻮﺍﻥ ﺑﻘﻴﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺍﻳﺠﺎﺩ ﻛﺮﺩ‪ .‬ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ‬
‫‪8‬‬ ‫‪0.68232780382802‬‬
‫ﻭﺗﺮﻱ ﺩﺭ ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ ﺧﻼﺻﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۵۶‬‬

‫ﻣﺜﺎﻝ ‪)۱۹−۵‬ﻣﻘﺎﯾﺴەﺍﯼ ﺑﯿﻦ ﺭﻭﺵ ﻭﺗﺮﯼ ﻭ ﺭﻭﺵ ﻧﯿﻮﺗﻦ(‬

‫ﺩﺭ ﻣﺜﺎﻝ ‪ ،۱۵-۵‬ﺩﻳﺪﻳﻢ ﻛﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0 = 10‬ﻭ ﺑﺎ ‪ ۶‬ﺗﻜﺮﺍﺭ‪ ،‬ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺭﻳﺸﻪ ﺗﺎﺑﻊ ‪f (x) := x −‬‬
‫‪ 0.8 sin(x) − 2π‬ﺑﺎ ﺩﻗﺖ ‪ 6.7 e − 17‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺭﻳﺸﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺭﺍ ﺑﺎ ﺭﻭﺵ ﻭﺗﺮﻱ ﺑﻴﺎﺑﻴﻢ‪.‬‬‫‪10‬‬
‫ﺩﺭ ﺭﻭﺵ ﻭﺗﺮﻱ ﺑﻪ ﺩﻭ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ ﻛﻪ ﺍﻳﻦ ﺣﺪﺱﻫﺎ ﺭﺍ ‪ x0 = 10‬ﻭ ‪ x1 = 9‬ﺍﺧﺘﻴﺎﺭ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻧﺘﺎﻳﺞ ﺭﻭﺵ ﻭﺗﺮﻱ‬
‫ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪k‬‬ ‫‪xk‬‬ ‫) ‪NCSD(xk‬‬ ‫| ‪f (xk ) |xk − xk−1‬‬ ‫|‪|xk − α‬‬

‫‪-1‬‬ ‫‪10.00000000000000‬‬ ‫‪-‬‬ ‫‪9.8 e+00‬‬ ‫‪-‬‬ ‫‪8.6 e+00‬‬
‫‪0‬‬ ‫‪9.000000000000000‬‬ ‫‪-‬‬ ‫‪8.0 e+00‬‬ ‫‪-‬‬ ‫‪7.6 e+00‬‬
‫‪1‬‬ ‫‪4.443405590022644‬‬ ‫‪0‬‬ ‫‪4.6 e+00 4.6 e+00‬‬ ‫‪3.0 e+00‬‬
‫‪2‬‬ ‫‪-1.604053630441728‬‬ ‫‪0‬‬ ‫‪1.4 e+00 6.0 e+00‬‬ ‫‪3.0 e+00‬‬
‫‪3‬‬ ‫‪-0.164496664293685‬‬ ‫‪0‬‬ ‫‪6.6 e-01 1.4 e+00‬‬ ‫‪1.6 e+00‬‬
‫‪4‬‬ ‫‪1.071183035085434‬‬ ‫‪1‬‬ ‫‪2.6 e-01 1.2 e+00‬‬ ‫‪3.5 e-01‬‬
‫‪5‬‬ ‫‪1.867467661217371‬‬ ‫‪1‬‬ ‫‪4.7 e-01 8.0 e-01‬‬ ‫‪4.5 e-01‬‬
‫‪6‬‬ ‫‪1.352752469944594‬‬ ‫‪2‬‬ ‫‪5.7 e-02 5.1 e-01‬‬ ‫‪6.6 e-02‬‬
‫‪7‬‬ ‫‪1.407668777555117‬‬ ‫‪2‬‬ ‫‪1.0 e-02 5.5 e-02‬‬ ‫‪1.1 e-02‬‬
‫‪8‬‬ ‫‪1.419488966717839‬‬ ‫‪4‬‬ ‫‪3.1 e-04 1.2 e-02‬‬ ‫‪3.5 e-04‬‬
‫‪9‬‬ ‫‪1.419133954196839‬‬ ‫‪6‬‬ ‫‪1.6 e-06 3.6 e-04‬‬ ‫‪1.8 e-06‬‬
‫‪10‬‬ ‫‪1.419135783539985‬‬ ‫‪10‬‬ ‫‪2.6 e-10 1.8 e-06‬‬ ‫‪2.9 e-10‬‬
‫‪11‬‬ ‫‪1.419135783830583‬‬ ‫‪16‬‬ ‫‪2.2 e-16 2.9 e-10‬‬ ‫‪2.2 e-16‬‬
‫‪12‬‬ ‫‪1.419135783830583‬‬ ‫‪16‬‬ ‫‪2.2 e-16 0.0 e+00‬‬ ‫‪2.2 e-16‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﺭﻭﺵ ﻭﺗﺮﻱ ﺑﺎ ‪ ۱۱‬ﺗﻜﺮﺍﺭ ﺟﻮﺍﺑﻲ ﺑﺎ ﺩﻗﺖ ‪ 2.2e−16‬ﺭﺍ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﺩ ﻛﻪ ‪ ۶‬ﺗﻜﺮﺍﺭ ﺑﻴﺸﺘﺮ ﺍﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬
‫ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﻧﻜﺘﻪ ﺍﺯ ﻗﺒﻞ ﻫﻢ ﻗﺎﺑﻞ ﺣﺪﺱ ﺑﻮﺩ‪.‬‬

‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫‪۷.۵‬‬

‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺍﺑﺰﺍﺭﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺑﺮﺍﻱ ﺑﺮﺭﺳﻲ ﻭ ﻣﻘﺎﻳﺴﻪ ﺳﺮﻋﺖ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺍﺭﺍﻳﻪ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺍﺑﺘﺪﺍ‬
‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺗﻌﺮﻳﻒ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺩﻧﺒﺎﻟﻪ‬ ‫ﺗﻌﺮﯾﻒ ‪۱۲−۵‬‬

‫ﻣﺮﺗﺒﻪ ﻫﻤ ﺮﺍﯾﯽ ﯾ‬
‫‬ ‫∞‬
‫= ‪ .xn‬ﮔﻮﻳﻴﻢ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﻓﻮﻕ ﺑﺮﺍﺑﺮ‬
‫ﻓﺮﺽ ﻛﻨﻴﻢ ‪ xk k=1‬ﺩﻧﺒﺎﻟﻪﺍﻱ ﺑﺎﺷﺪ ﻛﻪ ﺑﻪ ‪ p‬ﻣﻴﻞ ﻣﻲﻛﻨﺪ ﻭ ﺩﺭ ﺿﻤﻦ ‪̸ p‬‬
‫‪ σ‬ﺍﺳﺖ‪ ،‬ﻫﺮﮔﺎﻩ ‪ λ > 0‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ ﻛﻪ‬
‫| ‪| xk+1 − p‬‬
‫‪lim‬‬ ‫‪= λ,‬‬
‫‪k→∞ | xk − p |σ‬‬

‫ﺧﺎﻃﺮ ﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺑﺎﻳﺪ ﺍﻭﻻ ﺣﺪ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺑﺪﺍﻧﻴﻢ ﻭ ﺛﺎﻧﻴﺎ ﺣﺪ ﻓﻮﻕ ﻣﻮﺟﻮﺩ‬
‫ﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﻣﻮﺭﺩ ﺍﺭﺗﺒﺎﻁ ﻣﻘﺪﺍﺭ ‪ σ‬ﻭ ‪ λ‬ﻗﻀﻴﻪ ﺯﻳﺮ ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵۷‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫ﻗﻀﯿﻪ ‪۱۳−۵‬‬

‫ﺍﮔﺮ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺑﺎ ﺗﻌﺮﻳﻒ ﻓﻮﻕ ﺩﺍﺭﺍﻱ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ‪ σ‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻟﺰﻭﻣﺎ ‪ .σ ≥ 1‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺻﻮﺭﺗﻲ ﻛﻪ ‪ σ = 1‬ﺁﻥﮔﺎﻩ‬
‫‪.λ ≤ 1‬‬

‫ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ ﻣﻔﻬﻮﻡ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﺑﻪ ﻣﺜﺎﻝ ﺯﻳﺮ ﺗﻮﺟﻪ ﻛﻨﻴﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۲۰−۵‬ﻣﺮﺗﺒﻪ ﻫﻤ ﺮﺍﯾﯽ ﺩﻭ ﺩﻧﺒﺎﻟﻪ(‬

‫∞ ‪1‬‬
‫ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﺮﺍﺑﺮ ﻳﻚ ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‬ ‫‪k‬‬ ‫‪k=1‬‬
‫◀ ﺩﻧﺒﺎﻟﻪ‬
‫| ‪k+1 − 0‬‬
‫‪1‬‬
‫|‬
‫‪lim‬‬ ‫‪= 1.‬‬
‫| ∞→‪k‬‬
‫|‪k −0‬‬
‫‪1‬‬ ‫‪1‬‬
‫‬ ‫∞‬
‫ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻧﻴﺰ ﺑﺮﺍﺑﺮ ‪ ۱‬ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‬ ‫‪1‬‬
‫‪kk‬‬ ‫‪k=1‬‬
‫◀ ﺩﻧﺒﺎﻟﻪ‬
‫|‬ ‫‪1‬‬
‫‪(k+1)k‬‬
‫|‪−0‬‬
‫‪lim‬‬ ‫‪= 1.‬‬
‫∞→‪k‬‬ ‫‪| k1k −‬‬ ‫‪0 |1‬‬
‫‬ ‫∞‬
‫ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﻧﻴﺰ ﺑﺮﺍﺑﺮ ‪ ۱‬ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‬ ‫‪1‬‬
‫‪2k‬‬ ‫‪k=1‬‬
‫◀ ﺩﻧﺒﺎﻟﻪ‬
‫| ‪| 2k+1 − 0‬‬
‫‪1‬‬
‫‪1‬‬
‫‪lim‬‬ ‫‪= .‬‬
‫| ‪k→∞ | k − 0‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪2‬‬
‫‪2‬‬
‫‪n‬‬ ‫∞‪o‬‬
‫ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻣﺎ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﺮﺍﺑﺮ ‪ ۲‬ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‬ ‫‪22k‬‬
‫‪1‬‬
‫◀ ﺩﻧﺒﺎﻟﻪ‬
‫‪k=1‬‬
‫|‬ ‫‪1‬‬
‫|‪−0‬‬
‫‪22k+1‬‬
‫‪lim‬‬ ‫‪= 1.‬‬
‫‪k→∞ | 1k −‬‬ ‫‪0 |2‬‬
‫‪22‬‬
‫ﺑﺮﺍﻱ ﺩﻳﺪ ﺑﻬﺘﺮ‪ ،‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺑﻴﺴﺖ ﺟﻤﻠﻪ ﺍﻭﻝ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪﻫﺎ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬ ‫‪1‬‬
‫‪k‬‬ ‫‪k2‬‬ ‫‪k10‬‬
‫‪k‬‬ ‫‪2k‬‬ ‫‪22‬‬
‫‪k‬‬

‫‪1‬‬ ‫‪1.000000000000000‬‬ ‫‪1.000000000000000‬‬ ‫‪1.000000000000000‬‬ ‫‪0.500000000000000‬‬ ‫‪0.250000000000000‬‬

‫‪2‬‬ ‫‪0.500000000000000‬‬ ‫‪0.250000000000000‬‬ ‫‪0.000976562500000‬‬ ‫‪0.250000000000000‬‬ ‫‪0.062500000000000‬‬
‫‪3‬‬ ‫‪0.333333333333333‬‬ ‫‪0.111111111111111‬‬ ‫‪0.000016935087808‬‬ ‫‪0.125000000000000‬‬ ‫‪0.003906250000000‬‬
‫‪4‬‬ ‫‪0.250000000000000‬‬ ‫‪0.062500000000000‬‬ ‫‪0.000000953674316‬‬ ‫‪0.062500000000000‬‬ ‫‪0.000015258789063‬‬
‫‪5‬‬ ‫‪0.200000000000000‬‬ ‫‪0.040000000000000‬‬ ‫‪0.000000102400000‬‬ ‫‪0.031250000000000‬‬ ‫‪0.000000000232831‬‬
‫‪6‬‬ ‫‪0.166666666666667‬‬ ‫‪0.027777777777778‬‬ ‫‪0.000000016538172‬‬ ‫‪0.015625000000000‬‬ ‫‪5.42e-20‬‬
‫‪7‬‬ ‫‪0.142857142857143‬‬ ‫‪0.020408163265306‬‬ ‫‪0.000000003540133‬‬ ‫‪0.007812500000000‬‬ ‫‪2.93e-39‬‬
‫‪8‬‬ ‫‪0.125000000000000‬‬ ‫‪0.015625000000000‬‬ ‫‪0.000000000931323‬‬ ‫‪0.003906250000000‬‬ ‫‪8.63e-78‬‬
‫‪9‬‬ ‫‪0.111111111111111‬‬ ‫‪0.012345679012346‬‬ ‫‪0.000000000286797‬‬ ‫‪0.001953125000000‬‬ ‫‪7.45e-155‬‬
‫‪10‬‬ ‫‪0.100000000000000‬‬ ‫‪0.010000000000000‬‬ ‫‪0.000000000100000‬‬ ‫‪0.000976562500000‬‬
‫‪11‬‬ ‫‪0.090909090909091‬‬ ‫‪0.008264462809917‬‬ ‫‪0.000000000038554‬‬ ‫‪0.000488281250000‬‬
‫‪12‬‬ ‫‪0.083333333333333‬‬ ‫‪0.006944444444444‬‬ ‫‪0.000000000016151‬‬ ‫‪0.000244140625000‬‬
‫‪13‬‬ ‫‪0.076923076923077‬‬ ‫‪0.005917159763314‬‬ ‫‪0.000000000007254‬‬ ‫‪0.000122070312500‬‬
‫‪14‬‬ ‫‪0.071428571428571‬‬ ‫‪0.005102040816327‬‬ ‫‪0.000000000003457‬‬ ‫‪0.000061035156250‬‬
‫‪15‬‬ ‫‪0.066666666666667‬‬ ‫‪0.004444444444444‬‬ ‫‪0.000000000001734‬‬ ‫‪0.000030517578125‬‬
‫‪16‬‬ ‫‪0.062500000000000‬‬ ‫‪0.003906250000000‬‬ ‫‪0.000000000000909‬‬ ‫‪0.000015258789063‬‬
‫‪17‬‬ ‫‪0.058823529411765‬‬ ‫‪0.003460207612457‬‬ ‫‪0.000000000000496‬‬ ‫‪0.000007629394531‬‬
‫‪18‬‬ ‫‪0.055555555555556‬‬ ‫‪0.003086419753086‬‬ ‫‪0.000000000000280‬‬ ‫‪0.000003814697266‬‬
‫‪19‬‬ ‫‪0.052631578947368‬‬ ‫‪0.002770083102493‬‬ ‫‪0.000000000000163‬‬ ‫‪0.000001907348633‬‬
‫‪20‬‬ ‫‪0.050000000000000‬‬ ‫‪0.002500000000000‬‬ ‫‪0.000000000000098‬‬ ‫‪0.000000953674316‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﻧﺒﺎﻟﻪ ﺑﺎ ﻫﻤﮕﺮﺍﻳﻲ ﻣﺮﺗﺒﻪ ﺩﻭ ﺧﻴﻠﻲ ﺳﺮﻳﻊﺗﺮ ﺍﺯ ﺩﻧﺒﺎﻟﻪﻫﺎﻱ ﻣﺮﺗﺒﻪ ﻳﻚ ﻫﻤﮕﺮﺍ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۵۸‬‬

‫ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﺩﻧﺒﺎﻟﻪﻫﺎﻱ ﺑﺎ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻳﻚ‪ ،‬ﺳﺮﻋﺖ ﻣﻴﻞ ﻛﺮﺩﻥ ﺑﻪ ﺻﻔﺮ ﻳﻜﺴﺎﻥ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﭼﻨﻴﻦ ﺑﻪ ﻧﻈﺮ‬
‫ﻣﻲﺭﺳﺪ ﻛﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻣﻌﻴﺎﺭ ﻭ ﻳﺎ ﻣﺤﻚ ﻛﺎﻣﻼ ﻣﻨﺎﺳﺒﻲ ﺑﺮﺍﻱ ﺳﻨﺠﺶ ﻭ ﻣﻘﺎﻳﺴﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪﻫﺎ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺑﺎﻳﺪ‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﻌﻴﺎﺭ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺟﺎﻳﮕﺎﻩ ﺧﻮﺩ ﺭﺍ ﺩﺭ ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﻱ ﺩﺍﺭﺩ ﻭ ﻳﻚ ﺍﺑﺰﺍﺭ ﻃﺒﻘﻪﺑﻨﺪﻱ ﺩﻧﺒﺎﻟﻪﻫﺎ ﺑﻪ ﻟﺤﺎﻅ ﺳﺮﻋﺖ‬
‫ﻫﻤﮕﺮﺍ ﺷﺪﻥ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﻧﺒﺎﻟﻪﻫﺎﻳﻲ ﺑﺎ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻭ ﺑﺎ ﺳﺮﻋﺖ ﺑﺴﻴﺎﺭ ﺑﺎﻻﻳﻲ ﺑﻪ ﺣﺪ ﺧﻮﺩ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫ﺣﺎﻝ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﺑﺮ ﻣﻲﮔﺮﺩﻳﻢ‪ .‬ﺭﻭﺵﻫﺎﻱ ﺗﻜﺮﺍﺭﻱ ﻳﻚ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﻣﻲﻛﻨﻨﺪ ﻛﻪ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ )ﻭ ﻳﺎ ﺭﻳﺸﻪ(‬
‫ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺗﺒﻌﺎ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺭﻭﺵ ﺗﻜﺮﺍﺭﻱ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻫﻤﺎﻥ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﺷﺪﻩ ﺗﻌﺮﻳﻒ ﻧﻤﻮﺩ‪.‬‬
‫ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻣﺮﺗﺒﻪ ﻫﻤ ﺮﺍﯾﯽ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۱۴−۵‬‬
‫‬ ‫∞‬
‫ﺭﺍ ﺗﻮﻟﻴﺪ ﻛﻨﺪ ﻭ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ‪ p‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ ﻭ ﺩﺭ ﺿﻤﻦ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ‬ ‫ﺍﮔﺮ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺩﻧﺒﺎﻟﻪ‬
‫‪xk k=0‬‬
‫ﺩﻧﺒﺎﻟﻪ ﺑﺮﺍﺑﺮ ‪ σ‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺁﻥ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﺑﺮ ‪ σ‬ﺍﺳﺖ‪.‬‬

‫ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ )‪ g(x‬ﺑﺎ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ x0‬ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩﺍﻳﻢ ﻭ ﺩﻧﺒﺎﻟﻪ‬
‫‬ ‫∞‬
‫‪ xk k=0‬ﺗﻮﻟﻴﺪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ ﻣﺬﻛﻮﺭ ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺑﺎﻳﺪ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﺗﻮﻟﻴﺪ ﻛﻨﻴﻢ ﻭ‬
‫ﺩﺭ ﺑﻲﻧﻬﺎﻳﺖ ﺭﻓﺘﺎﺭ ﺁﻥ ﺭﺍ ﺑﺮﺭﺳﻲ ﻛﻨﻴﻢ‪ .‬ﻛﻪ ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﺍﻳﻦ ﻛﺎﺭ ﻏﻴﺮﻣﻤﻜﻦ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﻗﻀﻴﻪﺍﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺑﺎ ﻛﻤﻚ ﺁﻥ‬
‫ﻣﻲﺗﻮﺍﻥ ﺳﺎﺩﻩﺗﺮ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻳﻚ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺭﺍ ﻳﺎﻓﺖ‪ .‬ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﺭﻭﺵ ﺗﮑﺮﺍﺭ ﺳﺎﺩﻩ‬ ‫ﻣﺤﺎﺳﺒﻪ ﻣﺮﺗﺒﻪ ﻫﻤ ﺮﺍﯾﯽ ﯾ‬ ‫ﻗﻀﯿﻪ ‪۱۵−۵‬‬

‫‬ ‫∞‬
‫ﺍﮔﺮ ‪ xk k=0‬ﺩﻧﺒﺎﻟﻪ ﺗﻮﻟﻴﺪ ﺷﺪﻩ ﺗﻮﺳﻂ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ )‪ g(x‬ﺑﺎﺷﺪ ﻭ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻧﻘﻄﻪ‬
‫ﺛﺎﺑﺖ ‪ p‬ﻫﻤﮕﺮﺍ ﮔﺮﺩﺩ‪ ،‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ‪:‬‬
‫‬ ‫∞‬
‫‪ xk‬ﺑﺮﺍﺑﺮ ﻳﻚ ﺍﺳﺖ‪.‬‬ ‫‪k=0‬‬
‫• ﺍﮔﺮ ‪ ،g ′ (p) ̸= 0‬ﺁﻥﮔﺎﻩ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ‬
‫‬ ‫∞‬
‫‪ xk‬ﺑﺮﺍﺑﺮ ﺩﻭ ﺍﺳﺖ‪.‬‬ ‫‪k=0‬‬
‫• ﺍﮔﺮ ‪ g ′ (p) = 0‬ﻭ ‪ ،g ′′ (p) ̸= 0‬ﺁﻥﮔﺎﻩ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ‬
‫‬ ‫∞‬
‫‪xk‬‬ ‫‪k=0‬‬
‫• ﺍﮔﺮ ‪ g ′ (p) = g ′′ (p) = · · · = g (s−1) (p) = 0‬ﻭ ‪ g (s) (p) ̸= 0‬ﺁﻥﮔﺎﻩ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻧﺒﺎﻟﻪ‬
‫ﺑﺮﺍﺑﺮ ‪ s‬ﺍﺳﺖ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۲۱−۵‬ﻣﺸﺨﺺ ﮐﺮﺩﻥ ﻣﺮﺗﺒﻪ ﻫﻤ ﺮﺍﯾﯽ(‬

‫
‬
‫ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ﺗﺎﺑﻊ ‪ g(x) = 15 4x + x34‬ﺑﺎ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ‪ ،‬ﺑﺎ ﺍﻧﺘﺨﺎﺏ ﻳﻚ ﺣﺪﺱ ﺍﻭﻟﻴﻪ ‪ ،x0‬ﺩﻧﺒﺎﻟﻪ‬
‫‬ ‫∞‬
‫‪ xk k=0‬ﺗﻮﻟﻴﺪ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺻﻮﺭﺕ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﺩﻧﺒﺎﻟﻪ ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬
‫‬ ‫‬
‫‪1‬‬ ‫‪3‬‬
‫=‪p‬‬ ‫‪4p + 4‬‬
‫‪5‬‬ ‫‪p‬‬
‫√‬
‫ﻭ ﺑﻌﺪ ﺍﺯ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺟﺒﺮﻱ‪ ،‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ‪ .p = 5 3‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺩﻧﺒﺎﻟﻪ ﺣﺎﺻﻞ ﺍﺯ ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ )ﺩﺭ ﺻﻮﺭﺕ‬
‫√‬
‫ﻫﻤﮕﺮﺍﻳﻲ( ﺑﻪ ﻧﻘﻄﻪ ﺛﺎﺑﺖ ‪ p = 5 3‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺍﻳﻦ ﺩﻧﺒﺎﻟﻪ ﺭﺍ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪.‬‬
‫
‬
‫ﺩﺍﺭﻳﻢ‪ .g ′ (x) = 15 4 − x125 :‬ﻟﺬﺍ‬
‫√‬
‫‪g ′ (p) = g ′ ( 3) = 0.‬‬
‫‪5‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺣﺪﺍﻗﻞ ﺑﺮﺍﺑﺮ ﺩﻭ ﺍﺳﺖ‪ .‬ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﺩﻗﻴﻖﺗﺮ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﻣﺸﺘﻖ ﺩﻭﻡ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۵۹‬‬ ‫ﺭﻭﺵﻫﺎﯼ ﻋﺪﺩﯼ ﺑﺮﺍﯼ ﺭﯾﺸﻪﯾﺎﺑﯽ‬ ‫ﻓﺼﻞ ﭘﻨﺠﻢ‪:‬‬

‫= )‪ .g ′′ (x‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬ ‫‪12‬‬
‫‪x6‬‬ ‫ﺗﺎﺑﻊ ‪ g‬ﺭﺍ ﺑﻪ ﺳﺖ ﻣﻲﺍﻭﺭﻳﻢ ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪g ′′ (p) ̸= 0.‬‬
‫ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﺩﻗﻴﻘﺎ ﺑﺮﺍﺑﺮ ﺩﻭ ﺍﺳﺖ‪.‬‬

‫ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‬ ‫�‬

‫ﻗﺒﻼ ﺑﻴﺎﻥ ﺷﺪ ﻛﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻳﻚ ﺭﻭﺵ ﺳﺮﻳﻊ ﺑﺮﺍﻱ ﺭﻳﺸﻪﻳﺎﺑﻲ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻨﺠﺎ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻳﻦ ﺭﻭﺵ ﺭﺍ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﻣﺮﺗﺒﻪ‬
‫)‪ g(x) = x − ff′(x‬ﺩﺭ‬‫ﻫﻤﮕﺮﺍﻳﻲ ﻣﻄﺎﻟﻌﻪ ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ‪ f (x) = 0‬ﺭﻭﺵ ﺗﻜﺮﺍﺭ ﺳﺎﺩﻩ ﻭ ﺗﺎﺑﻊ )‪(x‬‬
‫ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﻓﺮﺽ ﻛﻨﻴﻢ ﺩﻧﺒﺎﻟﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﻪ ‪ α‬ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫)‪[f ′ (x)]2 − f (x)f ′′ (x‬‬
‫‪g ′ (x) = 1 −‬‬
‫‪[f ′ (x)]2‬‬
‫ﻭ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ ،f (p) = 0‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ‪:‬‬
‫‪′‬‬
‫‪[f (p)]2‬‬
‫‪g ′ (p) = 1 −‬‬
‫‪[f ′ (p)]2‬‬
‫ﻭ ﺍﮔﺮ ‪ ،f ′ (p) ̸= 0‬ﺁﻥﮔﺎﻩ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ‪ .g ′ (p) = 0‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‪:‬‬
‫ﺣﺪﺍﻗﻞ‬ ‫ﺍﮔﺮ ‪ α‬ﺭﻳﺸﻪ ﺳﺎﺩﻩ ‪f‬ﺑﺎﺷﺪ )‪ (f ′ (α) ̸= 0‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﺭﻳﺸﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺍﺭﺍﻱ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ‬
‫ﺩﻭ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺍﺭﺍﻱ ﺭﻳﺸﻪ ﺑﺎ ﻣﺮﺗﺒﻪ ﺩﻭ ﻳﺎ ﺑﺎﻻﺗﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺁﻥ ﺗﺎﺑﻊ‪ ،‬ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺩﻭ ﺭﺍ ﺍﺯ‬
‫ﺩﺳﺖ ﻣﻲﺩﻫﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶۰‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۶‬‬
‫ﺑﺴﻂ ﺗﯿﻠﻮﺭ‬
‫ﻓﺼﻞ ﺷﺸﻢ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۱۶۲‬‬ ‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۶‬‬
‫‪۱۶۴‬‬ ‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۶‬‬
‫‪۱۶۵‬‬ ‫ﺧﻄﺎﻱ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . n‬‬ ‫‪۱.۲.۶‬‬
‫‪۱۶۸‬‬ ‫∗ ﻓﺮﻡﻫﺎﻱ ﺩﻳﮕﺮ ﺑﺮﺍﻱ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ‪. . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۶‬‬

‫‪۱۶۱‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶۲‬‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ‪ ۱‬ﻳﻚ ﻧﻤﺎﻳﺶ ﺑﻪ ﺻﻮﺭﺕ ﺳﺮﻱ ﺑﺮﺍﻱ ﺑﺮﺧﻲ ﺗﻮﺍﺑﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﺑﺴﻂ ﺩﺍﺭﺍﻱ ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺯﻳﺎﺩﻱ ﺩﺭ ﺭﻳﺎﺿﻴﺎﺕ ﻭ‬
‫ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ ﺍﺯ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺗﻮﺍﺑﻊ ﻭ ﺁﻧﺎﻟﻴﺰ ﺧﻄﺎ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ‪ ،‬ﻣﺒﺎﻧﻲ‬
‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺍﺭﺍﻳﻪ ﺧﻮﺍﻫﺪ ﮔﺮﺩﻳﺪ‪.‬‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ‬ ‫‪۱.۶‬‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ‪ ،‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻳﻚ ﺳﺮﻱ ﺗﻮﺍﻧﻲ ‪ ۲‬ﻛﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ‬ ‫ﺳﺮﯼ ﺗﯿﻠﻮﺭ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۱−۶‬‬

‫ﺍﮔﺮ ‪ f‬ﺗﺎﺑﻌﻲ ﺣﻘﻴﻘﻲ ﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺑﻲﻧﻬﺎﻳﺖ ﺑﺎﺭ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ ﻧﻘﻄﻪ ‪x = α‬‬
‫ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪(x − α) ′‬‬ ‫‪(x − α) ′′‬‬ ‫‪2‬‬
‫)‪(x − α) (n‬‬ ‫‪n‬‬
‫‪f (α) +‬‬ ‫‪f (α) +‬‬ ‫‪f (α) + · · · +‬‬ ‫· · · ‪f (α) +‬‬ ‫)‪(۱.۶‬‬
‫!‪1‬‬ ‫!‪2‬‬ ‫!‪n‬‬
‫ﺑﺴﻂ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﻓﺸﺮﺩﻩ ﺯﻳﺮ ﻧﻴﺰ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‬
‫∞‬
‫‪X‬‬ ‫)‪(x − α) (n‬‬ ‫‪n‬‬
‫)‪f (α‬‬ ‫)‪(۲.۶‬‬
‫‪n=0‬‬
‫!‪n‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﺮﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩﻥ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ‪ ،‬ﺑﺎﻳﺪ ﺩﺭ ﺍﺑﺘﺪﺍ ﻣﺸﺨﺺ ﺷﻮﺩ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺭﺍ ﺣﻮﻝ‬
‫ﭼﻪ ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺎﻳﺪ ﺍﺯ ﺑﻲﻧﻬﺎﻳﺖ ﻣﺸﺘﻖﭘﺬﻳﺮ ﺑﻮﺩﻥ ﺗﺎﺑﻊ ﺩﺭ ﺁﻥ ﻧﻘﻄﻪ ﻧﻴﺰ ﺍﻃﻤﻴﻨﺎﻥ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ .‬ﺩﺭ ﺯﻳﺮ‬
‫ﭼﻨﺪ ﻣﺜﺎﻝ ﺍﺯ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﻮﺍﺑﻊ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱−۶‬ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﭼﻨﺪ ﺗﺎﺑﻊ(‬

‫ﺩﺭ ﺯﻳﺮ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﭼﻨﺪ ﺗﺎﺑﻊ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﺎﺑﻊ‬ ‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬ ‫ﻓﺮﻡ ﻓﺸﺮﺩﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬
‫∞‪P‬‬
‫= )‪f (x‬‬ ‫‪1‬‬
‫‪1−x‬‬ ‫‪1 + x + x + ··· + x + ...‬‬
‫‪2‬‬ ‫‪n‬‬
‫‪n=0‬‬ ‫‪xn‬‬
‫∞‪P‬‬
‫= )‪f (x‬‬ ‫‪1‬‬
‫‪1+x‬‬ ‫‪1 − x + x2 + · · · + (−x)n + . . .‬‬ ‫‪n n‬‬
‫‪n=0 (−1) x‬‬
‫‪2‬‬ ‫‪n‬‬ ‫∞‪P‬‬ ‫‪n‬‬

‫‪2! + · · · + n! + . . .‬‬
‫‪x‬‬ ‫‪x‬‬ ‫‪x‬‬ ‫‪x‬‬
‫‪f (x) = e‬‬ ‫‪1+x+‬‬ ‫‪n=0‬‬
‫‪3‬‬ ‫‪5‬‬
‫‪x2n+1‬‬
‫‪P∞ n! n x2n+1‬‬
‫)‪f (x) = sin(x‬‬ ‫‪x−‬‬ ‫‪x‬‬
‫!‪3‬‬ ‫!)‪+ x5! · · · + (−1)n (2n+1‬‬ ‫‪+ ...‬‬ ‫!)‪n=0 (−1) (2n+1‬‬
‫‪x2‬‬ ‫‪x4‬‬ ‫‪n x2n‬‬
‫∞‪P‬‬ ‫‪n x2n‬‬
‫)‪f (x) = cos(x‬‬ ‫‪1−‬‬ ‫!‪2‬‬ ‫‪+ 4! · · · + (−1) (2n)! + . . .‬‬ ‫!)‪n=0 (−1) (2n‬‬
‫‪x2‬‬ ‫‪3‬‬ ‫‪n‬‬ ‫∞‪P‬‬ ‫‪n−1 xn‬‬
‫)‪f (x) = ln(1 + x‬‬ ‫‪x−‬‬ ‫‪2‬‬ ‫‪+ x3 · · · + (−1)n−1 xn + . . .‬‬ ‫)‪n=0 (−1‬‬ ‫‪n‬‬

‫ﺳﻮﺍﻟﻲ ﻛﻪ ﭘﻴﺶ ﻣﻲﺁﻳﺪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﭼﻪ ﺍﺭﺗﺒﺎﻃﻲ ﺑﺎ ﺧﻮﺩ ﺗﺎﺑﻊ ﺩﺍﺭﺩ؟ ﺍﺻﻮﻻ ﺍﻧﺘﻈﺎﺭ ﻣﻲﺭﻭﺩ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬
‫ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﺧﻮﺩ ﺗﺎﺑﻊ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺍﻳﻦﮔﻮﻧﻪ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ‪ ،‬ﺩﺭ ﻧﻘﺎﻃﻲ ﻫﻤﮕﺮﺍ ﻧﺒﺎﺷﺪ‪ .‬ﺗﺒﻌﺎ ﺩﺭ‬
‫ﺍﻳﻦ ﻧﻘﺎﻁ‪ ،‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺑﺎ ﺧﻮﺩ ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫‪1 Taylor‬‬ ‫‪expansion‬‬ ‫‪2 Power‬‬ ‫‪series‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۶۳‬‬ ‫ﺑﺴﻂ ﺗﯿﻠﻮﺭ‬ ‫ﻓﺼﻞ ﺷﺸﻢ‪:‬‬

‫ﻣﺜﺎﻝ ‪)۲−۶‬ﻫﻤ ﺮﺍﯾﯽ ﻭ ﺩﺍﻣﻨﻪ ﻫﻤ ﺮﺍﯾﯽ ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﭼﻨﺪ ﺗﺎﺑﻊ(‬

‫‪P‬‬
‫∞ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺁﺯﻣﻮﻥﻫﺎﻱ ﻫﻤﮕﺮﺍﻳﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ‬ ‫◀ﺩﻳﺪﻳﻢ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f (x) = 1−x‬ﺑﻪ ﺻﻮﺭﺕ ‪n=0 x‬‬
‫‪n‬‬ ‫‪1‬‬

‫ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ ﺍﻳﻦ ﺳﺮﻱ ﻓﻘﻂ ﺩﺭ ﺑﺎﺯﻩ )‪ (−1, 1‬ﻫﻤﮕﺮﺍ ﻣﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺑﻘﻴﻪ ﺟﺎﻫﺎ‪ ،‬ﺳﺮﻱ ﻭﺍﮔﺮﺍ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﻳﻦ ﺳﺮﻱ ﺗﻨﻬﺎ ﺩﺭ‬
‫ﺑﺎﺯﻩ )‪ (−1, 1‬ﺍﺭﺯﺵ ﺩﺍﺭﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ ﺍﻳﻦ ﺳﺮﻱ ﺩﺭ ﺑﺎﺯﻩ ﻓﻮﻕ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ ‪ f‬ﻫﻤﮕﺮﺍ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻃﻮﺭ‬
‫‪ f (x) = 1−x‬ﺗﻨﻬﺎ ﺩﺭ ﺑﺎﺯﻩ )‪ (−1, 1‬ﺑﺎ ﺧﻮﺩ ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‪.‬‬
‫‪1‬‬
‫ﺧﻼﺻﻪ‪ ،‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ‬
‫◀ﺍﻣﺎ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﻮﺍﺑﻊ ‪ sin(x) ،e‬ﻭ )‪ cos(x‬ﻫﻤﻪ ﺟﺎ ﻫﻤﮕﺮﺍ ﻣﻲﺑﺎﺷﺪ ﻭ ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺳﺮﻱﻫﺎ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ‬
‫‪x‬‬

‫ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﺜﺎﻝﻫﺎﻱ ﻓﻮﻕ ﻣﻤﻜﻦ ﺍﺳﺖ ﭼﻨﻴﻦ ﺣﺪﺱ ﺯﺩﻩ ﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻳﻚ ﺗﺎﺑﻊ ﺩﺭ ﻳﻚ ﻧﻘﻄﻪﺍﻱ ﻣﺎﻧﻨﺪ ‪ x‬ﻫﻤﮕﺮﺍ‬
‫ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺑﺴﻂ ﺑﻪ )‪ f (x‬ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﻣﺎ ﭼﻨﻴﻦ ﺣﺪﺳﻲ ﺩﺭﺳﺖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺍﮔﺮ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺩﺭ ﻳﻚ‬
‫ﻧﻘﻄﻪ ﻫﻤﮕﺮﺍ ﺑﺎﺷﺪ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻪ ﻣﻘﺪﺍﺭﻱ ﻏﻴﺮ ﺍﺯ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﺁﻥ ﻧﻘﻄﻪ ﻫﻤﮕﺮﺍ ﺷﻮﺩ‪ .‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺗﺎﺑﻊ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ(‬ ‫ﻣﺜﺎﻝ ‪)۳−۶‬ﻫﻤ ﺮﺍ ﻧﺒﻮﺩﻥ ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﯾ‬

‫‪1.5‬‬ ‫ﺗﺎﺑﻊ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‬

‫‪‬‬
‫‪1‬‬ ‫‪0,‬‬ ‫‪x = 0,‬‬
‫‪0.5‬‬ ‫=‪f (x) :‬‬
‫‪x‬‬ ‫‪e−1/x ,‬‬ ‫‪2‬‬
‫‪x ̸= 0.‬‬
‫‪−4‬‬ ‫‪−2‬‬ ‫‪−0.5‬‬ ‫‪2‬‬ ‫‪4‬‬
‫ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺭ ﺷﻜﻞ ﻣﻘﺎﺑﻞ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻣﻲﺗﻮﺍﻥ ﻧﺸﺎﻥ ﺩﺍﺩ ﻛﻪ ﺗﻤﺎﻣﻲ ﻣﺸﺘﻘﺎﺕ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ α = 0‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ‪ .f (n) (0) = 0, n = 0, 1, . . .‬ﻟﺬﺍ‬
‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﻓﻮﻕ ﺣﻮﻝ ‪ α = 0‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪0 + 0 · x + 0 · x2 + · · · = 0‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻫﻤﻪ ﺟﺎ ﺑﻪ ﺻﻔﺮ ﻫﻤﮕﺮﺍ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﻣﺎ ﺗﺎﺑﻊ ‪ f‬ﻓﻘﻂ ﺩﺭ ﻧﻘﻄﻪ ‪ x = 0‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ ﻭ‬
‫ﺩﺭ ﺑﻘﻴﻪ ﻧﻘﺎﻁ ﻏﻴﺮﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ‪ :‬ﻫﺮﭼﻨﺪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻫﻤﻪ ﻧﻘﺎﻁ ﺩﺍﺭﺍﻱ ﻣﺸﺘﻘﺎﺕ ﺍﺯ ﻫﺮ ﻣﺮﺗﺒﻪﺍﻱ ﻣﻲﺑﺎﺷﺪ‬
‫ﻭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺁﻥ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ ،‬ﺍﻣﺎ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﻓﻘﻂ ﺩﺭ ﻧﻘﻄﻪ ‪ x = 0‬ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﻣﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺑﻘﻴﻪ ﻧﻘﺎﻁ ﺑﻪ ﺗﺎﺑﻊ‬
‫ﻫﻤﮕﺮﺍ ﻧﻤﻲﺑﺎﺷﺪ‪.‬‬

‫ﺩﺭﺧﺼﻮﺹ ﻫﻤﮕﺮﺍﻳﻲ ﻳﺎ ﻋﺪﻡ ﻫﻤﮕﺮﺍﻳﻲ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ‪ ،‬ﻣﻔﻬﻮﻡ ﺗﺤﻠﻴﻠﻲ ‪ ۳‬ﻣﻄﺮﺡ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻧﻘﻄﻪ‬ ‫ﺗﺎﺑﻊ ﺩﺭ ﯾ‬ ‫ﺗﺤﻠﯿﻠ ﺑﻮﺩﻥ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۶‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ f‬ﻳﻚ ﺗﺎﺑﻊ ﻭ ‪ α‬ﻳﻚ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ ‪ x = α‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺩﺭ‬
‫ﻳﻚ ﻫﻤﺴﺎﻳﮕﻲ ﺍﺯ ﻧﻘﻄﻪ ‪ ،α‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﺷﻮﺩ ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ α‬ﺗﺤﻠﻴﻠﻲ ﺍﺳﺖ‪.‬‬

‫‪ 1−x‬ﺗﻨﻬﺎ ﺩﺭ ﺑﺎﺯﻩ )‪ (−1, 1‬ﺗﺤﻠﻴﻠﻲ‬

‫‪1‬‬
‫ﺗﻮﺍﺑﻊ ‪ sin(x) ،ex‬ﻭ )‪ cos(x‬ﺩﺭ ﻛﻞ ﻧﻘﺎﻁ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ﺗﺤﻠﻴﻠﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻣﺎ ﺗﺎﺑﻊ‬
‫ﺍﺳﺖ ﻭ ﺗﺎﺑﻊ ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺩﺭ ﻣﺜﺎﻝ ‪ ۳-۶‬ﻫﻴﭻﺟﺎ ﺗﺤﻠﻴﻠﻲ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺗﺤﻠﻴﻠﻲ ﺑﻮﺩﻥ ﻳﻚ ﻣﻔﻬﻮﻡ ﻗﻮﻱﺗﺮ ﺍﺯ ﺑﻲﻧﻬﺎﻳﺖ ﺑﺎﺭ ﻣﺸﺘﻖﭘﺬﻳﺮ‬
‫ﺑﻮﺩﻥ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﺍﮔﺮ ﺗﺎﺑﻌﻲ ﺩﺭ ﻳﻚ ﻧﻘﻄﻪ ﺗﺤﻠﻴﻠﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺁﻥ ﺗﺎﺑﻊ ﺩﺭ ﺁﻥ ﻧﻘﻄﻪ ﺗﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺑﺎﺭ ﻣﺸﺘﻖﭘﺬﻳﺮ‬
‫‪3 Analitic‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶۴‬‬

‫ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﻋﻜﺲ ﺁﻥ ﺩﺭﺳﺖ ﻧﻴﺴﺖ‪.‬‬

‫ﺍﮔﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ α‬ﺗﺤﻠﻴﻠﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻳﻚ ﻣﺠﻤﻮﻋﻪ ﺷﺎﻣﻞ ‪ α‬ﺑﺎ ﺧﻮﺩ ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‪ .‬ﺍﻳﻦ‬
‫ﻧﻜﺘﻪ ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﻭ ﻣﺸﺘﻘﺎﺕ ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺩﺭ ﻧﻘﻄﻪ ‪ α‬ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﻛﻞ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻳﻚ‬
‫ﻣﺠﻤﻮﻋﻪ ﺍﺯ ﻧﻘﺎﻁ ﺍﻃﺮﺍﻑ ‪ α‬ﺗﺎﺑﻊ ‪ f‬ﺑﺴﺎﺯﻳﻢ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺗﺎﺑﻊ ‪ sin‬ﺩﺭ ﻛﻞ ‪ R‬ﺗﺤﻠﻴﻠﻲ ﺍﺳﺖ ﻭ ﻣﺎ ﺍﻃﻼﻋﺎﺕ ﺗﺎﺑﻊ ﻭ ﻣﺸﺘﻘﺎﺗﺶ‬
‫ﺭﺍ ﺩﺭ ﻧﻘﻄﻪ ‪ α = 0‬ﺩﺍﺭﻳﻢ‪ .‬ﺑﺎ ﺍﻳﻦ ﺍﻃﻼﻋﺎﺕ ﻣﻲﺗﻮﺍﻥ ﻛﻞ ﺗﺎﺑﻊ ‪ sin‬ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ ﻭ ﺑﺎﺯﺳﺎﺯﻱ ﻛﺮﺩ‪.‬‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ‬ ‫‪۲.۶‬‬

‫ﺧﺎﻃﺮﻧﺸﺎﻥ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺗﺎ ﺑﻴﻨﻬﺎﻳﺖ ﺑﺎﺭ ﻣﺸﺘﻖﭘﺬﻳﺮ ﻧﺒﺎﺷﺪ ﺁﻥﮔﺎﻩ ﻧﻤﻲﺗﻮﺍﻥ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺭﺍ ﻧﻮﺷﺖ‬
‫ﻭ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ‪ ،‬ﺍﺯ ﺟﻤﻠﻪ ﺩﺭ ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ‪ ،‬ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺑﺎ ﺑﻲﻧﻬﺎﻳﺖ ﺟﻤﻠﻪ ﻣﺸﻜﻞ‬
‫ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﺗﺎﺑﻊ ﻣﻄﺮﺡ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺗﺎﺑﻊ ﺣﻘﯿﻘ‬ ‫ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۶‬‬
‫ﺍﮔﺮ ﻣﺸﺘﻘﺎﺕ ﺗﺎ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﺗﺎﺑﻊ ‪ f‬ﺑﻪﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‬
‫‪(x − α) ′‬‬ ‫‪(x − α)2 ′′‬‬ ‫)‪(x − α)n (n‬‬
‫‪Tα,n (x) := f (α) +‬‬ ‫‪f (α) +‬‬ ‫‪f (α) + · · · +‬‬ ‫)‪f (α‬‬
‫!‪1‬‬ ‫!‪2‬‬ ‫!‪n‬‬

‫ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ ،n‬ﺑﺮﺵ ﺧﻮﺭﺩﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺍﺳﺖ‪ .‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ‪ ،‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻲﺑﺎﺷﺪ ﻛﻪ‬
‫ﺑﻪ ﺭﺍﺣﺘﻲ ﻣﻲﺗﻮﺍﻥ ﺁﻥ ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻭ ﺍﺭﺯﻳﺎﺑﻲ ﻧﻤﻮﺩ‪.‬‬

‫■ ﻧﻜﺎﺗﻲ ﺩﺭ ﺧﺼﻮﺹ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ‬

‫• )ﺗﻘﺮﻳﺐ ﺧﻄﻲ ﻭ ﻣﺮﺑﻌﻲ( ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﻳﻚ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ ‪ α‬ﺑﻪ ﺗﻘﺮﻳﺐ ﺧﻄﻲ ‪ f‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺑﺴﻂ‬
‫ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﻳﻚ ﻫﻤﺎﻥ ﺧﻂ ﻣﻤﺎﺱ ﺑﺮ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ α‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﺩﻭﻡ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ ‪ α‬ﺑﻪ ﺗﻘﺮﻳﺐ ﻣﺮﺑﻌﻲ‬
‫ﻳﺎ ﺩﺭﺟﻪ ﺩﻭﻡ ‪ f‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﺩﻭﻡ ﺗﺎﺑﻊ ‪ ،f‬ﻳﻚ ﺗﺎﺑﻊ ﺩﺭﺟﻪ ﺩﻭﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• )ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻌﻤﻴﻤﻲ ﺍﺯ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﮕﻴﻦ( ﻗﻀﻴﻪ ﺗﻴﻠﻮﺭ ﺑﻪ ﻧﻮﻋﻲ ﺗﻌﻤﻴﻢ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﮕﻴﻦ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻃﺒﻖ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﮕﻴﻦ ‪ ۴‬ﺍﮔﺮ ]‪ ،f ∈ C[a, x‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‬
‫)‪ f (x) = f (a) + (x−a‬ﺑﺎﺯﻧﻮﻳﺴﻲ ﻛﺮﺩ ﻛﻪ‬
‫)‪ .f (x) − f (a) = f (η)(x − a‬ﻛﻪ ﻣﻲﺗﻮﺍﻥ ﺁﻧﺮﺍ ﺑﻪ ﻓﺮﻡ )‪1! f (η‬‬
‫‪′‬‬ ‫‪′‬‬

‫ﻫﻤﺎﻥ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﺻﻔﺮ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ ‪ x = a‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• )ﺗﻄﺎﺑﻖ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺑﺎ ﺗﺎﺑﻊ( ﺍﮔﺮ )‪ Tα,n (x‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﺑﺮﺍﻱ ﺗﺎﺑﻊ )‪ f (x‬ﺣﻮﻝ ﻧﻘﻄﻪ ‪ α‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ‬
‫)‪ f (x‬ﻭ ﺑﺴﻂ )‪ Tα,n (x‬ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﺑﺮﺍﺑﺮﻧﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻣﺸﺘﻘﺎﺕ ﺗﺎ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﺁﻥﻫﺎ ﻧﻴﺰ ﺩﺭ ﻧﻘﻄﻪ‬
‫‪ x = α‬ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﺑﺮﺍﺑﺮﻧﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻗﻴﻖﺗﺮ‬
‫‪Tα,n‬‬
‫)‪(k‬‬
‫‪(α) = f (k) (α), k = 0, · · · , n.‬‬

‫• )ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻭ ﻳﺎ ﺍﺯ ﺩﺭﺟﻪ ‪ (n‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﺯ ﻭﺍﮊﻩ ﺩﺭﺟﻪ ‪ ۵‬ﺑﻪ ﺟﺎﻱ ﻭﺍﮊﻩ ﻣﺮﺗﺒﻪ ﺩﺭ ﺗﻌﺮﻳﻒ ‪ ۳-۶‬ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪.‬‬
‫ﺍﻣﺎ ﻭﺍﮊﻩ ﻣﺮﺗﺒﻪ ﻣﻨﺎﺳﺐﺗﺮ ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ )‪ f (n) (α‬ﺻﻔﺮ ﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﺑﺴﻂ‬
‫‪4 Mean‬‬ ‫)‪Value Theorem (for Derivatives‬‬ ‫‪5 Degree‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۶۵‬‬ ‫ﺑﺴﻂ ﺗﯿﻠﻮﺭ‬ ‫ﻓﺼﻞ ﺷﺸﻢ‪:‬‬

‫ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪n‬ﺍﻡ ﺍﻳﻦ ﺗﺎﺑﻊ ﺍﺯ ﺩﺭﺟﻪ ‪ n‬ﻧﺨﻮﺍﻫﺪ ﺑﻮﺩ ﻭ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻥ ﻋﺒﺎﺭﺕ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺩﺭﺟﻪ ‪n‬ﺍﻡ ﻣﻨﺎﺳﺐ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫ﻣﺜﺎﻝ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﭼﻬﺎﺭ ﺗﺎﺑﻊ ‪ sin x‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ ۳‬ﻣﻲﺑﺎﺷﺪ ﻭ ﺍﺳﺘﻨﺎﺩ ﻛﺮﺩﻥ ﺑﻪ ﺍﻳﻦ ﺑﺴﻂ ﺑﺎ ﻋﺒﺎﺭﺕ‬
‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺩﺭﺟﻪ ﭼﻬﺎﺭﻡ ﺍﻳﺠﺎﺩ ﺍﺑﻬﺎﻡ ﻣﻲﻛﻨﺪ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﺗﻮﺿﻴﺤﺎﺕ ﻣﻲﺗﻮﺍﻧﻴﻢ ﺑﮕﻮﻳﻴﻢ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ﺳﻮﻡ ﻭ‬
‫ﭼﻬﺎﺭﻡ ﺗﺎﺑﻊ ‪ sin x‬ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﺴﺎﻭﻱﺍﻧﺪ‪.‬‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺁﻥ ﺗﺎﺑﻊ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻛﻪ ﻧﺴﺒﺖ ﺑﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬
‫ﺷﻬﻮﺩ ﭘﻴﺪﺍ ﻛﻨﻴﻢ‪ ،‬ﺑﺴﻂﻫﺎﻱ ﺗﻴﻠﻮﺭ ﺍﺯ ﻣﺮﺗﺒﻪ ‪ ۹ ،۷ ،۵ ،۳ ،۱‬ﻭ ‪ ۱۱‬ﺗﺎﺑﻊ ‪ sin‬ﺭﺍ ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﺯﻳﺮ ﺭﺳﻢ ﮔﺮﺩﻳﺪﻩ ﺍﺳﺖ ﺗﺎ ﻣﺸﺨﺺ‬
‫ﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺑﺴﻂﻫﺎ ﭼﻘﺪﺭ ﺑﻪ ﺗﺎﺑﻊ ‪ sin‬ﻧﺰﺩﻳﻚﺍﻧﺪ ﻭ ﭼﮕﻮﻧﻪ ﺗﻘﺮﻳﺒﻲ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬

‫‪x3‬‬
‫‪T0,3 (x) = x −‬‬ ‫!‪3‬‬
‫)‪sin(x‬‬ ‫‪x‬‬ ‫)‪sin(x‬‬ ‫‪x‬‬

‫‪T0,1 (x) = x‬‬

‫‪x3‬‬ ‫‪x5‬‬ ‫‪x7‬‬

‫‪T0,7 (x) = x −‬‬ ‫!‪3‬‬ ‫=‬ ‫!‪5‬‬ ‫‪−‬‬ ‫!‪7‬‬
‫)‪sin(x‬‬ ‫‪x‬‬ ‫)‪sin(x‬‬ ‫‪x‬‬

‫‪x3‬‬ ‫‪x5‬‬
‫‪T0,5 (x) = x −‬‬ ‫!‪3‬‬ ‫‪+‬‬ ‫!‪5‬‬

‫‪x3‬‬ ‫‪x5‬‬ ‫‪x7‬‬ ‫‪x9‬‬ ‫‪x11‬‬

‫‪T0,11 (x) = x −‬‬ ‫!‪3‬‬ ‫=‬ ‫!‪5‬‬ ‫‪−‬‬ ‫!‪7‬‬ ‫‪+‬‬ ‫!‪9‬‬ ‫‪−‬‬ ‫!‪11‬‬
‫)‪sin(x‬‬ ‫‪x‬‬ ‫)‪sin(x‬‬ ‫‪x‬‬

‫‪x3‬‬ ‫‪x5‬‬ ‫‪x7‬‬ ‫‪x9‬‬

‫‪T0,9 (x) = x −‬‬ ‫!‪3‬‬ ‫=‬ ‫!‪5‬‬ ‫‪−‬‬ ‫!‪7‬‬ ‫‪+‬‬ ‫!‪9‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﺩﺭ ﻧﺰﺩﻳﻜﻲ ﻧﻘﻄﻪ ‪ ،α = 0‬ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ﺑﻪ ﺗﺎﺑﻊ ﻧﺰﺩﻳﻚ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻭﻟﻲ ﺑﺎ ﻓﺎﺻﻠﻪ ﮔﺮﻓﺘﻦ‬
‫ﺍﺯ ﺁﻥ‪ ،‬ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺑﺪ ﻭ ﺑﺪﺗﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﻣﻮﺿﻮﻉ ﺑﺮﺍﻱ ﺑﺴﻂﻫﺎﻱ ﺗﻴﻠﻮﺭ ﻣﺮﺍﺗﺐ ﺑﺎﻻﺗﺮ ﻫﻢ ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ ﻭ ﻓﻘﻂ ﻣﺮﺗﺒﻪ ﺑﺎﻻ ﺑﺎﻋﺚ‬
‫ﻣﻲﺷﻮﺩ ﻓﺎﺻﻠﻪ ﮔﺮﻓﺘﻦ ﺗﺎﺑﻊ ﻭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺁﻥ ﺑﺎ ﺗﺎﺧﻴﺮ ﺍﻧﺠﺎﻡ ﮔﺮﺩﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﭼﻨﻴﻦ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺣﻮﻝ‬
‫ﻧﻘﻄﻪ ‪ α‬ﺑﻪ ﺍﺯﺍﻱ ﻧﻘﺎﻁ ﻧﺰﺩﻳﻚ ﺑﻪ ‪ α‬ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺒﻲ ﺑﺮﺍﻱ ‪ f‬ﻣﻲﺑﺎﺷﺪ ﻭ ﻫﺮﭼﻪ ‪ x‬ﺍﺯ ‪ α‬ﻓﺎﺻﻠﻪ ﺑﻴﺸﺮﻱ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ‬
‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺑﻴﺸﺘﺮ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺧﻄﺎﻱ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪n‬‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺗﺎﺑﻊ ﻣﻌﺮﻓﻲ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺑﺮﺍﻱ ﺁﻥ ﻛﻪ ﺩﻗﺖ ﺍﻳﻦ ﺗﻘﺮﻳﺐ‬
‫ﺭﺍ ﺑﺴﻨﺠﻴﻢ‪ ،‬ﻻﺯﻡ ﺍﺳﺖ ﺧﻄﺎﻱ ﺗﻘﺮﻳﺐ ﺭﺍ ﻣﺸﺨﺺ ﻭ ﺗﻮﺻﻴﻒ ﻧﻤﺎﻳﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ ﻗﻀﻴﻪ ﺯﻳﺮ ﻣﻮﺳﻮﻡ ﺑﻪ ﻗﻀﻴﻪ ﺗﻴﻠﻮﺭ ﻫﻤﺮﺍﻩ ﺑﺎ‬
‫ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﻻﮔﺮﺍﻧﮋ ﺍﺭﺍﻳﻪ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶۶‬‬

‫ﻗﻀﯿﻪ ﺗﯿﻠﻮﺭ ﺑﺎ ﺑﺎﻗ ﻣﺎﻧﺪﻩ ﻻﮔﺮﺍﻧﮋ ]؟ [‬ ‫ﻗﻀﯿﻪ ‪۴−۶‬‬

‫‪ f‬ﺑﺮﺭﻭﻱ ]‪ [a, b‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ )‪ ،α ∈ (a, b‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ )‪x ∈ (a, b‬‬ ‫)‪(n+1‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ]‪ f ∈ C [a, b‬ﻭ‬
‫‪n‬‬

‫ﺩﺍﺭﻳﻢ‬
‫)‪f (x) = Tα,n (x) + Rn (x‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ Tα,n‬ﻫﻤﺎﻥ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﺣﻮﻝ ‪ α‬ﺍﺳﺖ ﻭ )‪ Rn (x‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺸﺨﺺ ﻣﻲﺷﻮﺩ‬
‫)‪(x − α)n+1 (n+1‬‬
‫=‪Rn (x) :‬‬ ‫‪f‬‬ ‫)‪(η‬‬ ‫)‪(۳.۶‬‬
‫!)‪(n + 1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻧﻘﻄﻪﺍﻱ ﻣﺎﺑﻴﻦ ‪ x‬ﻭ ‪ α‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺩﺭ ﻗﻀﻴﻪ ﻓﻮﻕ‪ Rn (x) ،‬ﺑﻪ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﻻﮔﺮﺍﻧﮋ ‪ ۶‬ﻳﺎ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ‪ ۷‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۴−۶‬ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻭ ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﺮﺍﯼ ﭼﻨﺪ ﺗﺎﺑﻊ(‬

‫◀ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f (x) = ex‬ﺣﻮﻝ ‪ α = 0‬ﻭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺍﻳﻦ ﺑﺴﻂ ﻋﺒﺎﺭﺕ ﺍﺳﺖ‬
‫‪x2‬‬ ‫‪xn‬‬ ‫‪xn+1 η‬‬
‫‪Tα,n (x) = 1 + x +‬‬ ‫‪+ ··· +‬‬ ‫‪,‬‬ ‫= )‪Rn (x‬‬ ‫‪e‬‬
‫!‪2‬‬ ‫!‪n‬‬ ‫!)‪(n + 1‬‬
‫ﺑﻪ ﻃﻮﺭﻳﻜﻪ ‪ η‬ﻋﺪﺩﻱ ﺑﻴﻦ ‪ α = 0‬ﻭ ‪ x‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫◀ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﻫﻤﺎﻥ ﺗﺎﺑﻊ ‪ f (x) = ex‬ﺣﻮﻝ ‪ α = 1‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪(x − 1)2‬‬ ‫‪(x − 1)n‬‬
‫‪Tα,n (x) = e + (x − 1)e +‬‬ ‫‪e + ··· +‬‬ ‫‪e,‬‬
‫!‪2‬‬ ‫!‪n‬‬
‫‪(x − 1)n+1 η‬‬
‫= )‪Rn (x‬‬ ‫‪e ,‬‬
‫!)‪(n + 1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺑﻴﻦ ‪ α = 1‬ﻭ ‪ x‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫◀ﻫﻤﻴﻦﻃﻮﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺍﺯ ﻣﺮﺗﺒﻪ ‪ 2n + 1‬ﺗﺎﺑﻊ )‪ f (x) = sin(x‬ﺣﻮﻝ ‪ α = 0‬ﻭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺁﻥ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪x3‬‬ ‫‪x5‬‬ ‫‪x2n+1‬‬
‫‪T2n+1 (x) = x −‬‬ ‫‪+‬‬ ‫‪− · · · + (−1)n‬‬ ‫‪,‬‬
‫!‪3‬‬ ‫!‪5‬‬ ‫!)‪(2n + 1‬‬
‫‬ ‫‬
‫‪x2n+2‬‬ ‫‪d2n+2‬‬ ‫‪x2n+2‬‬
‫= )‪Rn (x‬‬ ‫)‪sin(x‬‬ ‫=‬ ‫)‪(−1)n+1 sin(η‬‬
‫‪(2n + 2)! dx2n+2‬‬ ‫‪x=η‬‬ ‫!)‪(2n + 2‬‬

‫ﺩﺭ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ ،n‬ﺩﻭ ﻋﺎﻣﻞ ‪ (x − α)n+1‬ﻭ )‪ f (n+1) (η‬ﺗﺎﺛﻴﺮ ﮔﺬﺍﺭ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻭﺟﻮﺩ ﻋﺎﻣﻞ‬
‫‪ (x − α)n+1‬ﻧﺸﺎﻥ ﺩﻫﻨﺪﻩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻫﺮ ﭼﻪ ﻓﺎﺻﻠﻪ ‪ x‬ﺍﺯ ‪ α‬ﺑﻴﺸﺘﺮ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﺗﻘﺮﻳﺐ ﻧﻴﺰ ﺑﻴﺸﺘﺮ ﺍﺳﺖ‪ .‬ﺍﻳﻦ‬
‫ﻧﻜﺘﻪ ﺭﺍ ﺩﺭ ﺷﻜﻞﻫﺎﻱ ﻣﺮﺑﻮﻁ ﺑﻪ ﺑﺴﻂﻫﺎﻱ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ ۱‬ﺍﻟﻲ ‪ ۱۱‬ﺗﺎﺑﻊ ‪ sin‬ﻣﻼﺣﻈﻪ ﻧﻤﻮﺩﻳﻢ‪ .‬ﺑﻪ ﺍﻳﻦ ﺧﺎﻃﺮ ﺍﻳﻦ ﺧﺎﺻﻴﺖ ﺑﺴﻂ‬
‫ﺗﻴﻠﻮﺭ‪ ،‬ﺍﺻﻄﻼﺣﺎ ﮔﻔﺘﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻳﻚ ﺗﻘﺮﻳﺐ ﻣﺤﻠﻲ ‪ ۸‬ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻋﺎﻣﻞ ﺩﻭﻡ ﺩﺭ ﺧﻄﺎﻱ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ‬
‫)‪ f (n+1) (η‬ﻣﻲﺑﺎﺷﺪ ﻭ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ ‪ η‬ﻧﻘﻄﻪﺍﻱ ﻣﺎﺑﻴﻦ ‪ x‬ﻭ ‪ α‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ‪ η‬ﺗﺎﺑﻌﻲ ﺍﺯ ‪ x‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﺯ ﻃﺮﻓﻲ ﻫﻤﻮﺍﺭﻩ‬
‫ﺍﻳﻦ ﺍﻣﻜﺎﻥ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ ﻛﻪ ﺑﺘﻮﺍﻥ ‪ η‬ﺭﺍ ﺩﻗﻴﻘﺎ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ ﻧﻤﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻳﻦ ﻋﺎﻣﻞ ﺩﺭ ﺟﻤﻠﻪ ﺑﺎﻗﻲ ﻣﺎﻧﺪﻩ ﺭﺍ‬
‫ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺩﺭ ﻋﻤﻞ ﺑﺮﺍﻱ ﺗﺤﻠﻴﻞ ﺧﻄﺎ‪ ،‬ﻛﺮﺍﻥﻫﺎﻱ )‪ f (n+1) (η‬ﻣﻮﺭﺩ ﺍﺳﺘﻔﺎﺩﻩ ﻗﺮﺍﺭ ﻣﻲﮔﻴﺮﺩ‪ .‬ﻣﺜﺎﻝﻫﺎﻱ ﺯﻳﺮ‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬
‫‪6 Lagrange‬‬ ‫‪reminder term‬‬ ‫‪8 Local‬‬ ‫‪approximation‬‬
‫‪7 Truncation‬‬ ‫‪error‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۶۷‬‬ ‫ﺑﺴﻂ ﺗﯿﻠﻮﺭ‬ ‫ﻓﺼﻞ ﺷﺸﻢ‪:‬‬

‫ﻣﺜﺎﻝ ‪)۵−۶‬ﺗﻘﺮﯾﺐ ﺑﺎ ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﻭ ﺗﺨﻤﯿﻦ ﺍﻧﺪﺍﺯﻩ ﺧﻄﺎ(‬

‫ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n = 3‬ﺗﺎﺑﻊ ‪ f (x) = ex‬ﺣﻮﻝ ‪ α = 0‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫‪x2‬‬ ‫‪x3‬‬
‫‪T0,3 (x) = 1 + x +‬‬ ‫‪+‬‬
‫!‪2‬‬ ‫!‪3‬‬
‫ﺍﺯ ﺑﺴﻂ ﻓﻮﻕ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺗﻘﺮﻳﺒﻲ ‪) e = e1‬ﻋﺪﺩ ﻧﭙﺮﻳﻦ( ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬
‫‪12‬‬ ‫‪13‬‬
‫‪e ≃ T0,3 (1) = 1 + 1 +‬‬ ‫‪+‬‬ ‫̄‪= 2.6‬‬
‫!‪2‬‬ ‫!‪3‬‬
‫= )‪R3 (x‬‬ ‫‪x4 η‬‬
‫‪4! e‬‬ ‫ﺣﺎﻝ ﻣﻴﺨﻮﺍﻫﻴﻢ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺧﻄﺎﻱ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﺭﺍ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪ .‬ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﺑﻪ ﺍﺯﺍﻱ ‪ x = 1‬ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ ‪ R3 (1) = 4!1 eη‬ﺍﺳﺖ‪ ،‬ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺑﻴﻦ ‪ α = 0‬ﻭ ‪ x = 1‬ﺍﺳﺖ‪ .‬ﻧﻤﻲﺗﻮﺍﻥ‬
‫ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ‪ η‬ﺭﺍ ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﻟﺬﺍ ﻣﺤﺎﺳﺒﻪ ﺩﻗﻴﻖ )‪ R3 (1‬ﻏﻴﺮ ﻣﻤﻜﻦ ﺍﺳﺖ‪ .‬ﺩﺭ ﻋﻮﺽ ﺳﻌﻲ ﻣﻲﻛﻨﻴﻢ ﻛﺮﺍﻥ ﺑﺎﻻﻱ )‪R3 (1‬‬
‫ﺭﺍ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺑﺪﺗﺮﻳﻦ ﺣﺎﻟﺖ ﺑﺮﺍﻱ )‪ R3 (1‬ﺭﺍ ﺑﻪﻋﻨﻮﺍﻥ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎ ﻣﺸﺨﺺ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫= )‪R3 (1‬‬ ‫‪1 η‬‬

‫‪4! e‬‬

‫≤‬ ‫‪1 1‬‬

‫‪4! e‬‬
‫‪1‬‬
‫<‬ ‫‪4! 2.8‬‬ ‫̄‪= 0.116‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺣﺪﺍﻛﺜﺮ ﺧﻄﺎﻱ ﻣﺮﺗﻜﺐ ﺷﺪﻩ ﺩﺭ ﺗﻘﺮﻳﺐ ﻓﻮﻕ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ̄‪ .0.116‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫̄‪e − 2.6̄ ≤ 0.116‬‬
‫ﺫﻛﺮ ﺍﻳﻦ ﻧﻜﺘﻪ ﻻﺯﻡ ﺍﺳﺖ ﻛﻪ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﺧﻄﺎﻱ ﻣﺮﺗﻜﺐ ﺷﺪﻩ ﺑﺮﺍﺑﺮ · · · ‪ e − 2.6̄ = 0.05162‬ﺍﺳﺖ ﻭ ﻣﻘﺪﺍﺭ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ‬
‫ﺩﺭ ﻓﻮﻕ‪ ،‬ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎ ﺭﺍ ﻧﺸﺎﻥ ﻣﻲﺩﻫﺪ‪.‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ ﻋﻜﺲ ﻓﺮﺍﻳﻨﺪ ﻣﺜﺎﻝ ﻗﺒﻞ ﻋﻤﻞ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۶−۶‬ﻣﺮﺗﺒﻪ ﺑﺴﻂ ﺗﯿﻠﻮﺭ ﻭ ﺍﻧﺪﺍﺯﻩ ﺧﻄﺎ(‬

‫ﻣﻴﺨﻮﺍﻫﻴﻢ ﺑﺎ ﻛﻤﻚ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f (x) = ex‬ﻣﻘﺪﺍﺭ ‪ e‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺍﺧﺘﻼﻑ ﻣﻘﺪﺍﺭ ﺗﻘﺮﻳﺒﻲ ﻭ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ‬
‫ﺣﺪﺍﻛﺜﺮ ‪ 10−4‬ﺑﺎﺷﺪ‪ .‬ﻣﺮﺗﺒﻪ ﺑﺴﻂ ﭼﻪ ﺑﺎﻳﺪ ﺑﺎﺷﺪ؟‬
‫ﺍﮔﺮ ﺍﺯ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺣﻮﻝ ‪ α = 0‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ ﺑﺮﺍﻱ ﺑﺮﺁﻭﺭﺩﻩ ﺷﺪﻥ ﻧﺘﻴﺠﻪ ﺑﺎﻳﺪ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‬
‫‪1n+1 η‬‬
‫= |)‪|Rn (1‬‬ ‫‪e < 10−4‬‬
‫!)‪(n + 1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺑﻴﻦ ‪ 0‬ﻭ ‪ 1‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﺑﺮﺁﻭﺭﺩﻩ ﺷﺪﻥ ﺷﺮﻁ ﻛﺎﻓﻲ ﺍﺳﺖ ﻗﺮﺍﺭ ﺩﻫﻴﻢ‬
‫‪1n+1‬‬
‫‪2.8 < 10−4‬‬
‫!)‪(n + 1‬‬
‫‪ .(eη‬ﺍﺯ ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‬ ‫ﭼﺮﺍ ﻛﻪ ﻳﻚ ﻛﺮﺍﻥ ﺑﺎﻻ ﺑﺮﺍﻱ ﻣﻘﺪﺍﺭ ‪ eη‬ﺑﺮﺍﺑﺮ ‪ 2.8‬ﻣﻲﺑﺎﺷﺪ )‪≤ e1 < 2.8‬‬
‫‪ .(n + 1)! > 104 ∗ 2.8 = 28000‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﺭﺍﺑﻄﻪ ﻭ ﺑﺎ ﻛﻤﻚ ﺳﻌﻲ ﻭ ﺧﻄﺎ ﻣﻘﺪﺍﺭ ‪ n‬ﺭﺍ ﺣﺪﺍﻗﻞ ‪ 7‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﻳﻢ‪.‬‬
‫ﻟﺬﺍ ﻣﻘﺪﺍﺭ ‪ e‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺎﻛﻤﻚ )‪ T0,7 (1‬ﻭ ﺑﺎ ﺩﻗﺖ ‪ 10−4‬ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪.‬‬
‫‪12‬‬ ‫‪16‬‬ ‫‪17‬‬ ‫‪685‬‬
‫‪e ≃ T0,7 (1) = 1 + 1 +‬‬ ‫‪+ ··· +‬‬ ‫‪+‬‬ ‫=‬ ‫‪= 2.71825396‬‬
‫!‪2‬‬ ‫!‪6‬‬ ‫!‪7‬‬ ‫‪252‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۶۸‬‬

‫ﺍﺯ ﺍﺳﺘﺪﻻﻝ ﺑﺎﻻ ﻧﻤﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ ﻛﻪ ‪ n = 6‬ﻣﻨﺠﺮ ﺑﻪ ﺩﻗﺖ ‪ 10−4‬ﻧﻤﻲﺷﻮﺩ‪ .‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﺎ ‪ n = 6‬ﺑﻪ ﺩﻗﺖ ‪10−4‬‬
‫ﺑﺮﺳﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺩﺭ ﺑﺎﻻ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺘﻴﻢ ﻛﻪ ‪ n = 7‬ﺗﻀﻤﻴﻦ ﻛﻨﻨﺪﻩ ﺩﻗﺖ ‪ 10−4‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺍﻳﻦ‬
‫ﻣﺜﺎﻝ ﺗﺎﺛﻴﺮ ﺧﻄﺎﻱ ﺭﻭﻧﺪﺁﻑ ﺻﺮﻑ ﻧﻈﺮ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫∗ ﻓﺮﻡﻫﺎﻱ ﺩﻳﮕﺮ ﺑﺮﺍﻱ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ‬ ‫�‬

‫ﺩﺭ ﻗﻀﻴﻪ ‪ ۴-۶‬ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﻛﻪ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﻌﺮﻓﻲ ﺷﺪ‪ ،‬ﺑﻪ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺩﺭ ﻓﺮﻡ ﻻﮔﺮﺍﻧﮋ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻋﻼﻭﻩ ﺑﺮ‬
‫ﻓﺮﻡ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺑﻪ ﻓﺮﻡﻫﺎﻱ ﺩﻳﮕﺮﻱ ﻧﻴﺰ ﺑﻴﺎﻥ ﻛﺮﺩ ﻛﻪ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﺁﻧﻬﺎ ﺍﺷﺎﺭﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﻓﺮﻡ ﺍﻧﺘﮕﺮﺍﻟﻲ ﻓﺮﻡ ﺍﻧﺘﮕﺮﺍﻟﻲ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺑﺎ ﺍﻋﻤﺎﻝ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﮕﻴﻦ ﺍﻧﺘﮕﺮﺍﻝ ‪ ۹‬ﺭﻭﻱ ﻓﺮﻡ ﻻﮔﺮﺍﻧﮋ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪.‬‬
‫ﺩﺍﺭﻳﻢ‬
‫)‪(x − α)n+1 (n+1‬‬
‫= )‪Rn (x‬‬ ‫‪f‬‬ ‫)‪(η‬‬
‫!)‪(n + 1‬‬
‫‪Z x‬‬
‫‪(x − t)n‬‬
‫)‪= f (n+1) (η‬‬ ‫‪dt.‬‬
‫‪α‬‬ ‫!‪n‬‬
‫ﺑﺎ ﺍﻋﻤﺎﻝ ﻗﻀﻴﻪ ﻣﻘﺪﺍﺭ ﻣﻴﺎﻧﮕﻴﻦ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻲﺗﻮﺍﻥ ﻧﺘﻴﺠﻪ ﮔﺮﻓﺖ‬
‫‪Z‬‬ ‫‪x‬‬
‫‪1‬‬
‫= )‪Rn (x‬‬ ‫‪(x − t)n f n+1 (t)dt‬‬
‫!‪n‬‬ ‫‪α‬‬
‫ﻓﺮﻡ ﻓﻮﻕ ﺑﻪ ﻓﺮﻡ ﺍﻧﺘﮕﺮﺍﻟﻲ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫ﻓﺮﻡ ﻛﺸﻲ ﺍﮔﺮ ]‪ f ∈ C n+1 [a, b‬ﻭ )‪ α ∈ (a, b‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ }‪ ،x ∈ (a, b) − {α‬ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺑﺴﻂ‬
‫ﺗﻴﻠﻮﺭ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺗﻮﺻﻴﻒ ﻛﺮﺩ‬
‫‪(x − η)n‬‬
‫= )‪Rn (x‬‬ ‫)‪(x − α)f (n+1) (η‬‬
‫!‪n‬‬
‫ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺑﻪ ﻓﺮﻡ ﻛﺸﻲ ‪ ۱۰‬ﺑﺮﺍﻱ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺩﺭ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﻣﺮﺗﺒﻪ ‪ n‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻧﻤﺎﺩ ﺍُﻭﻱ ﺑﺰﺭﮒ)‪ (O‬ﻣﻌﻤﻮ ًﻻ ﺑﺮﺍﻱ ﺧﻼﺻﻪ ﻧﻮﻳﺴﻲ ﺟﻤﻠﻪ ﺑﺎﻗﻲﻣﺎﻧﺪﻩ ﺑﺎ ﻧﻤﺎﺩ ﺍُﻭﻱ ﺑﺰﺭﮒ ﺗﻮﺻﻴﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ‬
‫ﻣﺜﺎﻝ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﺴﻂ ﺗﻴﻠﻮﺭ ﺗﺎﺑﻊ ‪ f‬ﺍﺯ ﻣﺮﺗﺒﻪ ‪ n‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﮔﺮﺩﺩ‬
‫‪(x − α) ′‬‬ ‫)‪(x − α)n (n‬‬ ‫‬
‫‪f (x) = f (α) +‬‬ ‫‪f (α) + · · · +‬‬ ‫‪f (α) + O (x − α)n+1‬‬
‫!‪1‬‬ ‫!‪n‬‬

‫‪9 Mean‬‬ ‫‪Value Theorem for Integrals‬‬ ‫‪10‬‬ ‫‪Cauchy‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۷‬‬
‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬
‫ﻓﺼﻞ ﻫﻔﺘﻢ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۱۷۰‬‬ ‫ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۷‬‬
‫‪۱۷۱‬‬ ‫ﺩﺭﻭﻧﻴﺎﺏ ﻳﻚ ﺗﺎﺑﻊ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۷‬‬
‫‪۱۷۳‬‬ ‫ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۱.۷‬‬
‫‪۱۷۵‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۷‬‬
‫‪۱۷۵‬‬ ‫ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۷‬‬
‫‪۱۷۷‬‬ ‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۷‬‬
‫‪۱۷۹‬‬ ‫ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۷‬‬
‫‪۱۸۰‬‬ ‫ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۷‬‬
‫‪۱۸۰‬‬ ‫ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۳.۷‬‬
‫‪۱۸۵‬‬ ‫ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻤﺪﺟﻤﻠﻪﺍﻱ ‪. . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۷‬‬
‫‪۱۸۹‬‬ ‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺁﻥ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۷‬‬
‫‪۱۸۹‬‬ ‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﻻﮔﺮﺍﻧﮋ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۵.۷‬‬
‫‪۱۹۱‬‬ ‫ﻧﺘﺎﻳﺠﻲ ﺍﺯ ﻗﻀﻴﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۵.۷‬‬
‫‪۱۹۲‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۶.۷‬‬
‫‪۱۹۲‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . .‬‬ ‫‪۱.۶.۷‬‬
‫‪۱۹۳‬‬ ‫ﻣﻘﺎﻳﺴﻪ ﭘﺎﻳﺪﺍﺭﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ ‪. . . . . . . . . . . .‬‬ ‫‪۲.۶.۷‬‬
‫‪۱۹۳‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۷.۷‬‬
‫‪۱۹۳‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻨﺤﻨﻴﻬﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۸.۷‬‬

‫‪۱۶۹‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۷۰‬‬

‫ﺩﺭﻭﻧﻴﺎﺑﻲ ‪ ۱‬ﻳﻜﻲ ﺍﺯ ﻣﻔﺎﻫﻴﻢ ﻣﻬﻢ ﻭ ﭘﺮﻛﺎﺭﺑﺮﺩ ﺩﺭ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻓﺼﻞ‪ ،‬ﺩﺭ ﺍﺑﺘﺪﺍ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺳﭙﺲ ﺭﻭﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺗﻤﺮﻛﺰ ﻣﻲﺷﻮﺩ ﻭ ﭼﻨﺪﻳﻦ ﺭﻭﺵ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻭ ﺧﻄﺎﻱ‬
‫ﺑﺮﺷﻲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻧﻴﺰ ﺑﺮﺭﺳﻲ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ‬ ‫‪۱.۷‬‬

‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ‪ ،‬ﻧﻘﺎﻁ‬
‫) ‪(x0 , y0 ) , (x1 , y1 ) , . . . , (xn , yn‬‬ ‫)‪(۱.۷‬‬
‫ﻣﻮﺳﻮﻡ ﺑﻪ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ ۲‬ﺩﺍﺩﻩ ﺷﺪﻩﺍﻧﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪xi‬ﻫﺎ ﻣﺘﻤﺎﻳﺰ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﺎﺑﻌﻲ ﻣﺎﻧﻨﺪ )‪ p(x‬ﺭﺍ ﺑﻴﺎﺑﻴﻢ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ‬
‫ﻓﻮﻕ ﺑﮕﺬﺭﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪p(xi ) = yi , i = 0, . . . , n.‬‬
‫ﺩﺭﻭﻧﻴﺎﺑﻲ ‪۴‬‬ ‫ﺗﺎﺑﻊ ‪ p‬ﺑﻪ ﺻﻮﺭﺕ ﻓﻮﻕ ﺑﻪ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ‪ ۳‬ﻧﻘﺎﻁ )‪ (۱.۷‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﮔﻔﺘﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ‪ p‬ﻧﻘﺎﻁ )‪ (۱.۷‬ﺭﺍ‬
‫ﻣﻲﻛﻨﺪ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﻧﻘﺎﻁ‬
‫)‪(1, 2), (2, 4), (3, 5), (4, 3‬‬
‫ﺭﺍ ﺩﺭ ﺻﻔﺤﻪ ﺩﺭﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺗﺎﺑﻊ‬
‫‪p(x) := − 13 x3 + 32 x2 − 16 x + 1‬‬
‫ﺍﺯ ﻛﻠﻴﻪ ﻧﻘﺎﻁ ﻓﻮﻕ ﻣﻲﮔﺬﺭﺩ‪ .‬ﻟﺬﺍ ﺗﺎﺑﻊ )‪ p(x‬ﻳﻚ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﻱ ﻧﻘﺎﻁ ﻓﻮﻕ ﺍﺳﺖ‪ .‬ﺑﻪ ﺑﻴﺎﻥ ﺩﻳﮕﺮ‪ ،‬ﺗﺎﺑﻊ )‪ p(x‬ﻧﻘﺎﻁ ﻓﻮﻕ ﺭﺍ‬
‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ‪ .‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪6‬‬ ‫‪6‬‬
‫)‪(3, 5‬‬ ‫)‪(3, 5‬‬
‫‪5‬‬ ‫‪5‬‬
‫)‪(2, 4‬‬ ‫)‪(2, 4‬‬
‫‪4‬‬ ‫‪4‬‬
‫)‪(4, 3‬‬ ‫)‪(4, 3‬‬
‫‪3‬‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ‬ ‫‪3‬‬
‫)‪(1, 2‬‬ ‫)‪(1, 2‬‬
‫‪2‬‬ ‫‪2‬‬

‫‪1‬‬ ‫‪1‬‬
‫‪p(x) = − 13 x3 + 32 x2 − 16 x + 1‬‬

‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬

‫‪1 Interpolation‬‬ ‫‪3 Interpolating‬‬ ‫‪function‬‬

‫‪2 Nodes‬‬ ‫‪4 Interpolate‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷۱‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫‪6‬‬ ‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﻓﻮﻕ ﺍﺯ ﻧﻮﻉ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬

‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬ ‫ﺍﺳﺖ ﻭ ﺍﻟﺒﺘﻪ ﺩﺭﻭﻧﻴﺎﺏﻫﺎﻱ ﺩﻳﮕﺮﻱ ﺍﺯ ﻧﻴﺰ ﺑﺮﺍﻱ ﻧﻘﺎﻁ ﻗﻮﻕ‬
‫)‪(3, 5‬‬ ‫)‪p(x‬‬ ‫ﻭﺟﻮﺩ ﺩﺍﺭﻧﺪ ﻛﻪ ﺍﺯ ﻧﻮﻉ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻧﻤﻲﺑﺎﺷﻨﺪ‪.‬‬
‫‪5‬‬
‫)‪r(x‬‬ ‫ﺍﺻﻮﻻ ﺑﻲﻧﻬﺎﻳﺖ ﺗﺎﺑﻊ ﺩﻳﮕﺮ ﻣﻲﺗﻮﺍﻥ ﻳﺎﻓﺖ ﻛﻪ ﺍﺯ ﭼﻬﺎﺭ‬
‫)‪(2, 4‬‬
‫)‪s(x‬‬ ‫ﻧﻘﻄﻪ ﻗﻮﻕ ﻣﻲﮔﺬﺭﻧﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺑﻲﻧﻬﺎﻳﺖ ﺗﺎﺑﻊ‬
‫‪4‬‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﺩﻳﮕﺮ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﭼﻬﺎﺭ ﻧﻘﻄﻪ ﻓﻮﻕ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪.‬‬
‫)‪(4, 3‬‬ ‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺗﻮﺍﺑﻊ‬
‫‪3‬‬
‫‪1 13x2 − 59x + 10‬‬
‫=‪r(x) :‬‬ ‫‪,‬‬
‫)‪(1, 2‬‬ ‫‪2‬‬ ‫‪2x − 11‬‬
‫‪2‬‬ ‫‪1‬‬ ‫
‬ ‫

‬
‫=‪s(x) :‬‬ ‫‪7 − 3 sin π2 x − cos‬‬ ‫‪π‬‬
‫‪2x‬‬
‫‪2‬‬
‫ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪ .‬ﺩﺭ ﺷﻜﻞ ﺭﻭﺑﺮﻭ‪ ،‬ﻫﺮ ﺳﻪ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ‬
‫‪1‬‬
‫ﻓﻮﻕ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﺑﺮﺍﻱ ﻳﻚ ﻣﺠﻤﻮﻋﻪ‬
‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬ ‫ﺍﺯ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‪ ،‬ﻣﺎﻧﻨﺪ )‪ ،(۱.۷‬ﺑﻲﻧﻬﺎﻳﺖ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﻭﺟﻮﺩ ‪6‬‬
‫ﺩﺍﺭﺩ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﻣﺤﺪﻭﺩ ﺑﻪ ﺧﺎﻧﻮﺍﺩﻩ ﺧﺎﺻﻲ‬
‫ﺍﺯ ﺗﻮﺍﺑﻊ ﻣﻲﻛﻨﻴﻢ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﻣﻨﺤﺼﺮ ﺑﻪ‬
‫ﻓﺮﺩ ﺷﻮﺩ‪ .‬ﻣﻲﺗﻮﺍﻥ ﺍﻳﻦ ﺧﺎﻧﻮﺍﺩﻩ ﺭﺍ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺑﺎ ﺩﺭﺟﻪﺍﻱ ﻣﺸﺨﺺ ﻭ ﻳﺎ ﺗﻮﺍﺑﻊ ﻛﺴﺮﻱ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺑﺴﺘﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ﺍﻳﻦ‬
‫ﺧﺎﻧﻮﺍﺩﻩ ﭼﮕﻮﻧﻪ ﺍﻧﺘﺨﺎﺏ ﮔﺮﺩﺩ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺗﻮﺍﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﺩﺳﺘﻪ ﺑﻨﺪﻱ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺯﻳﺮ‪ ،‬ﺑﺮﺧﻲ ﺍﺯ ﺍﻧﻮﺍﻉ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺁﻭﺭﺩﻩ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺧﻄﻲ ‪۵‬‬ ‫• ﺩﺭﻭﻧﻴﺎﺑﻲ ﺗﻜﻪﺍﻱ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪۶‬‬ ‫• ﺩﺭﻭﻧﻴﺎﺑﻲ‬

‫• ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻣﺜﻠﺜﺎﺗﻲ ‪) ۷‬ﺩﺭﻭﻧﻴﺎﺑﻲ ﻓﻮﺭﻳﻪ ‪(۸‬‬

‫ﺍﺳﭙﻼﻳﻨﻲ ‪۹‬‬ ‫• ﺩﺭﻭﻧﻴﺎﺑﻲ‬
‫ﻛﺴﺮﻱ ‪۱۰‬‬ ‫• ﺩﺭﻭﻧﻴﺎﺑﻲ‬

‫ﺩﺭﻭﻧﻴﺎﺏ ﻳﻚ ﺗﺎﺑﻊ‬ ‫�‬

‫ﺗﺎﻛﻨﻮﻥ ﺩﺭ ﺧﺼﻮﺹ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪ ﻧﻘﻄﻪ ﺩﺭ ﺻﻔﺤﻪ ﺻﺤﺒﺖ ﺷﺪ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺗﺎﺑﻊ ﻣﻌﺮﻓﻲ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺗﺎﺑﻊ‬
‫‪ f‬ﺩﺭ ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺗﺎﺑﻌﻲ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ‬
‫)) ‪(x0 , f (x0 )), . . . , (xn , f (xn‬‬
‫ﻣﻲﮔﺬﺭﺩ ﺑﻪ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫‪5 Piecewise‬‬‫‪linear interpolation‬‬ ‫‪8 Fourier‬‬‫‪interpolation‬‬
‫‪6 Polynomial‬‬ ‫‪interpolation‬‬ ‫‪9 Spline‬‬‫‪interpolation‬‬
‫‪7 Trigonometric interpolation‬‬ ‫‪10 Rational Interpolation‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۷۲‬‬

‫ﻣﺜﺎﻝ ‪)۱−۷‬ﺩﺭﻭﻧﯿﺎﺏ ﺗﺎﺑﻊ ‪ sin‬ﻣﺒﺘﻨ ﺑﺮ ﭘﻨ ﻧﻘﻄﻪ(‬

‫ﺗﺎﺑﻊ )‪ f (x) = sin(x‬ﻭ ﻧﻘﺎﻁ‬

‫‪π‬‬ ‫‪3π‬‬
‫= ‪x0 = 0, x1 = , x2 = π, x3‬‬ ‫‪, x4 = 2π,‬‬
‫‪2‬‬ ‫‪2‬‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻳﻚ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﻓﻮﻕ ﺑﺮﺍﻱ )‪ f (x) = sin(x‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﺎﺑﻌﻲ ﻛﻪ ﻧﻘﺎﻁ ﺯﻳﺮ ﺭﺍ‬
‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ‪:‬‬
‫‪2 , −1), (2π, 0).‬‬
‫‪(0, 0), ( π2 , 1), (π, 0), ( 3π‬‬ ‫)‪(۲.۷‬‬
‫‪ p(x) := 3π8 3 x3 − π82 x2 + 3π‬ﻧﻘﺎﻁ ﻓﻮﻕ ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ‪ p‬ﻳﻚ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﻱ‬
‫‪16‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﺗﺎﺑﻊ ‪x‬‬
‫‪ sin‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , . . . , x4‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ‪ ،‬ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ‪ ،p‬ﺑﻪ ﻫﻤﺮﺍﻩ ﺗﺎﺑﻊ ‪ sin‬ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ‬
‫ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ )‪ p(x‬ﻭ ﺗﺎﺑﻊ ‪ sin‬ﻫﻢﺩﻳﮕﺮ ﺭﺍ ﺩﺭ ﻧﻘﺎﻁ )‪ (۲.۷‬ﻗﻄﻊ ﻣﻲﻛﻨﻨﺪ‪.‬‬

‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬
‫)‪sin(x‬‬ ‫‪+1‬‬
‫‪8‬‬
‫‪3π 3 x‬‬
‫‪3‬‬
‫‪−‬‬ ‫‪8 2‬‬
‫‪π2 x‬‬ ‫‪+‬‬ ‫‪16‬‬
‫‪3π x‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪3π‬‬ ‫‪2π‬‬
‫‪2‬‬ ‫‪2‬‬
‫‪−1‬‬

‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺍﮔﺮ ﺗﺎﺑﻊ ‪ f‬ﻭ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ x0 , . . . , xn‬ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﺑﻲﻧﻬﺎﻳﺖ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺯ ﺟﻤﻠﻪ‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‪ ،‬ﻛﺴﺮﻱ‪ ،‬ﻣﺜﻠﺜﺎﺗﻲ ﻭ ﻏﻴﺮﻩ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ x0 , . . . , xn‬ﺍﺭﺍﻳﻪ ﻛﺮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﻫﺮﭼﻨﺪ ﻛﻪ‬
‫ﻧﻮﻉ ﻭ ﺿﺎﺑﻄﻪ ﺍﻳﻦ ﺩﺭﻭﻧﻴﺎﺏﻫﺎ ﺑﺎ ﻫﻢ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪ ،‬ﺍﻣﺎ ﺗﻤﺎﻣﻲ ﺁﻥﻫﺎ ﺍﺯ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﻣﻲﮔﺬﺭﻧﺪ‪.‬‬

‫■ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺑﺮﻭﻥﻳﺎﺑﻲ‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺁﻥ ﺗﺎﺑﻊ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺍﮔﺮ ‪ p‬ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ‬
‫ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ p‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ f‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺑﻪ ﻃﻮﺭ ﺧﺎﺹ ﺍﮔﺮ ‪ x‬ﻋﺪﺩﻱ ﺣﻘﻴﻘﻲ‬
‫ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ )‪ p(x‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﺨﻤﻴﻨﻲ ﺍﺯ ‪ f‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺣﺎﻝ ﺍﮔﺮ ‪ x‬ﺑﻴﻦ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻭﺍﻗﻊ ﺑﺎﺷﺪ ﺍﺯ ﻭﺍﮊﻩ ﺩﺭﻭﻧﻴﺎﺑﻲ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﻘﺎﺑﻞ ﺍﮔﺮ ‪ x‬ﺑﻴﻦ ﻫﻴﭻ ﺩﻭ ﻧﻘﻄﻪ ﮔﺮﻩﺍﻱ ﻧﺒﺎﺷﺪ‪ ،‬ﺍﺯ ﻭﺍﮊﻩ ﺑﺮﻭﻧﻴﺎﺑﻲ ‪ ۱۱‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺑﻴﺎﻥ ﺩﻗﻴﻖﺗﺮ‪،‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ]‪ [a, b‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﺑﺎﺯﻩ ﺷﺎﻣﻞ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﺎﺷﺪ‪ ،‬ﻳﻌﻨﻲ‬
‫‪a := min{x0 , . . . , xn },‬‬ ‫‪b := max{x0 , . . . , xn },‬‬
‫ﺣﺎﻝ ﺍﮔﺮ ]‪ x ∈ [a, b‬ﻭ )‪ f (x‬ﺭﺍ ﺑﺎ )‪ p(x‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﻧﺠﺎﻡ ﺩﺍﺩﻩﺍﻳﻢ‪ .‬ﻭﻟﻲ ﺍﮔﺮ ]‪ x ∈/ [a, b‬ﻭ‬
‫)‪ f (x‬ﺭﺍ ﺑﺎ )‪ p(x‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺑﺮﻭﻧﻴﺎﺑﻲ ﺍﻧﺠﺎﻡ ﺩﺍﺩﻩﺍﻳﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﺑﺮﻭﻧﻴﺎﺑﻲ ﺍﺯ‬
‫ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪11 Extrapolation‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷۳‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬

‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬
‫‪16 3‬‬ ‫‪50‬‬
‫ﺗﺎﺑﻊ ‪f‬‬ ‫=‪p(x) :‬‬ ‫‪x − 18x2 + x − 2‬‬
‫‪3‬‬ ‫‪3‬‬
‫ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ‬ ‫ﻳﻚ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﻱ ﺗﺎﺑﻊ‬
‫‪f (x) := x ∗ sin(πx) + 2‬‬
‫ﺑﺎﺯﻩ ]‪[0.5, 2‬‬ ‫ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ 12 , 1, 32 , 2‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﻛﻮﭼﻚﺗﺮﻳﻦ ﺑﺎﺯﻩ ﺷﺎﻣﻞ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺣﺎﻝ ﺗﻘﺮﻳﺐ‬
‫)‪ f (0.75‬ﺑﺎ )‪ p(0.75‬ﻋﻤﻞ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭ‬
‫ﺣﺎﻟﻲ ﻛﻪ ﺗﻘﺮﻳﺐ )‪ f (0‬ﺑﺎ )‪ p(0‬ﺑﺮﻭﻧﻴﺎﺑﻲ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺷﻜﻞ ﺭﻭﺑﺮﻭ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺩﺭ‬
‫ﺑﺮﻭﻧﻴﺎﺑﻲ‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ‬ ‫ﺑﺮﻭﻧﻴﺎﺑﻲ‬ ‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﺩﻗﺖ ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺐ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺮﻭﻧﻴﺎﺑﻲ ﺩﻗﺖ‬
‫ﺗﻘﺮﻳﺐ ﻣﻨﺎﺳﺐ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﺧﺼﻮﺹ ﻭﻗﺘﻲ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬
‫ﺩﻭﺭ ﻣﻲﺷﻮﻳﻢ‪ .‬ﺍﺻﻮﻻ ﺗﻘﺮﻳﺐ ﺑﺮﻭﻧﻴﺎﺑﻲ ﻭﻗﺘﻲ ﻣﻨﺎﺳﺐ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺑﺎﺯﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻧﺰﺩﻳﻚ ﺑﺎﺷﻴﻢ‪.‬‬

‫ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ‬ ‫�‬

‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﺩﺍﺭﺍﻱ ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺯﻳﺎﺩﻱ ﻣﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺣﻞ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻣﺴﺎﻳﻞ ﻇﺎﻫﺮ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺍﺯ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺮﺍﻱ‬
‫ﻛﻮﭼﻚ ﻭ ﺑﺰﺭﮒ ﻛﺮﺩﻥ ‪ ۱۲‬ﺗﺼﺎﻭﻳﺮ ﺩﻳﺠﻴﺘﺎﻝ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺩﺍﺭﺍﻱ ﻛﺎﺭﺑﺮﺩﻫﺎﻳﻲ ﺩﺭ ﻓﺸﺮﺩﻩ ﻛﺮﺩﻥ ﻳﻚ‬
‫ﺳﻴﮕﻨﺎﻝ ﻳﺎ ﺗﺼﻮﻳﺮ ﺩﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﺁﻧﺎﻟﻴﺰ ﻋﺪﺩﻱ‪ ،‬ﺍﺯ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﻴﺸﺘﺮ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﻣﻮﺍﺭﺩﻱ ﺍﺷﺎﺭﻩ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺗﻘﺮﻳﺐ ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ‬

‫ﮔﺎﻫﻲ ﺑﺎ ﺗﻮﺍﺑﻌﻲ ﺳﺮﻭﻛﺎﺭ ﺩﺍﺭﻳﻢ ﺑﻪ ﺻﻮﺭﺕ ﺟﺪﻭﻟﻲ ﻣﺸﺨﺺ ﺷﺪﻩﺍﻧﺪ‪ .‬ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ‪ ،‬ﺗﻮﺍﺑﻌﻲ ﻫﺴﺘﻨﺪ ﻛﻪ ﺩﺍﺭﺍﻱ ﺿﺎﺑﻄﻪ ﻧﻤﻲﺑﺎﺷﻨﺪ‬
‫ﻭ ﻓﻘﻂ ﻣﻘﺪﺍﺭ ﺁﻥﻫﺎ ﺩﺭ ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﻧﻘﻄﻪ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻣﻌﻤﻮﻻ ﺁﻣﺎﺭﮔﻴﺮﻱﻫﺎ ﻭ ﻧﺘﺎﻳﺞ ﺁﺯﻣﺎﻳﺸﮕﺎﻫﻲ ﺑﻪ ﺻﻮﺭﺕ ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﻣﺜﻼ‪ ،‬ﻧﺘﺎﻳﺞ ﺣﺎﺻﻞ ﺍﺯ ﺳﺮﺷﻤﺎﺭﻱ ﺟﻤﻌﻴﺖ ﻛﺸﻮﺭ‪ ،‬ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺗﺎﺑﻊ ﺟﻤﻌﻴﺖ ﺭﺍ ﺩﺭ ﺑﺮﺧﻲ )ﺳﺎﻝﻫﺎ(‬
‫ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻟﻲ ﺩﻳﮕﺮ ﻛﻪ ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺍﻳﺠﺎﺩ ﻣﻲﺷﻮﺩ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺩﺭ ﺁﺯﻣﺎﻳﺸﮕﺎﻩ‪ ،‬ﻳﻚ ﺗﻜﻪ ﻳﺦ ﺭﺍ ﺩﺭ‬
‫ﻳﻚ ﻇﺮﻑ ﺁﺏ ﺑﺎ ﺩﻣﺎﻱ ﻣﺸﺨﺺ ﻣﻲﺍﻧﺪﺍﺯﻳﻢ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺑﺮﺭﺳﻲ ﻛﻨﻴﻢ ﻛﻪ ﺩﻣﺎﻱ ﺁﺏ ﺩﺭ ﻃﻮﻝ ﺯﻣﺎﻥ ﭼﮕﻮﻧﻪ ﺗﻐﻴﻴﺮ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ‬
‫ﻭﺍﻗﻊ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﺎﺑﻌﻲ ﺭﺍ ﺍﺭﺍﻳﻪ ﺩﻫﻴﻢ ﻛﻪ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﻣﺎﻱ ﺁﺏ ﺭﺍ ﺩﺭ ﺯﻣﺎﻥﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﻣﺸﺨﺺ ﻛﻨﺪ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻛﺎﺭ ﺩﺭ ﺁﺯﻣﺎﻳﺸﮕﺎﻩ‬
‫ﺍﺯ ﻳﻚ ﺩﻣﺎﺳﻨﺞ ﻭ ﻳﻚ ﺯﻣﺎﻥﺳﻨﺞ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻧﻜﺘﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻣﺎ ﻗﺎﺩﺭ ﻧﻤﻲﺑﺎﺷﻴﻢ ﺩﻣﺎﻱ ﺁﺏ ﺭﺍ ﺩﺭ ﺗﻤﺎﻣﻲ ﺯﻣﺎﻥﻫﺎ ﺍﻧﺪﺍﺯﻩ‬
‫ﺑﮕﻴﺮﻳﻢ‪ .‬ﺑﻠﻜﻪ ﻓﻘﻂ ﻣﻲﺗﻮﺍﻧﻴﻢ ﺩﺭ ﺯﻣﺎﻥﻫﺎﻱ ﺧﺎﺻﻲ ﺩﻣﺎﻱ ﺁﺏ ﺭﺍ ﺍﻧﺪﺍﺯﻩ ﻭ ﮔﺰﺍﺭﺵ ﺩﻫﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺣﺎﺻﻞ ﺁﺯﻣﺎﻳﺶ ﻣﺎ ﻳﻚ‬
‫ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺩﻣﺎﻱ ﺁﺏ ﺭﺍ ﺩﺭ ﺯﻣﺎﻥﻫﺎﻱ ﺧﺎﺻﻲ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺩﺍﺭﺍﻱ ﺿﺎﺑﻄﻪ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﺑﻠﻜﻪ ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺭ ﭼﻨﺪ ﻧﻘﻄﻪ ﻣﺸﺨﺺ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ‬
‫ﺩﻳﮕﺮ‪،‬ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻌﺪﺍﺩﻱ ﺯﻭﺝ ﻣﺮﺗﺐ ﺑﻪ ﺻﻮﺭﺕ‪:‬‬
‫‪(x0 , f0 ), (x1 , f1 ), . . . , (xn , fn ),‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪xi‬ﻫﺎ ﻣﺠﺰﺍ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬

‫‪12 Zoom‬‬ ‫‪in/Zoom out‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۷۴‬‬

‫‪xi‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫···‬ ‫‪xn‬‬

‫) ‪f (xi‬‬ ‫‪f0‬‬ ‫‪f1‬‬ ‫···‬ ‫‪fn‬‬

‫ﺣﺎﻝ ﺍﮔﺮ ﺑﺨﻮﺍﻫﻴﻢ ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺭﺍ ﺑﺎ ﻳﻚ ﺗﺎﺑﻊ ﻛﻪ ﺩﺍﺭﺍﻱ ﺿﺎﺑﻄﻪ ﺍﺳﺖ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻛﺮﺩ‪ .‬ﻣﺜﻼ‪ ،‬ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺟﻤﻌﻴﺖ ﻛﺸﻮﺭ ﺭﺍ ﺑﺎ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )ﻭ ﻳﺎ ﻫﺮ ﺧﺎﻧﻮﺍﺩﻩ ﺍﺯ ﺗﻮﺍﺑﻊ ﺩﻳﮕﺮ( ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﻴﻢ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‬
‫ﺟﻤﻌﻴﺖ ﻛﺸﻮﺭ ﺭﺍ ﺩﺭ ﻫﺮ ﺯﻣﺎﻥ )ﻭ ﻧﻪ ﻓﻘﻂ ﺩﺭ ﺯﻣﺎﻥﻫﺎﻱ ﺳﺮﺷﻤﺎﺭﻱ( ﻣﻲﺗﻮﺍﻧﻴﻢ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪.‬‬

‫■ ﺗﻘﺮﻳﺐ ﻳﻚ ﺗﺎﺑﻊ ﺑﺎ ﺗﺎﺑﻊ ﺩﻳﮕﺮﻱ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ﺁﻥﻫﺎ ﻓﻘﻂ ﺑﻪ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﻧﻴﺎﺯ ﺩﺍﺭﺩ‬
‫ﻣﺎﺷﻴﻨﻲ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ ﻛﻪ ﺗﻨﻬﺎ ﻗﺎﺩﺭ ﺍﺳﺖ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺟﻤﻊ‪ ،‬ﺗﻔﺮﻳﻖ‪ ،‬ﺿﺮﺏ ﻭ ﺗﻘﺴﻴﻢ ﺭﺍ ﺍﻧﺠﺎﻡ ﺩﻫﺪ‪ .‬ﺑﺎ ﺍﻳﻦ ﻣﺎﺷﻴﻦ‬
‫ﻣﻲﺗﻮﺍﻥ ﺗﻮﺍﺑﻌﻲ ﺭﺍ ﺍﺭﺯﻳﺎﺑﻲ ﻭ ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ ﻛﻪ ﺿﺎﺑﻄﻪ ﺁﻥﻫﺎ ﻓﻘﻂ ﺍﺯ ﭼﻬﺎﺭ ﻋﻤﻞ ﺍﺻﻠﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﺪ‪ .‬ﻣﺜﻼ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻭ‬
‫ﻳﺎ ﺗﻮﺍﺑﻊ ﻛﺴﺮﻱ‪ .‬ﺍﻣﺎ ﻣﺤﺎﺳﺒﻪ ﻣﺴﺘﻘﻴﻢ ﺗﻮﺍﺑﻊ ﭘﻴﭽﻴﺪﻩﺗﺮ ﻣﺎﻧﻨﺪ ﺗﻮﺍﺑﻊ ﻣﺜﻠﺜﺎﺗﻲ ﻭ ﻧﻤﺎﻳﻲ ﺑﺎ ﺍﻳﻦ ﭼﻨﻴﻦ ﻣﺎﺷﻴﻨﻲ ﻏﻴﺮﻣﻤﻜﻦ ﺍﺳﺖ‪ .‬ﺩﺭ‬
‫ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺗﻮﺍﺑﻊ ﭘﻴﭽﻴﺪﻩ ﺑﺎ ﻳﻚ ﺗﺎﺑﻊ ﺳﺎﺩﻩﺗﺮ ﻣﺎﻧﻨﺪ ﻳﻚ ﺗﺎﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻳﺎ ﻛﺴﺮﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﮔﺮﺩﺩ ﻭ ﺍﻳﻦ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺑﻪ‬
‫ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺍﺻﻠﻲ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﻣﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ ﺗﻘﺮﻳﺒﻲ ﺗﻮﺍﺑﻊ ﺑﺎ ﺿﺎﺑﻄﻪ ﭘﻴﭽﻴﺪﻩ ﺭﺍ ﺗﻨﻬﺎ ﺑﺎ ﭼﻬﺎﺭ‬
‫ﻋﻤﻞ ﺍﺻﻠﻲ ﻭ ﺍﻟﺒﺘﻪ ﺑﻪ ﺻﻮﺭﺕ ﺗﻘﺮﻳﺒﻲ ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﻣﺎﺷﻴﻦﻫﺎﻱ ﺍﻣﺮﻭﺯﻱ ﻧﻴﺰ ﺍﺯ ﺍﻳﻦ ﺗﻜﻨﻴﻚ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺑﺮﺧﻲ ﺗﻮﺍﺑﻊ ﺍﺯ‬
‫ﻗﺒﻴﻞ ﺗﻮﺍﺑﻊ ﻣﺜﻠﺜﺎﺗﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫■ ﺗﻘﺮﻳﺐ ﻳﻚ ﺗﺎﺑﻊ ﭘﻴﭽﻴﺪﻩ ﺑﺎ ﻳﻚ ﺗﺎﺑﻊ ﺳﺎﺩﻩﺗﺮ‬

‫ﮔﺎﻫﻲ ﺩﺭ ﻋﻤﻞ ﺑﺎ ﺗﻮﺍﺑﻌﻲ ﺳﺮﻭﻛﺎﺭ ﺩﺍﺭﻳﻢ ﻛﻪ ﺩﺍﺭﺍﻱ ﺿﺎﺑﻄﻪﺍﻱ ﭘﻴﭽﻴﺪﻩ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ﺁﻥﻫﺎ ﺩﺭ ﻳﻚ ﻧﻘﻄﻪ‬
‫ﭘﻴﭽﻴﺪﻩ ﻭ ﭘﺮﻫﺰﻳﻨﻪ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻋﻴﻦ ﺣﺎﻝ ﻣﺎ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ ﺍﻳﻦ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻧﻘﺎﻁ ﺯﻳﺎﺩﻱ ﺍﺭﺯﻳﺎﺑﻲ ﻛﻨﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ‬
‫ﻛﻪ ﺗﺎﺑﻌﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺍﺭﻳﻢ‬
‫‪Z‬‬ ‫‪x‬‬
‫=‪f (x) :‬‬ ‫‪esin(t) dt‬‬ ‫)‪(۳.۷‬‬
‫‪0‬‬
‫ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻫﺰﺍﺭﺍﻥ ﻧﻘﻄﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0, 2π‬ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﻣﺤﺎﺳﺒﻪ ﺗﺎﺑﻊ ﺩﺭ ﻫﺮ ﻧﻘﻄﻪ‬
‫ﻣﺴﺘﻠﺰﻡ ﻳﻚ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﻛﺎﺭﻱ ﭘﺮﻫﺰﻳﻨﻪ ﺍﺳﺖ‪.‬‬
‫‪xi‬‬ ‫) ‪f (xi‬‬
‫ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﺪ )ﺑﺎ ﻫﺮ ﺯﺣﻤﺘﻲ( ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﻧﻪ ﻧﻘﻄﻪ ﺑﻪ ﺻﻮﺭﺕ ﺩﺭﺝ ﺷﺪﻩ ﺩﺭ ﺟﺪﻭﻝ‬
‫‪0‬‬ ‫‪0‬‬
‫‪π‬‬
‫ﺭﻭﺑﺮﻭ ﻣﺤﺎﺳﺒﻪ ﻛﺮﺩﻩﺍﻳﻢ‪ .‬ﻳﻚ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﻱ ﻧﻪ ﻧﻘﻄﻪ ﻣﻨﺪﺭﺝ ﺩﺭ ﺟﺪﻭﻝ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪1.164644167‬‬
‫‪4‬‬
‫‪π‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺯﻳﺮ‪:‬‬
‫‪2‬‬ ‫‪3.104379018‬‬
‫‪3π‬‬
‫‪4‬‬ ‫‪5.044113868‬‬ ‫‪p(x) := 0.04282530 x8 − 0.1027883 x7 + 0.009911807 x6 +‬‬
‫‪π‬‬ ‫‪6.208758036‬‬ ‫‪0.6121665 x5 − 2.292691 x4 + 3.956304 x3 −‬‬
‫‪5π‬‬
‫‪4‬‬ ‫‪6.761365152‬‬ ‫‪2.521976 x2 + 1.914296 x‬‬
‫‪3π‬‬
‫‪7.081842278‬‬
‫‪2‬‬
‫‪7π‬‬
‫ﺣﺎﻝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ﺗﺎﺑﻊ )‪ f (x‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‬
‫‪7.402319405‬‬
‫‪4‬‬
‫ﻭ ﺑﻪ ﺍﺯﺍﻱ ﻫﺮ ] ‪ x ∈ [0, π2‬ﻣﻘﺪﺍﺭ )‪ f (x‬ﺭﺍ ﺑﺎ )‪ p(x‬ﺗﻘﺮﻳﺐ ﺯﺩ‪.‬‬
‫‪2π‬‬ ‫‪7.954926524‬‬

‫ﺑﻪ ﻣﻨﻈﻮﺭ ﻣﻘﺎﻳﺴﻪ ﺗﺎﺑﻊ ‪ f‬ﻭ ﺩﺭﻭﻧﻴﺎﺏ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺁﻥ‪ ،‬ﻳﻌﻨﻲ )‪ ،p(x‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺍﻳﻦ ﺩﻭ ﺗﺎﺑﻊ ﺩﺭ ﻛﻨﺎﺭ ﻫﻢ ﺭﺳﻢ ﺷﺪﻩﺍﻧﺪ‪.‬‬
‫ﻫﻤﻴﻦﻃﻮﺭ ﺑﺮﺍﻱ ﻣﻘﺎﻳﺴﻪ ﺑﻬﺘﺮ ﺍﻳﻦ ﺩﻭ ﺗﺎﺑﻊ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺗﻔﺎﺿﻞ ﺍﻳﻦ ﺩﻭ ﺗﺎﺑﻊ‪ ،‬ﻳﻌﻨﻲ )‪ f (x) − p(x‬ﻧﻴﺰ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷۵‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫‪9‬‬

‫‪8‬‬

‫‪7‬‬ ‫‪+0.02‬‬

‫‪6‬‬ ‫‪x‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪3π‬‬ ‫‪π‬‬ ‫‪5π‬‬ ‫‪3π‬‬ ‫‪7π‬‬ ‫‪2π‬‬
‫‪5‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬
‫)‪f (x‬‬ ‫‪−0.02‬‬
‫‪4‬‬ ‫)‪p(x‬‬

‫‪3‬‬ ‫‪−0.04‬‬

‫‪2‬‬ ‫‪−0.06‬‬
‫‪1‬‬
‫‪−0.08‬‬ ‫)‪f (x) − p(x‬‬
‫‪x‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪3π‬‬ ‫‪π‬‬ ‫‪5π‬‬ ‫‪3π‬‬ ‫‪7π‬‬ ‫‪2π‬‬
‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪−0.10‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﺎﺑﻊ )‪ p(x‬ﺗﻘﺮﻳﺐ ﺧﻮﺑﻲ ﺑﺮﺍﻱ ﺗﺎﺑﻊ )‪ f (x‬ﺍﺳﺖ ﻭ ﺍﺯ ﻃﺮﻓﻲ ﺿﺎﺑﻄﻪ )‪ p(x‬ﺑﺴﻴﺎﺭ ﺳﺎﺩﻩﺗﺮ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ‬ ‫‪۲.۷‬‬

‫ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻳﺎﻓﺘﻦ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﮕﺬﺭﺩ‪ .‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ‬
‫ﺗﻮﺍﺑﻊ ﺳﺎﺩﻩﺍﻱ ﻫﺴﺘﻨﺪ ﻭ ﺑﻪ ﺭﺍﺣﺘﻲ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺟﻤﻊ ﻭ ﺿﺮﺏ ﻣﻘﺪﺍﺭ ﺁﻥﻫﺎ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﻟﺤﺎﻅ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺁﻥﻫﺎ ﺑﺮﺍﻱ‬
‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﺟﺬﺍﺏ ﺑﻪ ﻧﻈﺮ ﻣﻲﺭﺳﺪ‪ ،‬ﭼﺮﺍ ﻛﻪ ﺑﺎ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﻛﻤﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﻣﻲﺗﻮﺍﻥ ﺗﻮﺍﺑﻊ ﻛﻠﻲ ﻣﺎﻧﻨﺪ ‪ sin‬ﺭﺍ ﺗﻨﻬﺎ ﺑﺎ ﺟﻤﻊ ﻭ‬
‫ﺿﺮﺏ ﺑﻪ ﺻﻮﺭﺕ ﺗﻘﺮﻳﺒﻲ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪.‬‬
‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺩﺭ ﺍﺑﺘﺪﺍ ﻣﻘﺪﻣﻪﺍﻱ ﺩﺭ ﺧﺼﻮﺹ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﻭ ﺧﻮﺍﺹ ﺁﻥﻫﺎ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻭ ﺳﭙﺲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‪،‬‬
‫ﺑﻪ ﻃﻮﺭ ﺩﻗﻴﻖ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ‪ .‬ﺳﭙﺲ ﺩﺭ ﺑﺨﺶﻫﺎﻱ ﺑﻌﺪ‪ ،‬ﻧﺤﻮﻩ ﻳﺎﻓﺘﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻭ ﺧﻄﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺁﻥ ﺷﺮﺡ ﺩﺍﺩﻩ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬ ‫�‬

‫ﻳﻚ ﺗﺎﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺑﻪ ﻓﺮﻡ ﻛﻠﻲ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪a0 + a1 x + · · · + an−1 xn−1 + an xn‬‬ ‫)‪(۴.۷‬‬
‫‪ n‬ﻣﻲﺑﺎﺷﺪ‪.‬‬ ‫ﺩﺭﺟﻪ ‪۱۳‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ai‬ﻫﺎ ﺍﻋﺪﺍﺩﻱ ﺣﻘﻴﻘﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﮔﺮ ‪ ،an ̸= 0‬ﺁﻥﮔﺎﻩ ﮔﻮﻳﻴﻢ ﻛﻪ ﺗﺎﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻓﻮﻕ ﺍﺯ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺻﻔﺮ ﻳﻚ ﺗﺎﺑﻊ ﺛﺎﺑﺖ ‪ ۱۴‬ﺍﺳﺖ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﻳﻚ ﻫﻤﺎﻥ ﺗﺎﺑﻊ ﺧﻄﻲ ‪ ۱۵‬ﺍﺳﺖ ﻭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺩﻭ ﻳﻚ‬
‫ﺳﻬﻤﻲ ‪ ۱۶‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻫﻤﻪ ﺟﺎ ﭘﻴﻮﺳﺘﻪ‪ ،‬ﻣﺸﺘﻖﭘﺬﻳﺮ ﻭ ﺗﺤﻠﻴﻠﻲ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﺑﻪ ﺍﻳﻦ ﻟﺤﺎﻅ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺑﺴﻴﺎﺭ‬
‫ﻣﻄﻠﻮﺏ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﺯﻳﺮ‪ ،‬ﻣﻄﺎﻟﺒﻲ ﺩﺭ ﻣﻮﺭﺩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫• ﻫﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ‪ n‬ﺩﻗﻴﻘﺎ ﺩﺍﺭﺍﻱ ‪ n‬ﺭﻳﺸﻪ )ﺑﺎ ﻟﺤﺎﻅ ﻛﺮﺩﻥ ﻣﺮﺗﺒﻪ ﺭﻳﺸﻪ( ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺗﻤﺎﻡ ﻳﺎ ﺑﺮﺧﻲ ﺍﺯ ﺍﻳﻦ ﺭﻳﺸﻪﻫﺎ‬
‫‪13 Degree‬‬ ‫‪15 Linear‬‬ ‫‪Function‬‬
‫‪14 Constant‬‬ ‫‪Function‬‬ ‫‪16 Parabola‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۷۶‬‬

‫ﻣﺨﺘﻠﻂ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• ﻫﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ‪ n‬ﺣﺪﺍﻛﺜﺮ ﺩﺍﺭﺍﻱ ‪ n‬ﺭﻳﺸﻪ ﺣﻘﻴﻘﻲ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫• ﻫﺮ ﭼﻨﺪ ﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ n ≥ 1‬ﺩﺭ ∞ ‪ ±‬ﺑﻪ ﻣﺜﺒﺖ ﻳﺎ ﻣﻨﻔﻲ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪.‬‬

‫• ﻫﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ‪ n‬ﺣﺪﺍﻛﺜﺮ ﺩﺍﺭﺍﻱ ‪ n − 1‬ﻧﻘﻄﻪ ﻣﺎﻛﺰﻳﻤﻢ ﻳﺎ ﻣﻴﻨﻴﻤﻢ ﻣﻮﺿﻌﻲ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻫﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ‪n‬‬
‫ﺣﺪﺍﻛﺜﺮ ‪ n − 1‬ﺑﺎﺭ ﻧﻮﺳﺎﻥ ﻣﻲﻛﻨﺪ‪.‬‬

‫■ ﻧﻤﺎﻳﺶﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ‬

‫ﺩﺭ )‪ (۴.۷‬ﻳﻚ ﻓﺮﻡ ﻛﻠﻲ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﻣﻌﺮﻓﻲ ﺷﺪ‪ .‬ﻧﻜﺘﻪ ﻗﺎﺑﻞ ﺗﻮﺟﻪ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻪ‬
‫ﻓﺮﻡﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺩﻳﮕﺮﻱ ﻧﻴﺰ ﺑﻴﺎﻥ ﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺩﺭ ﺯﻳﺮ ﻧﻤﺎﻳﺶﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺩﻳﮕﺮﻱ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﺍﺭﺍﻳﻪ ﺷﺪﻩ‬
‫ﺍﺳﺖ‪:‬‬
‫‪pn (x) = a0 + a1 (x − c1 ) + a2 (x − c2 )2 + · · · + an (x − cn )n ,‬‬
‫) ‪pn (x) = a0 + a1 (x − c1 ) + a2 (x − c1 )(x − c2 ) + · · · + an (x − c1 ) . . . (x − cn‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ ci‬ﻫﺎ ﺍﻋﺪﺍﺩ ﺩﻟﺨﻮﺍﻫﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ‪ ،‬ﺑﻲﻧﻬﺎﻳﺖ ﻓﺮﻡ ﺩﻳﮕﺮ ﻧﻴﺰ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪ p4 (x) := x + 2x2 + 3x3‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺍﺭﺍﻱ ﻧﻤﺎﻳﺶﻫﺎﻱ‬
‫ﺩﻳﮕﺮ ﺯﻳﺮ ﻧﻴﺰ ﺍﺳﺖ‪:‬‬
‫‪2‬‬ ‫‪3‬‬
‫‪p4 (x) = −35 + 36 x + 29 (x − 2) + 3 (x − 3) ,‬‬
‫‪p4 (x) = −22 + 28 x + 20 (x − 1) (x − 2) + 3 (x − 1) (x − 2) (x − 3) ,‬‬
‫‪p4 (x) = 3 x (x − 2) (x − 3) − 17 x (x − 1) (x − 3) + 17 x (x − 1) (x − 2) .‬‬
‫ﺑﺮﺍﻱ ﻣﻌﺮﻓﻲ ﭼﻨﺪﭼﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺯ ﻫﺮ ﻳﻚ ﺍﺯ ﻓﺮﻡﻫﺎﻱ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﺍﻣﺎ ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺧﻮﺍﻫﻴﻢ‬
‫ﺩﻳﺪ‪ ،‬ﺑﻌﻀﻲ ﺍﺯ ﻓﺮﻡﻫﺎ‪ ،‬ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﺳﺎﺩﻩﺗﺮ ﻣﻲﻛﻨﻨﺪ‪.‬‬

‫■ ∗ ﻣﺤﺎﺳﺒﻪ ﻣﻘﺪﺍﺭ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭ ﻳﻚ ﻧﻘﻄﻪ ﻭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﻫﺮﻧﺮ‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ‪ n‬ﻭ ﺑﺎ ﻧﺎﻡ )‪ p(x‬ﺑﻪ ﻓﺮﻡ ﻛﻠﻲ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪p(x) := an xn + an−1 xn−1 + · · · + a1 x + a0 .‬‬ ‫)‪(۵.۷‬‬
‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﻌﺪﺍﺩ ﻣﺤﺎﺳﺒﺎﺕ ﻻﺯﻡ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪ p‬ﺩﺭ ﻧﻘﻄﻪ ‪ x = α‬ﺭﺍ ﻣﺸﺨﺺ ﻛﻨﻴﻢ‪ .‬ﺑﻪ ﺍﺻﻄﻼﺡ‪ ،‬ﻣﻲﺧﻮﺍﻫﻴﻢ‬
‫ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﻪ )‪ p(α‬ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻻﺯﻡ ﺍﺳﺖ ﺍﺑﺘﺪﺍ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﻣﺤﺎﺳﺒﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭ ﻳﻚ ﻧﻘﻄﻪ ﺭﺍ ﻣﺸﺨﺺ‬
‫ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺳﻪ ﺭﻭﺵ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﺍﻳﻦ ﻛﺎﺭ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺭﻭﺵ ﺍﻭﻝ‪ :‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ )‪ ،p(α‬ﺍﺑﺘﺪﺍ ﺑﺎ ‪ n−1‬ﻋﻤﻞ ﺿﺮﺏ ﻣﻘﺪﺍﺭ ‪ α‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ‪ an‬ﺭﺍ ﺩﺭ ﺁﻥ ﺿﺮﺏ‬
‫‪n‬‬

‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﺍﻧﺠﺎﻡ ‪ n‬ﺿﺮﺏ ﻣﻘﺪﺍﺭ ‪ an αn‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻫﻤﻴﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺑﺎ ‪ n − 1‬ﺿﺮﺏ ﻣﻘﺪﺍﺭ ‪ an−1 xn−1‬ﺭﺍ‬
‫ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ ﻭ ﺩﺭ ﻧﻬﺎﻳﺖ ﺑﺎ ‪ 1‬ﺿﺮﺏ ﻣﻘﺪﺍﺭ ‪ a1 x‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺗﻌﺪﺍﺩ ﻛﻞ ﺿﺮﺏﻫﺎﻱ ﻻﺯﻡ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫)‪n(n + 1‬‬
‫= ‪n + (n − 1) + · · · + 1‬‬
‫‪2‬‬
‫ﻫﻤﻴﻦﻃﻮﺭ ﺑﻪ ‪ n‬ﺟﻤﻊ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‪ ،‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ ‪ n 2+n‬ﺿﺮﺏ ﻭ ‪ n‬ﺟﻤﻊ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪ .‬ﺍﻣﺎ ﺑﻪ ﺭﺍﺣﺘﻲ ﻣﻲﺗﻮﺍﻥ‬
‫‪2‬‬

‫ﻓﻬﻤﻴﺪ ﻛﻪ‪ ،‬ﺑﺎ ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﻋﻤﻠﻴﺎﺕ ﺗﻜﺮﺍﺭﻱ‪ ،‬ﺗﻌﺪﺍﺩ ﺿﺮﺏﻫﺎ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﻛﺎﻫﺶ ﺩﺍﺩ‪ .‬ﺭﻭﺵ ﺩﻭﻡ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷۷‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﺭﻭﺵ ﺩﻭﻡ‪ :‬ﺍﻳﻦ ﺭﻭﺵ ﻫﻤﺎﻧﻨﺪ ﺭﻭﺵ ﺍﻭﻝ ﺍﺳﺖ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ‪ αi‬ﺍﺯ ﻣﺤﺎﺳﺒﺎﺕ ﻗﺒﻠﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ‬
‫ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﻭﻗﺘﻲ ﺑﺎ ﻳﻚ ﻋﻤﻞ ﺿﺮﺏ ‪ α2‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﺮﺩﻳﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ﺁﻥ ﺭﺍ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ α3‬ﻧﮕﻪ ﻣﻲ ﺩﺍﺭﻳﻢ ﺗﺎ ﺑﺘﻮﺍﻧﻴﻢ‬
‫ﺑﺎ ﺍﻧﺠﺎﻡ ﻳﻚ ﺿﺮﺏ ﺩﻳﮕﺮ ﻣﻘﺪﺍﺭ ‪ α3‬ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ‪ ،‬ﺑﺎ ﺍﻧﺠﺎﻡ ‪ n − 1‬ﺿﺮﺏ‪ ،‬ﻣﻘﺎﺩﻳﺮ ‪α2 , α3 , . . . , αn‬‬
‫ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺎ ﺍﻧﺠﺎﻡ ‪ n‬ﺿﺮﺏ ﻣﻘﺎﺩﻳﺮ ‪ an xn , an−1 xn−1 , . . . , ax‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ ﻭ ﺩﺭ ﻧﺤﺎﻳﺖ ﺑﺎ ﺍﻧﺠﺎﻡ ‪ n‬ﻋﻤﻞ‬
‫ﺟﻤﻊ ﻣﻘﺪﺍﺭ )‪ p(α‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‪ ،‬ﺩﺭ ﺭﻭﺵ ﺩﻭﻡ ﺑﻪ ‪ 2n − 1‬ﺿﺮﺏ ﻭ ‪ n‬ﺟﻤﻊ ﻧﻴﺎﺯ ﺩﺍﺭﻳﻢ‪.‬‬
‫ﺭﻭﺵ ﺳﻮﻡ)ﺭﻭﺵ ﻫﺮﻧﺮ(‪ :‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ‪ ،‬ﻛﻪ ﺑﻪ ﺭﻭﺵ ﻫﺮﻧﺮ ‪ ۱۷‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﺮﺍﻱ ﻛﺎﻫﺶ ﻫﺮﭼﻪ ﺑﻴﺸﺘﺮ ﺗﻌﺪﺍﺩ‬
‫ﺿﺮﺏﻫﺎ ﺩﺭ ﺭﻭﺵ ﺩﻭﻡ ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪ‪ .‬ﺍﻳﻦ ﺭﻭﺵ ﺑﺮ ﺍﻳﻦ ﻣﺒﻨﺎ ﺍﺳﺘﻮﺍﺭ ﺍﺳﺖ ﻛﻪ ﻣﻲﺗﻮﺍﻥ ﭼﻨﺪﭼﻤﻠﻪﺍﻱ )‪ p(x‬ﺩﺭ )‪ (۵.۷‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻧﻤﻮﺩ‬
‫))) ‪p(x) = a0 + x(a1 + x(a2 + x(a3 + · · · + x(an−1 + an x) . . .‬‬ ‫)‪(۶.۷‬‬
‫ﺣﺎﻝ ﺑﺎ ﺍﻟﻬﺎﻡ ﺍﺯ ﻓﺮﻡ ﻓﻮﻕ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ )‪ p(α‬ﺑﺎ ﺍﻧﺠﺎﻡ ﻳﻚ ﺟﻤﻊ ﻭ ﻳﻚ ﺿﺮﺏ‪ ،‬ﻣﻘﺪﺍﺭ ‪ an−1 + an α‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪.‬‬
‫ﺑﺎ ﻳﻚ ﺿﺮﺏ ﻭ ﻳﻚ ﺟﻤﻊ ﺩﻳﮕﺮ ﻣﻘﺪﺍﺭ )‪ an−2 + x(an−1 + an α‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺑﺎ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺭﻭﻧﺪ ﺩﺭ ﻧﻬﺎﻳﺖ ﺑﺎ ‪n‬‬
‫ﺟﻤﻊ ﻭ ‪ n‬ﺿﺮﺏ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﻘﺪﺍﺭ )‪ p(α‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ‪.‬‬

‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻳﺎﻓﺘﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻳﻲ ﻛﻪ ﺍﺯ ﺗﻌﺪﺍﺩﻱ ﻧﻘﻄﻪ ﺩﺍﺩﻩ ﺷﺪﻩ ﻋﺒﻮﺭ ﻛﻨﺪ‪.‬‬
‫ﺍﻣﺎ ﺳﻮﺍﻻﺗﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﺮﺍﻱ ﻫﺮ ﻣﺠﻤﻮﻋﻪ ﺍﺯ ﻧﻘﺎﻁ ﺩﺭ ﺻﻔﺤﻪ‪ ،‬ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ‬
‫ﻳﺎﻓﺖ‪ .‬ﺁﻳﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ؟ ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻻﺕ ﭘﺎﺳﺦ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺩﺭﻭﻧﯿﺎﺏ ﭼﻨﺪﺟﻤﻠەﺍﯼ‬ ‫ﻗﻀﯿﻪ ‪۱−۷‬‬
‫ﺍﮔﺮ‬
‫‪(xi , yi ), i = 0, . . . , n,‬‬
‫ﻧﻘﺎﻃﻲ ﺑﺎ ‪xi‬ﻫﺎﻱ ﻣﺠﺰﺍ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫ﺍﻟﻒ( ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻧﻘﺎﻁ ﻓﻮﻕ ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ ﻭ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ‬
‫ﺍﺳﺖ‪.‬‬
‫ﺏ( ﺑﻲﻧﻬﺎﻳﺖ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ n + 1‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ ﻓﻮﻕ ﻣﻲﮔﺬﺭﻧﺪ‪ .‬ﻓﺮﻡ ﻛﻠﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻳﻲ ﺍﺯ ﺩﺭﺟﻪ‬
‫‪ n + 1‬ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ ﻓﻮﻕ ﻣﻲﮔﺬﺭﻧﺪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪pn+1 (x) := pn (x) + c(x − x0 )(x − x1 ) . . . (x − xn ), c ̸= 0,‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ )‪ pn (x‬ﻫﻤﺎﻥ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱ ﺍﺳﺖ ﻛﻪ ﻭﺟﻮﺩﺵ ﻭ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩﻱﺍﺵ ﺩﺭ ﻗﺴﻤﺖ‬
‫)ﺍﻟﻒ( ﺛﺎﺑﺖ ﺷﺪ‪.‬‬
‫ﺝ( ﺑﻲﻧﻬﺎﻳﺖ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ ،m‬ﻛﻪ ‪ ،m > n + 1‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﻧﻘﺎﻁ ﻓﻮﻕ ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ ﻭ ﺑﻪ ﻓﺮﻡ ﻛﻠﻲ‬
‫ﺯﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‪:‬‬
‫‪pm (x) := pn (x) + qm−n (x − x0 )(x − x1 ) . . . (x − xn ),‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ qm−n‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ m − n‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪17 Horner’s‬‬ ‫‪method‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۷۸‬‬

‫ﺍﺛﺒﺎﺕ‪) .‬ﺍﻟﻒ‪ :‬ﺍﺛﺒﺎﺕ ﻭﺟﻮﺩ( ﺍﺑﺘﺪﺍ ﺑﺎ ﺍﺳﺘﻘﺮﺍ ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﻛﻪ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﻭﺟﻮﺩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭ ﺣﺎﻟﺖ‬
‫‪ n = 0‬ﺑﺪﻳﻬﻲ ﺍﺳﺖ‪ .‬ﭼﺮﺍ ﻛﻪ‪ ،‬ﺩﺭ ﺣﺎﻟﺖ ‪ n = 0‬ﻳﻚ ﻧﻘﻄﻪ ) ‪ (x0 , y0‬ﺭﺍ ﺩﺍﺭﻳﻢ ﻛﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺻﻔﺮ ‪ p0 (x) = y0‬ﺍﺯ‬
‫ﺍﻳﻦ ﻧﻘﻄﻪ ﻋﺒﻮﺭ ﻣﻲﻛﻨﺪ‪.‬‬
‫ﺣﺎﻝ ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺑﺮﺍﻱ ‪ k‬ﻧﻘﻄﻪ ‪ (xi , yi ), i = 0, . . . , k − 1‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺣﺪﺍﻛﺜﺮ ﺍﺯ ﺩﺭﺟﻪ ‪ k − 1‬ﺑﺎ ﻧﺎﻡ‬
‫)‪ pk−1 (x‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺛﺎﺑﺖ ﻛﻨﻴﻢ ﻛﻪ ﺑﺮﺍﻱ ﻧﻘﺎﻁ ‪(xi , yi ), i = 0, . . . , k‬‬
‫ﻧﻴﺰ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )‪ pk (x‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﻛﻨﻴﻢ‬
‫‪pk (x) := pk−1 (x) + β(x − x0 ) . . . (x − xk−1 ),‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫) ‪yk − pk−1 (xk‬‬
‫=‪β :‬‬ ‫‪.‬‬
‫) ‪(xk − x0 ) . . . (xk − xk−1‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺨﺮﺝ ﻛﺴﺮ ﻓﻮﻕ‪ ،‬ﻫﻴﭻﮔﺎﻩ ﺻﻔﺮ ﻧﻤﻲﺑﺎﺷﺪ‪ ،‬ﭼﺮﺍ ﻛﻪ ‪xi‬ﻫﺎ ﺭﺍ ﻣﺠﺰﺍ ﻓﺮﺽ ﻛﺮﺩﻳﻢ‪ .‬ﺣﺎﻝ‪ ،‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ‪ pk‬ﻳﻚ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ k‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﺯ ﺳﻮﻱ ﺩﻳﮕﺮ ﺑﻪ ﺍﺯﺍﻱ ‪ i = 0, . . . , k − 1‬ﺩﺍﺭﻳﻢ‪:‬‬
‫) ‪pk (xi ) =pk−1 (xi ) + β(xi − x0 ) . . . (xi − xi ) . . . (xi − xk−1‬‬
‫‪=yi + β × 0 = yi .‬‬
‫ﻫﻤﭽﻨﻴﻦ ﺩﺍﺭﻳﻢ‪:‬‬
‫) ‪pk (xk ) =pk−1 (xk ) + β(xk − x0 ) . . . (xk − xi ) . . . (xk − xk−1‬‬
‫) ‪yk − pk−1 (xk‬‬
‫‪=pk−1 (xk ) +‬‬ ‫‪(xk − x0 ) . . . (xk − xk−1 ) = yk .‬‬
‫) ‪(xk − x0 ) . . . (xk − xk−1‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )‪ pk (x‬ﻛﻪ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ k‬ﺍﺳﺖ‪ ،‬ﺍﺯ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , k‬ﻣﻲﮔﺬﺭﺩ‪.‬‬
‫)ﺍﻟﻒ‪ :‬ﺍﺛﺒﺎﺕ ﻣﻨﺤﺼﺮﺑﻪﻓﺮﺩﻱ( ﻓﺮﺽ ﻛﻨﻴﻢ ﺩﻭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )‪ pn (x‬ﻭ )‪ qn (x‬ﻣﻮﺟﻮﺩ ﺑﺎﺷﻨﺪ ﻛﻪ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪n‬‬
‫ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﻨﺪ‪ .‬ﺣﺎﻝ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪rn (x) := pn (x) − qn (x).‬‬
‫ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ )‪ r(x‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺍﺳﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺍﺭﻳﻢ‪:‬‬
‫) ‪rn (xi ) =pn (xi ) − qn (xi‬‬ ‫)‪(۷.۷‬‬
‫‪=yi − yi = 0, i = 0, 1, . . . , n.‬‬ ‫)‪(۸.۷‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﻘﺎﻁ ﻣﺘﻤﺎﻳﺰ ‪ x0 , x1 , . . . , xn‬ﺭﻳﺸﻪﻫﺎﻱ )‪ rn (x‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ )‪) rn (x‬ﻛﻪ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺣﺪﺍﻛﺜﺮ ﺍﺯ ﺩﺭﺟﻪ ‪n‬‬
‫ﺍﺳﺖ( ﺩﺍﺭﺍﻱ ‪ n + 1‬ﺭﻳﺸﻪ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﻧﺘﻴﺠﻪ ﺍﻣﻜﺎﻥ ﻧﺪﺍﺭﺩ ﻣﮕﺮ ﺁﻥ ﻛﻪ ‪ rn (x) = 0‬ﻭ ﻳﺎ )‪.pn (x) = qn (x‬‬

‫ﻃﺒﻖ ﻗﻀﻴﻪ ﻓﻮﻕ‪ ،‬ﺩﻗﻴﻘﺎ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﻛﻨﺪ‪ ،‬ﺍﻣﺎ‬
‫ﺑﻲﻧﻬﺎﻳﺖ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ‪ n‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻛﻨﺪ‪ .‬ﻃﺒﻴﻌﻲ ﺍﺳﺖ ﻛﻪ ﺭﻭﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﻣﺘﻤﺮﻛﺰ ﺷﻮﻳﻢ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺖ ﻭ ﻫﻤﭽﻨﻴﻦ ﺩﺭﺟﻪ ﻛﻮﭼﻚﺗﺮﻱ ﻧﺴﺒﺖ ﺑﻪ‬
‫ﺩﺭﻭﻧﻴﺎﺏﻫﺎﻱ ﺩﻳﮕﺮ ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﺍﺳﺘﺎ‪ ،‬ﺁﻥ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﻛﻪ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﺭﺍ ﺩﺭﻭﻧﻴﺎﺑﻲ‬
‫ﻣﻲﻛﻨﺪ ﺭﺍ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺁﻥ ﻧﻘﺎﻁ ﻣﻲﻧﺎﻣﻴﻢ ﻭ ﺁﻥ ﺭﺍ ﺑﺎ ﻧﻤﺎﺩ ‪ Pn‬ﻧﻤﺎﻳﺶ ﻣﻲﺩﻫﻴﻢ‪ .‬ﺑﺮﺍﻱ ﺗﺎﻛﻴﺪ ﺑﻴﺸﺘﺮ ﺗﻌﺮﻳﻒ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﺩﺭ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۷۹‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۷‬‬

‫ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ‪ n + 1‬ﻧﻘﻄﻪ ﺯﻳﺮ‬
‫) ‪(x0 , y0 ), (x1 , y1 ), . . . , (xn , yn‬‬
‫ﺩﺭ ﺻﻔﺤﻪ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺎﺷﻨﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪xi‬ﻫﺎ ﻣﺠﺰﺍ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ ﻓﻮﻕ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ ،n‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺍﺯ ﻧﻘﺎﻁ ﻓﻮﻕ ﺑﮕﺬﺭﺩ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻳﻲ ﻣﺎﻧﻨﺪ )‪ ،Pn (x‬ﺑﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ ﻓﻮﻕ ﻣﻮﺳﻮﻡ ﺍﺳﺖ ﻫﺮﮔﺎﻩ‬

‫ﺍﻟﻒ( ‪ Pn‬ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺑﺎﺷﺪ ﻭ‬

‫ﺏ( ‪.Pn (xi ) = yi , i = 0, 1, . . . , n‬‬

‫ﺑﻪ ﻃﻮﺭ ﻣﺸﺎﺑﻪ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻳﻚ ﺗﺎﺑﻊ ﻧﻴﺰ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻐﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫ﺗﺎﺑﻊ‬ ‫ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۷‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x0 , x1 , . . . , xn‬ﻧﻘﺎﻃﻲ ﻣﺘﻤﺎﻳﺰ ﺑﺎﺷﻨﺪ ﻭ ‪ f‬ﺗﺎﺑﻌﻲ ﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻧﻘﺎﻁ ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , x1 , . . . , xn‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ ﺯﻳﺮ‬
‫‪(x0 , f (x0 )) , (x1 , f (x1 )) , . . . , (xn , f (xn )) .‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ ،x0 , x1 , . . . , xn‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺍﺳﺖ‪ ،‬ﺑﻪ‬
‫ﻃﻮﺭﻱ ﻛﻪ ﺑﺎ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﺎﻁ ‪ x0 , x1 , . . . , xn‬ﻣﻄﺎﺑﻘﺖ ﺩﺍﺭﺩ‪.‬‬

‫ﺗﺎﻛﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ n + 1‬ﻧﻘﻄﻪ ﺩﺭ ﺻﻔﺤﻪ‪ ،‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﻭ ﻧﻪ ﺩﻗﻴﻘﺎ ‪ n‬ﺍﺳﺖ‪.‬‬
‫ﺑﺴﺘﻪ ﺑﻪ ﻧﻘﺎﻁ‪ ،‬ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺣﺘﻲ ﺍﺯ ﺩﺭﺟﻪ ﺻﻔﺮ ﻭ ﻳﺎ ﻳﻚ ﻧﻴﺰ ﺑﺎﺷﺪ‪.‬‬

‫ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ‬ ‫‪۳.۷‬‬

‫ﺗﺎ ﺍﻳﻨﺠﺎ ﻣﻔﺎﻫﻴﻢ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺸﺮﻳﺢ ﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺍﺩﺍﻣﻪ ﺑﻪ ﻧﺤﻮﻩ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺧﻮﺍﻫﻴﻢ‬
‫ﭘﺮﺩﺍﺧﺖ‪ .‬ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺭﻭﺵﻫﺎﻱ ﻣﺘﻨﻮﻋﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺩﺭ ﺯﻳﺮ ﺑﻪ ﺑﺮﺧﻲ ﺍﺯ ﺁﻥﻫﺎ ﺍﺷﺎﺭﻩ ﻣﻲﮔﺮﺩﺩ‪:‬‬

‫• ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ‬
‫• ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ‬
‫• ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ‬
‫ﻧﻮﻳﻞ ‪۱۸‬‬ ‫• ﺭﻭﺵ‬
‫• ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﻣﺘﻨﺎﻫﻲ‬
‫ﺑﺮﻳﺴﻨﺘﺮﻳﻚ ‪۱۹‬‬ ‫• ﺭﻭﺵ‬

‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ ﺭﺍ ﺍﺭﺍﻳﻪ ﺧﻮﺍﻫﻴﻢ ﻧﻤﻮﺩ‪.‬‬
‫‪18 Neville‬‬ ‫‪19 Barycentric‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸۰‬‬

‫ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬ ‫�‬

‫ﺩﺭ ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ ‪ ،(xi , yi ), i = 0, . . . , n‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ‬

‫‪Pn (x) := a0 + a1 x + · · · + an xn ,‬‬ ‫)‪(۹.۷‬‬
‫ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺩﺭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻓﻮﻕ‪ ،‬ﺿﺮﺍﻳﺐ ‪ a0 , a1 , . . . , an‬ﻣﺠﻬﻮﻝ ﻣﻲﺑﺎﺷﺪ ﻭ ﺑﺎﻳﺪ ﺁﻥﻫﺎ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺑﻴﺎﺑﻴﻢ ﻛﻪ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ )‪ (۹.۷‬ﺍﺯ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﺑﮕﺬﺭﺩ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪Pn (x0 ) = y0‬‬ ‫‪‬‬
‫‪‬‬ ‫‪a0 + a1 x0 + · · · + an xn0 = y0‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪Pn (x1 ) = y1‬‬ ‫‪a0 + a1 x1 + · · · + an xn1 = y1‬‬
‫‪..‬‬ ‫⇒‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪Pn (xn ) = yn‬‬ ‫‪a0 + a1 xn + · · · + an xnn = yn‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ‪ ،‬ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﺑﺎ ‪ n + 1‬ﻣﻌﺎﺩﻟﻪ ﻭ ﻣﺠﻬﻮﻻﺕ ‪ a0 , a1 , . . . , an‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ‬
‫ﺧﻄﻲ ﺑﻪ ﺻﻮﺭﺕ ‪ Ax = b‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪x0‬‬ ‫‪x20‬‬ ‫···‬ ‫‪xn0‬‬ ‫‪a0‬‬ ‫‪y0‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪1‬‬ ‫‪x1‬‬ ‫‪x21‬‬ ‫···‬ ‫‪xn1 ‬‬ ‫‪ a1 ‬‬ ‫‪ y1 ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪A :=  .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫‪x :=  .  ,‬‬ ‫‪b :=  .  ,‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪ .. ‬‬ ‫‪ .. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪1‬‬ ‫‪xn‬‬ ‫‪x2n‬‬ ‫···‬ ‫‪xnn‬‬ ‫‪an‬‬ ‫‪yn‬‬
‫ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﺧﺎﺹ ﻣﻮﺳﻮﻡ ﺑﻪ ﻣﺎﺗﺮﻳﺲ ﻭﻧﺪﺭﻣﻮﻧﺪ ‪ ۲۰‬ﻣﻲ ﺑﺎﺷﺪ‪ .‬ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﮔﺮ ‪xi‬ﻫﺎ ﺍﺯ ﻳﻚﺩﻳﮕﺮ‬
‫ﻣﺠﺰﺍ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺎﺗﺮﻳﺲ ‪ A‬ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ‪ Ax = b‬ﺩﺭ ﻓﻮﻕ‪ ،‬ﺩﺍﺭﺍﻱ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ‬
‫ﻓﺮﺩ ﺍﺳﺖ ﻭ ﺍﻳﻦ ﺟﻮﺍﺏ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻧﻤﺎﻳﺪ‪.‬‬

‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺟﺎﻱ ﻓﺮﻡ )‪ (۹.۷‬ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺯ ﻓﺮﻡ ﺩﻳﮕﺮﻱ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ ﻛﻪ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻧﻬﺎﻳﻲ ﺭﺍ ﺳﺎﺩﻩﺗﺮ ﻛﻨﺪ؟‬

‫ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬ ‫�‬

‫ﺍﺳﺎﺱ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ ﺍﺳﺖ ﻛﻪ ﺩﺭ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫‪20 Vandermond‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۸۱‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﭼﻨﺪﺟﻤﻠەﺍﯼﻫﺎﯼ ﭘﺎﯾەﺍﯼ ﻻﮔﺮﺍﻧﮋ‬ ‫ﺗﻌﺮﯾﻒ ‪۴−۷‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ‪ x0 , . . . , xn ،‬ﻧﻘﺎﻃﻲ ﻣﺘﻤﺎﻳﺰ ﺑﺎﺷﻨﺪ‪ .‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫) ‪(x − x1 )(x − x2 ) . . . (x − xn‬‬
‫=‪ℓ0 (x) :‬‬ ‫‪,‬‬
‫) ‪(x0 − x1 )(x0 − x2 ) . . . (x0 − xn‬‬
‫) ‪(x − x0 )(x − x2 ) . . . (x − xn‬‬
‫=‪ℓ1 (x) :‬‬ ‫‪,‬‬
‫) ‪(x1 − x0 )(x1 − x2 ) . . . (x1 − xn‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬
‫) ‪(x − x0 ) . . . (x − xk−1 )(x − xk+1 ) . . . (x − xn‬‬
‫=‪ℓk (x) :‬‬ ‫‪,‬‬
‫) ‪(xk − x0 ) . . . (xk − xk−1 )(xk − xk+1 ) . . . (xk − xn‬‬
‫‪..‬‬ ‫‪..‬‬
‫‪.‬‬ ‫‪.‬‬
‫) ‪(x − x0 )(x − x1 ) . . . (x − xn−1‬‬
‫=‪ℓn (x) :‬‬ ‫‪.‬‬
‫) ‪(xn − x0 )(xn − x1 ) . . . (xn − xn−1‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ‪ n + 1‬ﻧﻘﻄﻪ ‪ x0 , . . . , xn‬ﺑﻪ ﺗﻌﺪﺍﺩ ‪ n + 1‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ‬

‫)‪ℓ0 (x), ℓ1 (x), · · · , ℓn (x‬‬
‫ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪ .‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻧﻤﺎﺩ ﺣﺎﺻﻞﺿﺮﺏ ‪ ،Π‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﻓﺸﺮﺩﻩﺗﺮ ﺯﻳﺮ ﻧﻴﺰ ﻧﺸﺎﻥ ﺩﺍﺩ‪:‬‬
‫‪Y‬‬
‫‪n‬‬
‫‪x − xj‬‬
‫=‪ℓi (x) :‬‬ ‫‪, i = 0, . . . , n.‬‬ ‫)‪(۱۰.۷‬‬
‫‪xi − xj‬‬
‫‪j=0,j̸=i‬‬
‫ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ‬
‫‪x0 = −1, x1 = −0.5, x2 = 0, x3 = 0.5, x4 = 1‬‬
‫ﺭﺳﻢ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸۲‬‬

‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ )‪ (۱۰.۷‬ﺧﻮﺍﺹ ﺟﺎﻟﺒﻲ ﺩﺍﺭﻧﺪ ﻛﻪ ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﻪ ﺳﻪ ﻣﻮﺭﺩ ﺁﻥ ﺍﺷﺎﺭﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪ -۱‬ﻳﻚ ﺧﺎﺻﻴﺖ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ )‪ (۱۰.۷‬ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺗﻮﺍﺑﻊ ﺑﺮﺍﻱ ﻣﺠﻤﻮﻋﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺣﺪﺍﻛﺜﺮ ﺍﺯ‬
‫ﺩﺭﺟﻪ ‪ n‬ﺗﺸﻜﻴﻞ ﻳﻚ ﭘﺎﻳﻪ ﻣﻲﺩﻫﻨﺪ‪ .‬ﻳﻌﻨﻲ ﻫﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺗﺮﻛﻴﺐ ﺧﻄﻲ‬
‫ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ )‪ (۱۰.۷‬ﻧﻮﺷﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺍﮔﺮ )‪ pn (x‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺑﺎﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺍﻋﺪﺍﺩ ﺣﻘﻴﻘﻲ ‪ a0 , . . . , an‬ﻣﻮﺟﻮﺩﻧﺪ ﻛﻪ‬
‫‪pn (x) = a0 ℓ0 (x) + a1 ℓ1 (x) + · · · + an ℓn (x).‬‬
‫ﺑﻪ ﺍﻳﻦ ﻋﻠﺖ ﻧﺎﻡ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ ﺑﻪ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪ -۲‬ﺁﻥ ﭼﻴﺰﻱ ﻛﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ ﺭﺍ ﻣﺘﻤﺎﻳﺰ ﻛﺮﺩﻩ ﺍﺳﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺍﻳﻦ ﺗﻮﺍﺑﻊ ﺩﺭ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪x0 , . . . , xn‬‬
‫ﻛﺎﺭﺩﻳﻨﺎﻝ ‪ ۲۱‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﻣﻘﺪﺍﺭ )‪ ℓi (x‬ﺩﺭ ﻧﻘﻄﻪ ‪ xi‬ﺑﺮﺍﺑﺮ ‪ 1‬ﻭ ﺩﺭ ﺑﻘﻴﻪ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﺮﺍﺑﺮ ‪ 0‬ﺍﺳﺖ‪ .‬ﺑﻪ‬
‫ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪‬‬
‫‪1,‬‬ ‫‪i = j,‬‬
‫=‪ℓi (xj ) = δij :‬‬ ‫)‪(۱۱.۷‬‬
‫‪0,‬‬ ‫‪i ̸= j.‬‬

‫‪ -۳‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ )‪ (۱۰.۷‬ﺩﺍﺭﺍﻱ ﺧﺎﺻﻴﺖ ﺍﻓﺮﺍﺯ ﻭﺍﺣﺪ ‪ ۲۲‬ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ‬
‫‪X‬‬
‫‪n‬‬
‫‪ℓi (x) = 1.‬‬
‫‪i=0‬‬

‫ﺑﻌﺪ ﺍﺯ ﻣﻌﺮﻓﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﺑﻪ ﻧﺤﻮﻩ ﺍﻧﺠﺎﻡ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺮﻣﻲﮔﺮﺩﻳﻢ‪ .‬ﻧﻘﺎﻁ ‪ (xi , yi ), i = 0, . . . , n‬ﻛﻪ‬
‫‪xi‬ﻫﺎ ﻣﺠﺰﺍﻳﻨﺪ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺍﮔﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ‪Pn‬‬
‫ﺑﻨﺎﻣﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻃﺒﻖ ﺧﺎﺻﻴﺖ ‪ ۱‬ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ‪ Pn‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺗﺮﻛﻴﺐ ﺧﻄﻲ ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ ﻧﻮﺷﺖ‪:‬‬
‫‪Pn (x) = a0 ℓ0 (x) + a1 ℓ1 (x) + · · · + an ℓn (x).‬‬ ‫)‪(۱۲.۷‬‬
‫ﺣﺎﻝ ﻣﺸﺎﺑﻪ ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ‪ a0 , . . . , an‬ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ ‪ .Pn (xi ) = yi‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪Pn (x0 ) = y0‬‬
‫‪‬‬
‫‪‬‬ ‫‪a0 ℓ0 (x0 ) + a1 ℓ1 (x0 ) + · · · + an ℓn (x0 ) = y0‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪Pn (x1 ) = y1‬‬ ‫‪‬‬
‫‪a0 ℓ0 (x1 ) + a1 ℓ1 (x1 ) + · · · + an ℓn (x1 ) = y1‬‬
‫‪..‬‬ ‫⇒‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪Pn (xn ) = yn‬‬ ‫‪a0 ℓ0 (xn ) + a1 ℓ1 (xn ) + · · · + an ℓn (xn ) = yn‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ ﺑﺎ ‪ n + 1‬ﻣﻌﺎﺩﻟﻪ ﻭ ﻣﺠﻬﻮﻻﺕ ‪ a0 , a1 , . . . , an‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ‬
‫ﺧﻄﻲ ﺑﻪ ﺻﻮﺭﺕ ‪ Ax = b‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫) ‪ℓ0 (x0‬‬ ‫) ‪ℓ1 (x0‬‬ ‫···‬ ‫) ‪ℓn (x0‬‬ ‫‪a0‬‬ ‫‪y0‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫) ‪ ℓ0 (x1‬‬ ‫) ‪ℓ1 (x1‬‬ ‫···‬ ‫‪ℓn (x1 ) ‬‬ ‫‪ a1 ‬‬ ‫‪ y1 ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫‪A :=  .‬‬ ‫‪..‬‬ ‫‪.. ‬‬ ‫‪,‬‬ ‫‪x :=  .  ,‬‬ ‫‪b :=  .  ,‬‬
‫‪ ..‬‬ ‫‪.‬‬ ‫‪. ‬‬ ‫‪ .. ‬‬ ‫‪ .. ‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪ ‬‬
‫) ‪ℓ0 (xn‬‬ ‫) ‪ℓ1 (xn‬‬ ‫···‬ ‫) ‪ℓn (xn‬‬ ‫‪an‬‬ ‫‪yn‬‬

‫‪21 Cardinal‬‬ ‫‪22 Partition‬‬ ‫‪of unity‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۸۳‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﻃﺒﻖ ﺧﺎﺻﻴﺖ ‪ ۲‬ﺍﺯ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ‪ A‬ﺑﺮﺍﺑﺮ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺍﺳﺖ‪ .‬ﻳﻌﻨﻲ ﺩﺭ‬
‫ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻛﻪ ﺩﺭ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻳﻚ ﻣﺎﺗﺮﻳﺲ ﻫﻤﺎﻧﻲ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺩﺭ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﻟﺰﻭﻣﻲ ﺑﻪ‬
‫ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺑﻪ ﺳﺎﺩﮔﻲ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ‪ x = b‬ﻭ ﻳﺎ ‪ .ai = yi , i = 0, . . . , n‬ﺗﻮﺿﻴﺤﺎﺕ ﻓﻮﻕ ﺩﺭ ﻗﻀﻴﻪ‬
‫ﺯﻳﺮ ﻣﻮﺳﻮﻡ ﺑﻪ ﻗﻀﻴﻪ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺟﻤﻊﺑﻨﺪﻱ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻗﻀﯿﻪ ﺩﺭﻭﻧﯿﺎﺑﯽ ﻻﮔﺮﺍﻧﮋ‬ ‫ﻗﻀﯿﻪ ‪۵−۷‬‬

‫ﺍﮔﺮ ‪ (xi , yi ), i = 0, . . . , n,‬ﻧﻘﺎﻃﻲ ﺑﺎ ‪xi‬ﻫﺎﻱ ﻣﺠﺰﺍ ﺑﺎﺷﻨﺪ ﻭ ﺍﮔﺮ )‪ Pn (x‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﻳﻦ ﻧﻘﺎﻁ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫)‪ Pn (x‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻗﺎﺑﻞ ﻧﻤﺎﻳﺶ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪X‬‬
‫‪n‬‬
‫= )‪Pn (x) = y0 ℓ0 (x) + y1 ℓ1 (x) + · · · + yn ℓn (x‬‬ ‫‪yi ℓi (x).‬‬ ‫)‪(۱۳.۷‬‬
‫‪i=0‬‬

‫ﺩﺭ ﻣﺜﺎﻝﻫﺎﻱ ﺯﻳﺮ ﺍﺯ ﻗﻀﻴﻪ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۲−۷‬ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﯼ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ(‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻘﺎﻁ‬
‫‪1‬‬ ‫‪1‬‬
‫)‪(x0 , y0 ) = ( , 2), (x1 , y1 ) = ( , −1), (x2 , y2 ) = (1, 7‬‬
‫‪3‬‬ ‫‪4‬‬
‫ﺭﺍ ﺑﺎ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﻣﺮﺗﺐ ﺷﺪﻩ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‪ ،‬ﺩﺭ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﻣﺮﺗﺐ‬
‫ﻧﺒﻮﺩﻥ ﺍﻳﻦ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺍﻫﻤﻴﺘﻲ ﻧﺪﺍﺭﺩ‪.‬‬
‫ﻃﺒﻖ ﻗﻀﻴﻪ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻧﻤﻮﺩ‬
‫‪P2 (x) := y0 ℓ0 (x) + y1 ℓ1 (x) + y2 ℓ2 (x),‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫) ‪(x − x1 )(x − x2‬‬ ‫‪(x −‬‬ ‫)‪4 )(x − 1‬‬
‫‪1‬‬
‫=‪ℓ0 (x) :‬‬ ‫‪= 1‬‬ ‫)‪= −18(x − 41 )(x − 1‬‬
‫) ‪(x0 − x1 )(x0 − x2‬‬ ‫‪(3 −‬‬ ‫)‪4 )( 3 − 1‬‬
‫‪1 1‬‬

‫) ‪(x − x0 )(x − x2‬‬ ‫‪(x −‬‬ ‫)‪3 )(x − 1‬‬

‫‪1‬‬
‫=‪ℓ1 (x) :‬‬ ‫‪= 1‬‬ ‫)‪= 16(x − 31 )(x − 1‬‬
‫) ‪(x1 − x0 )(x1 − x2‬‬ ‫‪(4 −‬‬ ‫)‪3 )( 4 − 1‬‬
‫‪1 1‬‬

‫) ‪(x − x0 )(x − x1‬‬ ‫‪(x −‬‬ ‫) ‪3 )(x − 4‬‬

‫‪1‬‬ ‫‪1‬‬
‫=‪ℓ2 (x) :‬‬ ‫=‬ ‫‪= 2(x − 13 )(x − 41 ).‬‬
‫) ‪(x2 − x0 )(x2 − x1‬‬ ‫‪(1 −‬‬ ‫) ‪3 )(1 − 4‬‬
‫‪1‬‬ ‫‪1‬‬

‫ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﺩﺍﺭﻳﻢ‪:‬‬

‫‬ ‫‬ ‫‬ ‫‬ ‫‬ ‫‬
‫‪P2 (x) = 2 −18(x − 14 )(x − 1) + (−1) 16(x − 13 )(x − 1) + 7 2(x − 31 )(x − 41 ) .‬‬

‫ﻣﺜﺎﻝ ‪)۳−۷‬ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﯼ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ ﺗﺎﺑﻊ ﺳﯿﻨﻮﺱ(‬

‫= ‪x0 = 0, x1‬‬ ‫‪π‬‬

‫‪6,‬‬ ‫= ‪x2‬‬ ‫‪π‬‬
‫‪3,‬‬ ‫= ‪x3‬‬ ‫‪π‬‬
‫‪2‬‬ ‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ )‪ f (x) = sin(x‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ‬
‫ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﮔﺮﺩﺩ‪ .‬ﻃﺒﻖ ﻗﻀﻴﻪ ﻻﮔﺮﺍﻧﮋ‪ ،‬ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪P3 (x) := sin(0)ℓ0 (x) + sin( π6 )ℓ1 (x) + sin( π3 )ℓ2 (x) + sin( π2 )ℓ3 (x),‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸۴‬‬

‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬

‫
‬ ‫
‬ ‫
‬
‫‪x−‬‬ ‫‪π‬‬
‫
‪6‬‬ ‫‪x − π3‬‬ ‫‪x − π2‬‬ ‫‪36‬‬ ‫
‬ ‫
‬ ‫
‬
‫=‪ℓ0 (x) :‬‬ ‫
‬ ‫‪
= − 3 x − π6 x − π3 x − π2 ,‬‬
‫‪0−‬‬ ‫‪π‬‬
‫‪0 − π3‬‬ ‫‪0− 2‬‬‫‪π‬‬ ‫‪π‬‬
‫‪6‬‬
‫
‬ ‫
‬
‫)‪(x − 0‬‬ ‫‪x − π3‬‬ ‫‪x− 2‬‬‫‪π‬‬
‫‪108‬‬ ‫
‬ ‫
‬
‫=‪ℓ1 (x) :‬‬ ‫
‬ ‫
‬ ‫
‬
‫‪π = π 3 (x − 0) x − 3‬‬
‫‪π‬‬
‫‪x − π2 ,‬‬
‫‪6 −0‬‬ ‫‪6 − 3‬‬ ‫‪6 − 2‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬
‫
‬ ‫
‬
‫)‪(x − 0‬‬ ‫‪x − π6‬‬ ‫‪x − π2‬‬ ‫‪108‬‬ ‫
‬ ‫
‬
‫=‪ℓ2 (x) :‬‬ ‫
‬ ‫
‬ ‫
‬
‫‪π = − π 3 (x − 0) x − 6‬‬
‫‪π‬‬
‫‪x − π2 ,‬‬
‫‪3 −0‬‬ ‫‪3 − 6‬‬ ‫‪3 − 2‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬
‫
‬ ‫
‬
‫)‪(x − 0‬‬ ‫‪x − π6‬‬ ‫‪x − π3‬‬ ‫‪36‬‬ ‫
‬ ‫
‬
‫=‪ℓ3 (x) :‬‬ ‫
‬ ‫
‬ ‫
‬
‫‪π = π 3 (x − 0) x − 6‬‬
‫‪π‬‬
‫‪x − π3 .‬‬
‫‪2 −0‬‬ ‫‪2 − 6‬‬ ‫‪2 − 3‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬

‫ﺑﻌﺪ ﺍﺯ ﺟﺎﻳﮕﺬﺍﺭﻱ ﻭ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺟﺒﺮﻱ‪ ،‬ﺩﺭﻧﻬﺎﻳﺖ ﺩﺍﺭﻳﻢ‪:‬‬

‫! √‬ ‫! √‬ ‫! √‬
‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬
‫= )‪P3 (x‬‬ ‫‪−3‬‬ ‫‪−2‬‬ ‫‪−1‬‬
‫‪90 π − 54 3 x + −63 π + 36 2 x + 11 π − 9/2‬‬
‫‪3‬‬ ‫‪2‬‬
‫)‪x. (۱۴.۷‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ P3‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ sin‬ﺩﺭ ﺑﺎﺯﻩ ] ‪ [0, π2‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺑﺮﺍﻱ ﺁﻥ ﻛﻪ ﺩﻗﺖ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺭﺍ‬
‫ﺑﺴﻨﺠﻴﻢ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ )ﺳﻤﺖ ﭼﭗ( ﺗﺎﺑﻊ ‪ sin‬ﻭ ﺩﺭﻭﻧﻴﺎﺑﺶ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﮔﺮﺩﺩ ﻛﻪ ﺩﺭ ﺑﺎﺯﻩ ] ‪ ،[0, π2‬ﻧﻤﻮﺩﺍﺭ ﺍﻳﻦ‬
‫ﺩﻭ ﺗﺎﺑﻊ ﺑﻪ ﻫﻤﺪﻳﮕﺮ ﻧﺰﺩﻳﻚ ﺍﺳﺖ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﺑﻪ ﺳﺨﺘﻲ ﻣﻲﺗﻮﺍﻥ ﺁﻥﻫﺎ ﺭﺍ ﺍﺯ ﻫﻢ ﺗﻔﻜﻴﻚ ﻧﻤﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺑﻬﺘﺮ ﻣﺸﺨﺺ ﻛﺮﺩﻥ‬
‫ﺩﻗﺖ ﺗﻘﺮﻳﺐ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ )ﺳﻤﺖ ﺭﺍﺳﺖ( ﻧﻤﻮﺩﺍﺭ ﺧﻄﺎﻱ )‪ f (x) − P3 (x‬ﻧﻴﺰ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪2‬‬ ‫)‪sin(x) − p3 (x‬‬

‫)‪sin(x‬‬
‫)‪p3 (x‬‬

‫‪1‬‬ ‫‪+0.01‬‬
‫‪x‬‬
‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬
‫‪−0.01‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪2‬‬
‫‪x‬‬
‫‪1‬‬ ‫‪2‬‬

‫■ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ ﻭ ﺍﻟﮕﻮﺭﻳﺘﻢ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ‬

‫ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﺷﺪ‪ .‬ﺍﻛﻨﻮﻥ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﺭ ﻣﻮﺭﺩ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ ﺍﻳﻦ ﺭﻭﺵ ﺻﺤﺒﺖ ﻛﻨﻴﻢ‪ .‬ﺍﻧﺠﺎﻡ ﺍﻳﻦ‬
‫ﺭﻭﺵ ﺭﻭﻱ ﻛﺎﻏﺬ ﺳﺎﺩﻩ ﻭ ﺳﺮﺭﺍﺳﺖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﭘﻴﺎﺩﻩﺳﺎﺯﻱ ﺭﻭﺵ ﻣﺘﻔﺎﻭﺕ ﺑﺎ ﺍﻧﺠﺎﻡ ﺁﻥ ﺭﻭﻱ ﻛﺎﻏﺬ ﺍﺳﺖ‪ .‬ﺩﺭ ﻣﺜﺎﻝﻫﺎﻱ ﻗﺒﻞ‪،‬‬
‫ﺿﺎﺑﻄﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺳﺘﺨﺮﺍﺝ ﺷﺪ‪ .‬ﺍﻣﺎ ﺍﺳﺘﺨﺮﺍﺝ ﺿﺎﺑﻄﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﭼﻨﺪﺍﻥ ﺑﻪ ﻛﺎﺭ ﻧﻤﻲﺁﻳﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺫﺧﻴﺮﻩ‬
‫ﻭ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﺿﺎﺑﻄﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻧﻴﺎﺯ ﺑﻪ ﺑﺮﻧﺎﻣﻪﻧﻮﻳﺴﻲ ﭘﻴﭽﻴﺪﻩ ﺩﺍﺭﺩ‪.‬‬
‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ‪ n + 1‬ﺯﻭﺝ ﻧﻘﻄﻪ ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻭ ﻳﻚ ﻣﻘﺪﺍﺭ ﻣﺎﻧﻨﺪ ‪ a‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻭﺭﻭﺩﻱ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ‬
‫ﻣﻴﺨﻮﺍﻫﻴﻢ ﻣﻘﺪﺍﺭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﻓﻮﻕ ﺭﺍ ﻧﻘﻄﻪ ‪ a‬ﺑﻴﺎﺑﻴﻢ‪ .‬ﻻﺯﻡ ﻧﻴﺴﺖ ﻛﻪ ﺿﺎﺑﻄﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ‬
‫ﻣﺸﺨﺺ ﻛﻨﻴﻢ ﻭ ﺳﭙﺲ ﻣﻘﺪﺍﺭ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺭﺍ ﺩﺭ ﻧﻘﻄﻪ ‪ a‬ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﻳﻦ ﻛﺎﺭ ﻣﺴﺘﻠﺰﻡ ﻣﺤﺎﺳﺒﺎﺕ ﺳﻨﮕﻴﻦ ﻭ ﻧﻤﺎﺩﻳﻦ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۸۵‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫‪Pn‬‬
‫‪i=0‬‬ ‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻋﻮﺽ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﺍﻳﻦ ﻛﺎﺭ ﺍﺑﺘﺪﺍ ‪ ℓi (a), i = 0, · · · , n‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻣﻴﻜﻨﻴﻢ ﻭ ﺳﭙﺲ ﻣﻘﺪﺍﺭ )‪yi ℓi (a‬‬
‫ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﺧﺮﻭﺟﻲ ﻣﻌﺮﻓﻲ ﻣﻴﻜﻨﻴﻢ‪.‬‬
‫ﻭﺭﻭﺩﻱ‪ (xi , yi ), i = 0, · · · , n :‬ﻭ ‪a‬‬
‫ﺧﺮﻭﺟﻲ‪Pn (a) :‬‬

‫ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻤﺪﺟﻤﻠﻪﺍﻱ‬ ‫‪۴.۷‬‬

‫ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ ﻛﻪ ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ ﺑﻪ ﺭﻭﺵ ﻧﻴﻮﺗﻦ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ ،‬ﻳﻚ ﺭﻭﺵ ﺩﻳﮕﺮ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺩﺭﻭﻧﻴﺎﺏ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﻫﻤﺎﻧﻨﺪ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﻳﻚ ﻧﻤﺎﻳﺶ ﺑﺎ ﺿﺮﺍﻳﺐ ﻣﺠﻬﻮﻝ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﺩﺭﻭﻧﻴﺎﺏ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‬
‫ﻭ ﺳﭙﺲ ﺿﺮﺍﻳﺐ ﻣﺠﻬﻮﻝ ﺍﺯ ﺣﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ‪.‬‬
‫ﺩﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﭘﺎﻳﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ‬
‫) ‪1, (x − x0 ), (x − x0 )(x − x1 ), (x − x0 )(x − x1 )(x − x2 ), . . . , (x − x0 )(x − x1 ) . . . (x − xn−1‬‬
‫ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﻧﺪ ﻭ ﻟﺬﺍ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﻣﺘﻤﺎﻳﺰ ‪ x0 , x1 , . . . , xn‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ‬
‫ﻣﻲﺷﻮﺩ‬
‫) ‪Pn (x) := a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 ) + · · · + an (x − x0 ) . . . (x − xn−1‬‬
‫ﻭ ‪ai‬ﻫﺎ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﺸﺨﺺ ﻣﻲﺷﻮﻧﺪ ﻛﻪ ﺩﺭ ﻧﻘﺎﻁ ﻓﻮﻕ ﺑﺎ ﺗﺎﺑﻊ ‪ f‬ﺗﻄﺎﺑﻖ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻨﺪ‪ .‬ﻳﻌﻨﻲ‬
‫‪Pn (xi ) = f (xi ), i = 0, . . . , n.‬‬
‫ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫) ‪Pn (x0 ) = f (x0‬‬ ‫‪‬‬
‫‪‬‬ ‫) ‪a0 = f (x0‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫) ‪Pn (x1 ) = f (x1‬‬ ‫‪‬‬
‫‪‬‬ ‫) ‪a0 + a1 (x1 − x0 ) = f (x1‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫) ‪Pn (x2 ) = f (x2‬‬ ‫⇒‬ ‫‪a0 + a1 (x2 − x0 ) + a2 (x2 − x0 )(x2 − x1 ) = y2‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫) ‪P (x ) = f (x‬‬ ‫‪a + a ℓ (x − x ) + · · · + a (x − x ) . . . (x − x‬‬
‫‪n‬‬ ‫‪n‬‬ ‫‪n‬‬ ‫‪0‬‬ ‫‪1 1 n‬‬ ‫‪0‬‬ ‫‪n n‬‬ ‫‪0‬‬ ‫‪n‬‬ ‫) ‪n−1 ) = f (xn‬‬

‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺩﺳﺘﮕﺎﻩ ‪ n + 1‬ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﺑﺮﺣﺴﺐ ﻣﺠﻬﻮﻻﺕ ‪ a0 , a1 , . . . , an‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ‬
‫ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺑﻪ ﺻﻮﺭﺕ ‪ Ax = b‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪...‬‬ ‫‪0‬‬
‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫) ‪(x1 − x0‬‬ ‫‪0‬‬ ‫‪...‬‬ ‫‪...‬‬ ‫‪0‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫) ‪(x2 − x0‬‬ ‫) ‪(x2 − x0 )(x2 − x1‬‬ ‫‪‬‬
‫‪A := 1‬‬ ‫‪...‬‬ ‫‪...‬‬ ‫‪0‬‬ ‫‪,‬‬
‫‪.‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪‬‬
‫‪.‬‬ ‫‪‬‬
‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪‬‬
‫) ‪1 (xn − x0 ) (xn − x0 )(xn − x1‬‬ ‫‪...‬‬ ‫‪...‬‬ ‫) ‪(xn − x0 ) . . . (xn − xn−1‬‬
‫ﻭ‬
‫‪h‬‬ ‫⊤‪i‬‬ ‫‪h‬‬ ‫⊤‪i‬‬
‫‪x := a0‬‬ ‫‪a1‬‬ ‫‪a2‬‬ ‫‪...‬‬ ‫‪an‬‬ ‫‪,‬‬ ‫‪b := y0‬‬ ‫‪y1‬‬ ‫‪y2‬‬ ‫‪...‬‬ ‫‪yn‬‬ ‫‪.‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻇﺎﻫﺮ ﺷﺪﻩ ﺩﺭ ﺭﻭﺵ ﻧﻴﻮﺗﻦ‪ ،‬ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﭘﺎﻳﻴﻦﻣﺜﻠﺜﻲ ﺍﺳﺖ ﻛﻪ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺳﺎﺩﻩ ﺍﺳﺖ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸۶‬‬

‫ﻭ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﻴﺶﺭﻭ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﭘﻴﺶﺭﻭ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺍﻭﻝ ﻣﻘﺪﺍﺭ ‪ a0‬ﻭ ﺍﺯ ﻣﻌﺎﺩﻟﻪ ﺩﻭﻡ ﻣﻘﺪﺍﺭ ‪ a1‬ﻭ ﺍﺯ ﻣﻌﺎﺩﻟﻪ‬
‫ﺳﻮﻡ ‪ a2‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ‪:‬‬
‫‪a0 = f (x0 ),‬‬
‫) ‪f (x1 ) − f (x0‬‬
‫= ‪a1‬‬ ‫‪,‬‬
‫‪x1 − x0‬‬
‫) ‪f (x1 )−f (x0‬‬
‫‪f (x2 ) − f (x0 ) −‬‬ ‫‪x1 −x0‬‬ ‫‪(x2‬‬ ‫) ‪− x0‬‬
‫= ‪a2‬‬
‫) ‪(x2 − x0 )(x2 − x1‬‬
‫ﺍﮔﺮ ﺑﻪ ﻫﻤﻴﻦ ﺭﻭﺍﻝ ﺍﺩﺍﻣﻪ ﺩﻫﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺗﺮﺗﻴﺐ ﻣﻘﺎﺩﻳﺮ ‪ a3‬ﺗﺎ ‪ an‬ﻧﻴﺰ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺍﻣﺎ ﻫﺮﭼﻪ ﺑﻴﺸﺘﺮ ﭘﻴﺶ ﻣﻲﺭﻭﻳﻢ‪،‬‬
‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﻣﺤﺎﺳﺒﻪ ‪ a3‬ﺗﺎ ‪ an‬ﭘﻴﭽﻴﺪﻩ ﻭ ﭘﻴﭽﻴﺪﻩﺗﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺧﻮﺷﺒﺨﺘﺎﻧﻪ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺍﺑﺰﺍﺭ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ‪ ۲۳‬ﻣﻲﺗﻮﺍﻥ ﺭﻭﺍﻝ‬
‫ﻣﻨﺎﺳﺐ ﻭ ﺳﻴﺴﺘﻤﺎﺗﻴﻜﻲ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ai‬ﻫﺎ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﺍﺑﺰﺍﺭ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﯿﻢ ﺷﺪﻩ‬ ‫ﺗﻌﺮﯾﻒ ‪۶−۷‬‬
‫ﺍﮔﺮ ‪ f‬ﺗﺎﺑﻌﻲ ﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ﻧﻘﺎﻁ ‪ x0 , · · · , xn‬ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ‬

‫• ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﺮﺗﺒﻪ ﺻﻔﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﻄﻪ ‪ xi‬ﺑﺎ ] ‪ f [xi‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫) ‪f [xi ] = f (xi‬‬

‫• ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﺮﺗﺒﻪ ﻳﻚ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻧﻘﺎﻁ ‪ xi‬ﻭ ‪ xj‬ﺑﺎ ] ‪ f [xi , xj‬ﻧﺸﺎﻥ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ‬
‫ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫] ‪f [xi ] − f [xj‬‬
‫= ] ‪f [xi , xj‬‬
‫‪xi − xj‬‬
‫• ﺑﺎﺯﺍﻱ ‪ k > 0‬ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﺮﺗﺒﻪ ‪ k‬ﺩﺭ ﻧﻘﺎﻁ ‪xi0 , xi1 , · · · , xik‬ﻛﻪ ﺑﺎ ] ‪ f [xi0 , · · · , xik‬ﻧﺸﺎﻥ ﺩﺍﺩﻩ‬
‫ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫] ‪f [xi0 , · · · , xik−1 ] − f [xi1 , · · · , xik‬‬
‫= ] ‪f [xi0 , · · · , xik‬‬
‫‪x i0 − x ik‬‬

‫ﻣﻄﺎﺑﻖ ﺑﺎ ﺗﻌﺮﻳﻒ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﺩﺍﺭﺍﻱ ﻣﺎﻫﻴﺖ ﺑﺎﺯﮔﺸﺘﻲ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﺮﺗﺒﻪ‬
‫‪ k‬ﺑﺮﺣﺴﺐ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﺮﺗﺒﻪ ‪ k − 1‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۴−۷‬ﻣﺤﺎﺳﺒﻪ ﺗﻔﺎﺿﻞ ﺗﻘﺴﯿﻢ ﺷﺪﻩ(‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ‪ x0 = 0, x1 = 0.5, x2 = 1‬ﻭ ‪ .f (x) = 4x2 + 1‬ﻣﻲﺧﻮﺍﻫﻴﻢ ] ‪ f [x0 , x1 , x2‬ﺭﺍ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ‪ .‬ﺑﺎ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻌﺮﻳﻒ ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪:‬‬
‫] ‪f [x0 ]−f [x1‬‬ ‫] ‪f [x1 ]−f [x2‬‬
‫] ‪f [x0 , x1 ] − f [x1 , x2‬‬ ‫‪x0 −x1‬‬ ‫‪−‬‬ ‫‪x1 −x2‬‬
‫= ] ‪f [x0 , x1 , xn‬‬ ‫=‬
‫‪x0 − x2‬‬ ‫‪x0 − x2‬‬
‫‪1−2‬‬
‫‪− 2−5‬‬
‫‪= 0−0.5 0.5−1 = 4.‬‬
‫‪0−1‬‬

‫‪23 Divided‬‬ ‫‪difference‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۸۷‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﺍﺳﺘﻘﻼﻝ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﯿﻢ ﺷﺪﻩ ﺍﺯ ﺗﺮﺗﯿﺐ ﻧﻘﺎﻁ‬ ‫ﻗﻀﯿﻪ ‪۷−۷‬‬

‫ﺗﺮﺗﻴﺐ ﻧﻘﺎﻁ ﺩﺭ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﺩﺭ ﻣﻘﺪﺍﺭ ﺁﻥ ﺍﻫﻤﻴﺘﻲ ﻧﺪﺍﺭﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺍﮔﺮ } ‪{x0 , · · · , xk } = {y0 , · · · , yk‬‬
‫ﺁﻥﮔﺎﻩ ] ‪f [x0 , · · · , xk ] = f [y0 , · · · , yk‬‬

‫ﻣﺜﺎﻝ ‪۵−۷‬‬
‫ﺍﮔﺮ ‪x0 , x1‬ﻭ ‪ x2‬ﺳﻪ ﻧﻘﻄﻪ ﻣﺘﻤﺎﻳﺰ ﺑﺎﺷﻨﺪ ﺁﻥﮔﺎﻩ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺍﻳﻦ ﺳﻪ ﻧﻘﻄﻪ ﺑﻪ ﺻﻮﺭﺗﻬﺎﻱ ﺯﻳﺮ ﻗﺎﺑﻞ ﺗﻌﺮﻳﻒ‬
‫ﻣﻲﺑﺎﺷﺪ‬
‫] ‪f [x0 ,x1 ]−f [x1 ,x2‬‬
‫= ] ‪f [x0 , x1 , x2‬‬ ‫‪x0 −x2‬‬
‫] ‪f [x1 ,x0 ]−f [x0 ,x2‬‬
‫= ] ‪f [x1 , x0 , x2‬‬ ‫‪x1 −x2‬‬
‫] ‪f [x2 ,x0 ]−f [x0 ,x1‬‬
‫= ] ‪f [x2 , x0 , x1‬‬ ‫‪x2 −x1‬‬

‫ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ ﺍﮔﺮ ‪ A‬ﻟﻴﺴﺘﻲ ﺍﺯ ‪ m‬ﻋﻀﻮ ﺍﺯ ﻣﺠﻤﻮﻋﻪ } ‪ {x0 , · · · , xn‬ﺑﺎﺷﺪ ﻛﻪ ﺷﺎﻣﻞ ‪ xα‬ﻧﺒﺎﺷﺪ ﻭ ‪ B‬ﻧﻴﺰ ﻟﻴﺴﺖ ﻣﺘﻤﺎﻳﺰﻱ‬
‫ﺩﻳﮕﺮﻱ ﺑﺎﺷﺪ ﻛﻪ ﺷﺎﻣﻞ ‪ xβ‬ﻧﺒﺎﺷﺪ ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‬
‫]‪f [A] − f [B‬‬
‫= ] ‪f [x0 , · · · , xn‬‬
‫‪xβ − xα‬‬
‫ﺍﻛﻨﻮﻥ ﻛﻪ ﺍﺑﺰﺍﺭ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﺗﻌﺮﻳﻒ ﺷﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻗﻀﻴﻪ ﻧﻴﻮﺗﻦ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺭﺍ ﺍﺭﺍﻳﻪ ﻛﺮﺩ‪.‬‬

‫ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﯿﻢ ﺷﺪﻩ )ﺭﻭﺵ ﻧﯿﻮﺗﻦ( ﺑﺮﺍﯼ ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻗﻀﯿﻪ ‪۸−۷‬‬
‫ﺍﮔﺮ‬
‫) ‪Pn (x) := a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 ) + · · · + an (x − x0 ) . . . (x − xn−1‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﻣﺘﻤﺎﻳﺰ ‪ x0 , x1 , . . . , xn‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬
‫‪ai = f [x0 , . . . , xi ].‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪Pn (x) = f [x0 ]+(x − x0 )f [x0 , x1 ] + (x − x0 )(x − x1 )f [x0 , x1 , x2 ] + · · · +‬‬
‫] ‪(x − x0 ) · · · (x − xn−1 )f [x0 , · · · , xn‬‬

‫ﻓﺮﻡ ﻓﺸﺮﺩﻩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺩﺭ ﻗﻀﻴﻪ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻮﺷﺖ‬
‫‪X‬‬
‫‪n‬‬
‫= )‪Pn (x‬‬ ‫] ‪(x − x0 ) · · · (x − xi−1 )f [x0 , · · · , xi‬‬
‫‪i=0‬‬

‫ﺩﺭ ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﻧﻴﺎﺯ ﺑﻪ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ] ‪ f [x0 ], f [x0 , x1 ], · · · , f [x0 , · · · , xn‬ﺩﺍﺭﻳﻢ‪ .‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ‬
‫ﺳﻴﺴﺘﻤﺎﻧﻴﻚ ﻭ ﻛﺎﺭﺍﻱ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﻮﺭﺩ ﻧﻴﺎﺯ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻳﻚ ﺟﺪﻭﻝ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺗﻮﺿﻴﺢ ﺍﻳﻦ ﻧﻜﺘﻪ ﻓﺮﺽ ﻛﻨﻴﺪ‬
‫ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , x1 , x2 , x3‬ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ .‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﻮﺭﺩ ﻧﻴﺎﺯ‪،‬‬
‫ﺟﺪﻭﻟﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۸۸‬‬

‫) ‪x0 f [x0 ]=f (x0‬‬

‫‪f [x0 , x1 ]= f [xx00]−f‬‬
‫‪−x1‬‬
‫] ‪[x1‬‬

‫) ‪x1 f [x1 ]=f (x1‬‬ ‫‪f [x0 , x1 , x2 ]= f [x0 ,xx10]−f‬‬

‫‪−x2‬‬
‫] ‪[x1 ,x2‬‬

‫‪f [x1 , x2 ]= f [xx11]−f‬‬

‫‪−x2‬‬
‫] ‪[x2‬‬
‫‪f [x0 , x1 , x2 , x3 ]= f [x0 ,x1 ,xx20]−f‬‬
‫‪−x3‬‬
‫] ‪[x1 ,x2 ,x3‬‬

‫) ‪x2 f [x2 ]=f (x2‬‬ ‫‪f [x1 , x2 , x3 ]= f [x1 ,xx21]−f‬‬

‫‪−x3‬‬
‫] ‪[x2 ,x3‬‬

‫‪f [x2 , x3 ]= f [xx22]−f‬‬

‫‪−x3‬‬
‫] ‪[x3‬‬

‫) ‪x3 f [x3 ]=f (x3‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻣﻘﺪﺍﺭ ﻫﺮ ﺧﺎﻧﻪ ﺍﺯ ﺟﺪﻭﻝ‪ ،‬ﺑﺎ ﻛﻤﻚ ﻣﻘﺪﺍﺭ ﺩﻭ ﺧﺎﻧﻪ ﺩﺭ ﺳﺘﻮﻥ ﻣﺎﻗﺒﻞ ﺁﻥ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ ﻭ ﺑﻌﺪ ﺍﺯ ﺗﻜﻤﻴﻞ‬
‫ﺟﺪﻭﻝ‪ ،‬ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻣﻮﺭﺩ ﻧﻴﺎﺯ ﺩﺭ ﻳﺎﻝ ﺑﺎﻻﻳﻲ ﺟﺪﻭﻝ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۶−۷‬ﺍﺳﺘﺨﺮﺍﺝ ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ ﺑﺎ ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﯿﻢ ﺷﺪﻩ(‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ‬
‫‪3‬‬
‫‪x0 = 1, x1 = , x2 = 0, x3 = 2‬‬
‫‪2‬‬
‫ﺭﺍ ﺑﻴﺎﺑﻴﻢ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫‪13‬‬ ‫‪5‬‬
‫= ) ‪f (x0 ) = 3, f (x1‬‬ ‫= ) ‪, f (x2 ) = 3, f (x3‬‬
‫‪4‬‬ ‫‪3‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪P3 (x) = f [x0 ] + f [x0 , x1 ](x − x0 ) + f [x0 , x1 , x2 ](x − x0 )(x − x1 )+‬‬
‫) ‪f [x0 , x1 , x2 , x3 ](x − x0 )(x − x1 )(x − x2‬‬
‫ﺣﺎﻝ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻻﺯﻡ ﺟﺪﻭﻝ ﺯﻳﺮ ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪1‬‬ ‫‪f [x0 ]= 3‬‬ ‫‪13‬‬
‫‪3− 4‬‬ ‫‪1‬‬
‫=] ‪f [x0 , x1‬‬ ‫‪3‬‬ ‫=‬ ‫‪2‬‬
‫‪1− 2‬‬ ‫‪1 1‬‬
‫‪2−6‬‬
‫‪3‬‬ ‫‪13‬‬
‫‪2‬‬
‫=] ‪f [x1‬‬ ‫‪4‬‬ ‫‪13‬‬ ‫= ] ‪f [x0 , x1 , x2‬‬ ‫=‬ ‫‪1‬‬
‫‪−3‬‬ ‫‪1−0‬‬ ‫‪3‬‬ ‫‪1 −5‬‬
‫‪3− 3‬‬
‫‪1‬‬
‫‪f [x1 , x2 ]= 43‬‬ ‫=‬ ‫‪6‬‬ ‫‪1 −2‬‬ ‫= ] ‪f [x0 , x1 , x2 , x3‬‬ ‫‪= −2‬‬
‫‪2 −0‬‬
‫‪1−2‬‬
‫‪6− 3‬‬ ‫‪−5‬‬
‫‪0‬‬ ‫‪f [x2 ]= 3‬‬ ‫=] ‪f [x0 , x1 , x2‬‬ ‫‪3‬‬ ‫=‬ ‫‪3‬‬
‫‪3− 5‬‬ ‫‪−2‬‬ ‫‪2 −2‬‬
‫= ‪f [x1 , x2 ]= 0−23‬‬ ‫‪3‬‬
‫‪5‬‬
‫‪2‬‬ ‫=] ‪f [x3‬‬ ‫‪3‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﺑﻌﺪ ﺍﺯ ﺟﺎﻳﮕﺬﺍﺭﻱ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺩﺭ ﻓﺮﻡ ﻧﻴﻮﺗﻦ‪ ،‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫‪1‬‬ ‫‪1‬‬
‫‪P3 (x) = 3 +‬‬ ‫‪(x − 1) + (x − 1)(x −‬‬ ‫‪3‬‬
‫)‪2‬‬ ‫‪+ (−2)(x − 1)(x −‬‬ ‫‪3‬‬
‫‪2 )(x‬‬ ‫)‪− 0‬‬
‫‪2‬‬ ‫‪3‬‬

‫■ ﻧﻜﺎﻧﻲ ﺩﺭ ﺧﺼﻮﺹ ﺭﻭﺵ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﻧﻴﻮﺗﻦ‬

‫‪ .۱‬ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ ﻣﻘﺪﺍﺭﻱ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﺩﺭ ﺁﻥ ﻣﺤﺎﺳﺒﻪ ﻛﻨﻴﻢ ﺑﺴﺘﮕﻲ ﺗﺎﻡ ﻧﺪﺍﺭﺩ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺍﺑﺘﺪﺍ ﺟﺪﻭﻝ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﮔﺮﺩﺩ ﻭ ﺳﭙﺲ ﺑﻪﺍﺯﺍﻱ ﻣﻘﺎﺩﻳﺮ ﻣﺘﻔﺎﻭﺕ ﺍﺯ ﺍﻳﻦ ﺟﺪﻭﻝ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪ .۲‬ﺍﻳﻦ ﺭﻭﺵ ﺑﻪ ﺗﺮﺗﻴﺐ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺴﺘﮕﻲ ﻧﺪﺍﺭﺩ‬

‫‪ .۳‬ﻓﻀﺎﻱ ﻻﺯﻡ ﺑﺮﺍﻱ ﭘﻴﺎﺩﻫﺴﺎﺯﻱ ﺭﻭﺵ )‪ O(n‬ﻣﻲﺑﺎﺷﺪ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۸۹‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫‪ .۴‬ﺩﺭ ﺻﻮﺭﺕ ﺍﺿﺎﻓﻪ ﺷﺪﻥ ﺗﻌﺪﺍﺩﻱ ﻧﻘﻄﻪ ﺟﺪﻳﺪ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺳﺎﺩﮔﻲ ﺍﺯ ﻣﺤﺎﺳﺒﺎﺕ ﻗﺒﻠﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬

‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ ﺗﺎﺑﻊ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺁﻥ‬ ‫‪۵.۷‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺍﮔﺮ )‪ Pn (x‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ‪ n + 1‬ﻧﻘﻄﻪ ﻣﺠﺰﺍ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ )‪Pn (x‬‬
‫ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﺗﻘﺮﻳﺐ ﺍﺯ ﺗﺎﺑﻊ )‪ f (x‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺖ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‪:‬‬
‫‪f (x) − Pn (x).‬‬
‫ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺑﻪ ﺗﺤﻠﻴﻞ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ ﻭ ﺑﻪ ﺳﻮﺍﻻﺕ ﺯﻳﺮ ﭘﺎﺳﺦ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬

‫• ﻛﺮﺍﻥ ﺑﺎﻻﻱ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﭼﻪ ﻣﻲﺑﺎﺷﺪ؟‬

‫• ﺩﺭ ﭼﻪ ﺑﺎﺯﻩﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﻘﺮﻳﺐ ﺧﻮﺑﻲ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﻣﺤﺴﻮﺏ ﻣﻲﺷﻮﺩ؟‬

‫• ﺑﺮﺍﻱ ﺣﺼﻮﻝ ﺑﻪ ﻳﻚ ﺩﻗﺖ ﺧﺎﺹ‪ ،‬ﺍﺯ ﭼﻪ ﺗﻌﺪﺍﺩ ﻧﻘﻄﻪ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ؟‬

‫• ﺁﻳﺎ ﺑﺎ ﺍﻓﺰﺍﻳﺶ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ‪ ،‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻛﺎﻫﺶ ﻣﻲﻳﺎﺑﺪ؟‬

‫• ﺗﺎﺛﻴﺮ ﻣﺤﻞ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﭼﻪ ﻣﻲﺑﺎﺷﺪ ﻭ ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﻧﺘﺨﺎﺏ ﻛﺮﺩ ﻛﻪ‬
‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻛﺎﻫﺶ ﺑﻴﺎﺑﺪ؟‬

‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﻻﮔﺮﺍﻧﮋ‬ ‫�‬

‫ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﺧﻄﺎﯼ ﺑﺮﺷ ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ⁃ﻓﺮﻣﻮﻝ ﺧﻄﺎﯼ ﻻﮔﺮﺍﻧﮋ‬ ‫ﻗﻀﯿﻪ ‪۹−۷‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ‪:‬‬

‫ﺍﻟﻒ( ]‪ [a, b‬ﺑﺎﺯﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻭ ﻣﺠﺰﺍﻱ ‪ x0 , x1 , . . . , xn‬ﺑﺎﺷﺪ )ﻳﻌﻨﻲ } ‪ a = min{x0 , · · · , xn‬ﻭ‬
‫} ‪(b = max{x0 , · · · , xn‬‬

‫ﺏ( ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ ‪n + 1‬ﺍﻡ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪.(f ∈ C n+1 [a, b]) ،‬‬

‫ﺍﮔﺮ )‪ Pn (x‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ‪ x0 , · · · , xn‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺭ ﺑﺎﺯﻩ ]‪ ،[a, b‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ ‪ f‬ﺑﺎ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ Pn‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬

‫)‪(x − x0 ) · · · (x − xn ) (n+1‬‬
‫= )‪f (x) − Pn (x‬‬ ‫‪f‬‬ ‫]‪(ηx ), x ∈ [a, b‬‬ ‫)‪(۱۵.۷‬‬
‫!)‪(n + 1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ ηx‬ﻋﺪﺩﻱ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﺳﺖ‪.‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻗﻀﻴﻪ ﻓﻮﻕ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻌﻲ ﺑﺮﻗﺮﺍﺭ ﺍﺳﺖ ﻛﻪ ﺩﺍﺭﺍﻱ ﻣﺸﺘﻘﺎﺕ ﭘﻴﻮﺳﺘﻪ ﺗﺎ ﻣﺮﺗﺒﻪ ‪ n + 1‬ﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﺗﺎﺑﻊ ﻣﺘﻌﻠﻖ ﺑﻪ‬
‫]‪ C n+1 [a, b‬ﻧﺒﺎﺷﺪ ﺁﻥﮔﺎﻩ ﻧﻤﻲﺗﻮﺍﻥ ﺍﺯ ﻗﻀﻴﻪ ﻓﻮﻕ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۹۰‬‬

‫ﻣﺜﺎﻝ ‪)۷−۷‬ﺗﺨﻤﯿﻦ ﺧﻄﺎﯼ ﺑﺮﺷ ﺗﺎﺑﻊ ‪(cos‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ )‪ p4 (x‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ )‪ f (x) = cos(x‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ‬

‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬ ‫‪π‬‬
‫= ‪x0 = 0, x1‬‬ ‫= ‪, x2 = , x3 = , x4‬‬
‫‪6‬‬ ‫‪4‬‬ ‫‪3‬‬ ‫‪2‬‬
‫ﺑﺎﺷﺪ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ ‪ cos‬ﺑﺎ ‪ p4‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [a, b] = [0, π2‬ﺑﻴﺎﺑﻴﻢ‪ .‬ﻃﺒﻖ ﻗﻀﻴﻪ ‪ ۹-۷‬ﻭ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ‬
‫ﻣﺸﺘﻖ ﭘﻨﺠﻢ ﺗﺎﺑﻊ ‪ cos‬ﺑﺮﺍﺑﺮ ‪ − sin‬ﺍﺳﺖ‪ ،‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫) ‪(x − 0)(x − π6 )(x − π4 )(x − π3 )(x − π2‬‬
‫= )‪f (x) − pn (x‬‬ ‫)) ‪(− sin(ηx‬‬
‫!)‪(4 + 1‬‬
‫ﺣﺎﻝ ﺍﮔﺮ ] ‪ ،x ∈ [0, π2‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪|x − 0| x −‬‬ ‫‪x − π4 x −‬‬
‫‪π‬‬ ‫‪π‬‬
‫‪x−‬‬ ‫‪π‬‬
‫= |)‪|f (x) − pn (x‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪2‬‬
‫|) ‪|− sin(ηx‬‬
‫!)‪(4 + 1‬‬
‫‪π π π π π‬‬
‫) () () () () (‬
‫)‪≤ 2 3 4 3 2 (1‬‬
‫!)‪(4 + 1‬‬
‫‪=0.0177 . . .‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺗﻘﺮﻳﺐ‪ ،‬ﺣﺪﺍﻛﺜﺮ ﺑﺮﺍﺑﺮ ‪ 0.018‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۸−۷‬ﺗﺨﻤﯿﻦ ﺧﻄﺎﯼ ﺑﺮﺷ ﺗﺎﺑﻊ ‪(cos‬‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺗﺎﺑﻊ )‪ f (x) = sin(x‬ﺭﺍ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﺶ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [a, b] = [0, π2‬ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ ،‬ﺑﻪ‬
‫ﻃﻮﺭﻱ ﻛﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺍﻳﻦ ﺗﻘﺮﻳﺐ ﺣﺪﺍﻛﺜﺮ ‪ 10−8‬ﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ ﺣﺼﻮﻝ ﺑﻪ ﺍﻳﻦ ﺩﻗﺖ‪ ،‬ﺍﺯ ﭼﻨﺪ ﻧﻘﻄﻪ ﮔﺮﻩﺍﻱ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ؟‬
‫ﻓﺮﺽ ﻛﻨﻴﻢ ‪ x0 , x1 , . . . , xn‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺮ ﻣﺒﻨﺎﻱ ﺁﻥﻫﺎ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ‬
‫ﻛﻪ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪π‬‬
‫‪xi = i 2n‬‬ ‫‪, i = 0, . . . , n.‬‬
‫ﺣﺎﻝ ﺍﮔﺮ ] ‪ ،x ∈ [0, π2‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫} ‪|x − xi | ≤ max{xi , π2 − xi‬‬
‫‪, 2 − i 2n‬‬
‫‪π π‬‬
‫‪= max{i 2n‬‬ ‫‪π‬‬
‫}‬
‫=‬ ‫‪π‬‬
‫‪2n‬‬ ‫}‪max{i, n − i‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪:‬‬
‫‪
π‬‬ ‫
‬ ‫
‬ ‫
‬
‫‪|(x − x0 )(x − x1 ) . . . (x − xn )| ≤ 2n‬‬
‫‪π‬‬
‫‪max{0, n} 2n‬‬ ‫‪max{1, n − 1} · · · 2n‬‬ ‫‪π‬‬
‫}‪max{n − 1, 1‬‬ ‫‪π‬‬
‫‪2n‬‬ ‫}‪max{n, 0‬‬
‫‪ π n+1‬‬
‫=‬ ‫‪n(n − 1) · · · ⌈ n2 ⌉⌈ n2 ⌉ · · · (n − 1)n‬‬
‫‪2n‬‬
‫ ‪ π n+1‬‬ ‫!‪n‬‬
‫‪2‬‬
‫=‬
‫‪2n‬‬ ‫!)‪(⌈ n2 ⌉ − 1‬‬
‫ﺍﺯ ﻃﺮﻓﻲ ﭼﻮﻥ ] ‪ ،ηx ∈ [0, π2‬ﻟﺬﺍ ‪ . f (n) (ηx ) ≤ 1‬ﺣﺎﻝ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺩﻭ ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﺩﺭ ﻓﺮﻣﻮﻝ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻻﮔﺮﺍﻧﮋ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۹۱‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﺩﺍﺭﻳﻢ‪:‬‬
‫)‪|(x − x0 )(x − x1 ) . . . (x − xn )| (n‬‬
‫= |)‪|f (x) − pn (x‬‬ ‫) ‪f (ηx‬‬
‫!)‪(n + 1‬‬
‫‬ ‫‪2‬‬
‫
‬
‫‪π n+1‬‬ ‫!‪n‬‬
‫‪n‬‬
‫‪2n‬‬ ‫!)‪(⌈ 2 ⌉−1‬‬
‫≤‬
‫!)‪(n + 1‬‬
‫ﺣﺎﻝ ﺑﺎ ﺳﻌﻲ ﻭ ﺧﻄﺎ ‪ n‬ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ‬
‫‬ ‫‪2‬‬
‫
‬
‫‪π n+1‬‬ ‫!‪n‬‬
‫‪n‬‬
‫‪2n‬‬ ‫!)‪(⌈ 2 ⌉−1‬‬
‫‪≤ 10−8‬‬
‫!)‪(n + 1‬‬

‫ﻧﺘﺎﻳﺠﻲ ﺍﺯ ﻗﻀﻴﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ‬ ‫�‬

‫■ ﺩﺭﻭﻧﻴﺎﺏ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬
‫ﺩﺭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺟﻤﻠﻪ )‪ f (n+1‬ﻇﺎﻫﺮ ﺷﺪﻩ ﺍﺳﺖ ﻭ ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﺍﮔﺮ ‪ f‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ n‬ﺑﺎﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺍﻳﻦ ﺟﻤﻠﻪ ﻫﻤﻮﺍﺭﻩ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ ﻭ ﻟﺬﺍ ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺻﻔﺮ ﺑﻮﺩﻥ ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﻪ ﻣﻌﻨﻲ ﺁﻥ ﺍﺳﺖ ﻛﻪ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺎ ﺧﻮﺩ ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﺗﻮﺿﻴﺤﺎﺕ‪ ،‬ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺭﺍ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪.‬‬

‫ﭼﻨﺪﺟﻤﻠەﺍﯼ‬ ‫ﻧﺘﯿﺠﻪ ‪۱۰−۷‬‬

‫ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ ﯾ‬
‫ﺍﮔﺮ ‪ f‬ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ k‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ‪ n‬ﻧﻘﻄﻪ ﻛﻪ ‪ n > k‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺧﻮﺩ‬
‫ﺗﺎﺑﻊ ‪.f‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺍﮔﺮ ‪ f (x) := x4‬ﺭﺍ ﺑﺎ ‪ ۵‬ﻧﻘﻄﻪ ﻳﺎ ﺑﻴﺸﺘﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﺑﺮ ‪ P(x) = x4‬ﺧﻮﺍﻫﺪ‬
‫ﺑﻮﺩ‪.‬‬

‫■ ﻫﻤﮕﺮﺍﻳﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ‬

‫ﻭ‬ ‫) ‪(x−x0 )...(x−xn‬‬
‫!)‪(n+1‬‬ ‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻄﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺣﺎﺻﻞﺿﺮﺏ ﺩﻭ ﺟﻤﻠﻪ‬
‫)‪ .f (n+1‬ﺣﺎﻝ ﺍﮔﺮ ‪ n‬ﺑﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻛﻨﺪ )ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻛﻨﺪ( ﺁﻥﮔﺎﻩ‪ ،‬ﺟﻤﻠﻪ ﺍﻭﻝ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ‬
‫ﻣﻲﻛﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‬ ‫‬
‫) ‪(x − x0 ) . . . (x − xn‬‬
‫‪lim‬‬ ‫∈ ‪= 0, x‬‬ ‫‪min xi , max xi .‬‬
‫∞→‪n‬‬ ‫!)‪(n + 1‬‬ ‫‪i=0,...,n‬‬ ‫‪i=0,...,n‬‬

‫ﺍﻣﺎ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻛﺮﺩﻥ ﺟﻤﻠﻪ ﺍﻭﻝ ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ‪ ،‬ﺩﻟﻴﻞ ﺑﺮ ﺁﻥ ﻧﻤﻲﺷﻮﺩ ﻛﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﺻﻔﺮ ﻛﻨﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﺟﻤﻠﻪ‬
‫) ‪ f (n+1) (ηx‬ﻧﻴﺰ ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻭ ﻣﻤﻜﻦ ﺍﺳﺖ ﺍﻳﻦ ﺟﻤﻠﻪ ﺑﻪ ﺳﻤﺖ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻛﻨﺪ ﻭ ﻧﺮﺥ ﺑﻪ ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ‬
‫ﻛﺮﺩﻥ ﺁﻥ ﺑﺰﺭﮒﺗﺮ ﺍﺯ ﻧﺮﺥ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻛﺮﺩﻥ ﺟﻤﻠﻪ ﺍﻭﻝ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﭼﻨﻴﻦ ﺣﺎﻟﺘﻲ‪ ،‬ﺑﺎ ﺍﻓﺰﺍﻳﺶ ‪ ،n‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺩﺭ ﺑﺮﺧﻲ ﻧﻘﺎﻁ ﺑﻪ‬
‫ﺑﻲﻧﻬﺎﻳﺖ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﻫﺮﭼﻨﺪ ﭼﻨﻴﻦ ﺣﺎﻟﺘﻲ ﺑﻪ ﻧﺪﺭﺕ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ‪ ،‬ﺍﻣﺎ ﻣﺜﺎﻝ ﻣﻌﺮﻭﻓﻲ ﻣﻮﺳﻮﻡ ﺑﻪ ﻣﺜﺎﻝ ﺭﺍﻧﮓ ‪ ۲۴‬ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ‬
‫ﺍﻳﻦ ﺍﺗﻔﺎﻕ ﻣﻲﺍﻓﺘﺪ ﻭ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺁﻥ ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺑﻪ ﺧﻮﺩ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪.‬‬
‫‪24 Runge‬‬ ‫‪Example‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۹۲‬‬

‫ﻣﺜﺎﻝ ‪)۹−۷‬ﻣﺜﺎﻝ ﺭﺍﻧﮓ(‬

‫‪ f (x) := 1+x‬ﻣﺒﺘﻨﻲ ﺑﺮ ‪ n + 1‬ﻧﻘﻄﻪ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [−5, 5‬ﺭﺍ ﺑﺎ )‪ Pn (x‬ﻧﺸﺎﻥ‬‫‪1‬‬
‫ﺍﮔﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪2‬‬

‫ﺩﻫﻴﻢ ﺁﻥﮔﺎﻩ ﺑﻪﺍﺯﺍﻱ ﺑﺮﺧﻲ ‪x‬ﻫﺎ ﺩﺭ ]‪ Pn (x) ̸→ f (x) [−5, 5‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ﺑﻪ‬
‫‪ dx‬ﺍﻓﺰﺍﻳﺶ ﻣﻲﻳﺎﺑﺪ ﻭ ﺩﺭ ﻧﻬﺎﻳﺖ ﻣﻘﺪﺍﺭ ﺧﻄﺎﻱ‬ ‫‪dn‬‬
‫ﺧﻮﺩ ﺗﺎﺑﻊ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪ .‬ﻋﻠﺖ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﺎ ﺍﻓﺰﺍﻳﺶ ‪ n‬ﻛﺮﺍﻥ ﺑﺎﻻﻱ )‪n f (x‬‬

‫ﺑﺮﺷﻲ ﺑﻪ ﺻﻔﺮ ﻫﻤﮕﺮﺍ ﻧﻤﻲﺷﻮﺩ‪ .‬ﺑﺮﺭﺳﻲ ﻋﺪﻡ ﻫﻤﮕﺮﺍﺋﻲ ﺩﺭ ]؟؟[ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻫﻤﻮﺍﺭﻩ ﻧﻤﻲﺗﻮﺍﻥ ﺑﺎ ﺍﻓﺰﺍﻳﺶ ‪ n‬ﺩﻗﺖ ﺗﻘﺮﻳﺐ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺭﺍ ﺑﺎﻻ ﺑﺮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ﻣﺸﺘﻘﺎﺕ ﺍﺯ ﻫﺮ‬
‫ﻣﺮﺗﺒﻪ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻛﺮﺍﻧﺪﺍﺭ ﺑﺎﺷﻨﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﭼﻮﻥ ﻫﻤﻮﺍﺭﻩ ﺟﻤﻠﻪ ﺍﻭﻝ ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪،‬‬
‫ﻟﺬﺍ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻧﻴﺰ ﺑﻪ ﺻﻔﺮ ﻣﻴﻞ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻳﻦ ﻧﻜﺘﻪ ﺩﺭ ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻫﻤ ﺮﺍﯾﯽ ﭼﻨﺪﺟﻤﻠەﺍﯼﻫﺎﯼ ﺩﺭﻭﻧﯿﺎﺏ ﺑﺮﺍﯼ ﺗﻮﺍﺑﻌ ﺑﺎ ﻣﺸﺘﻘﺎﺕ ﮐﺮﺍﻥﺩﺍﺭ‬ ‫ﻧﺘﯿﺠﻪ ‪۱۱−۷‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ‪ f‬ﺗﺎﺑﻌﻲ ﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻲﻧﻬﺎﻳﺖ ﻣﺸﺘﻖ ﭘﺬﻳﺮ ﺍﺳﺖ ﻭ ﻋﻼﻭﻩ ﺑﺮ ﺁﻥ ﻋﺪﺩ ﺣﻘﻴﻘﻲ ‪ β‬ﻭﺟﻮﺩ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‬
‫ﻛﻪ ﻗﺪﺭ ﻣﻄﻠﻖ ﻣﺸﺘﻘﺎﺕ ﺍﺯ ﻫﺮ ﻣﺮﺗﺒﻪ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﻛﻮﭼﻚﺗﺮ ﺍﺯ ‪ β‬ﺑﺎﺷﺪ ﻭ ‪ β‬ﻣﺴﺘﻘﻞ ﺍﺯ ‪ n‬ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‬
‫‪f (n) (x) < β, ∀n, ∀x ∈ [a, b].‬‬
‫ﺣﺎﻝ ﺍﮔﺮ )‪ Pn (x‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ‪ n + 1‬ﻧﻘﻄﻪ ﻣﻨﺎﺳﺐ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫‪lim Pn (x) = f (x), ∀x ∈ [a, b].‬‬
‫∞→‪n‬‬

‫ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻧﺘﻴﺠﻪ ﻓﻮﻕ‪ ،‬ﻫﻤﮕﺮﺍﻳﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻌﻲ ﻣﺎﻧﻨﺪ ‪ ex ،cos ،sin‬ﻭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﺗﻮﺍﺑﻊ ﺩﻳﮕﺮ ﻧﺘﻴﺠﻪ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱۰−۷‬ﻫﻤ ﺮﺍﺋ ﭼﻨﺪﺟﻤﻠەﺍﯼ ﺩﺭﻭﻧﯿﺎﺏ )‪( sin(x‬‬

‫ﻫﻤﮕﺮﺍﺋﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ )‪ f (x) = sin(x‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [−π, π‬ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦﻛﻪ‬
‫‪dn+1‬‬
‫‪sin(x) ≤ 1‬‬
‫‪dxn+1‬‬
‫ﻟﺬﺍ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺟﻤﻠﻪ ﺧﻄﺎ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﺻﻔﺮ ﻫﻤﮕﺮﺍ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﻘﺎﻳﺴﻪ ﺭﻭﺵﻫﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺑﺎ ﻳﻚﺩﻳﮕﺮ‬ ‫‪۶.۷‬‬

‫ﺭﻭﺷﻬﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺍﺯ ﻟﺤﺎﻅ ﺟﺒﺮﻱ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﻌﺎﺩﻝ ﻣﻲﺑﺎﺷﻨﺪ ﻭﻟﻲ ﺩﺭ ﻋﻤﻞ ﻭ ﺍﺯ ﺩﻳﺪﮔﺎﻩ ﻣﺤﺎﺳﺒﺎﺕ‬
‫ﻋﺪﺩﻱ ﺍﻳﻦ ﺭﻭﺵﻫﺎ ﺑﺎ ﻳﻚﺩﻳﮕﺮ ﻣﺘﻔﺎﻭﺕ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﭼﺮﺍ ﻛﻪ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻭ ﻣﻜﺎﻧﻲ ﻣﻮﺭﺩ ﻧﻴﺎﺯ ﺩﺭ ﻫﺮﻛﺪﺍﻡ ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪.‬‬
‫ﻫﻤﭽﻨﻴﻦ ﺩﻗﺖ ﻧﺘﺎﻳﺞ ﻧﻴﺰ ﺍﺯ ﺟﻤﻠﻪ ﺗﻔﺎﻭﺕﻫﺎ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﻋﻤﻞ ﻭﻗﺘﻲ ‪ n‬ﻛﻮﭼﻚ ﺑﺎﺷﺪ ﺁﻥﮔﺎﻩ ﺍﺯ ﻫﺮ ﺭﻭﺵ ﺩﻟﺨﻮﺍﻩ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻧﻤﻮﺩ‪ .‬ﻭﻟﻲ ﺍﮔﺮ ‪ n‬ﺑﺰﺭﮒ ﺑﺎﺷﺪ ﺭﻭﺵﻫﺎﻱ ﺗﻔﺎﺿﻼﺕ ﻣﺘﻨﺎﻫﻲ ﻭ ﺗﻔﺎﺿﻼﺕ ﺗﻘﺴﻴﻢ ﺷﺪﻩ ﺍﺭﺟﺢ ﻣﻴﺒﺎﺷﻨﺪ‪.‬‬

‫ﻣﻘﺎﻳﺴﻪ ﭘﻴﭽﻴﺪﮔﻲ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ‬ ‫�‬

‫ﺣﺬﻑ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۹۳‬‬ ‫ﺩﺭﻭﻧﯿﺎﺑﯽ‬ ‫ﻓﺼﻞ ﻫﻔﺘﻢ‪:‬‬

‫ﻣﻘﺎﻳﺴﻪ ﭘﺎﻳﺪﺍﺭﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﻭﺵﻫﺎﻱ ﻣﺴﺘﻘﻴﻢ‪ ،‬ﻻﮔﺮﺍﻧﮋ ﻭ ﻧﻴﻮﺗﻦ‬ ‫�‬

‫ﺣﺬﻑ‬

‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ‬ ‫‪۷.۷‬‬

‫ﺍﮔﺮ )‪ f (x‬ﺗﺎﺑﻌﻲ ﺑﺎﺷﺪ ﻛﻪ ﺩﺭ ]‪ [a, b‬ﺩﺍﺭﺍﻱ ﻣﻌﻜﻮﺱ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺎ ﻛﻤﻚ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻲﺗﻮﺍﻥ ﻣﻌﻜﻮﺱ ﺗﺎﺑﻊ )‪ f (x‬ﻳﻌﻨﻲ‬
‫)‪ f −1 (x‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺯﺩ‪ .‬ﺍﻳﻦ ﻛﺎﺭ ﺑﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫ﻳﺎﻓﺘﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ )‪ .f −1 (x‬ﻳﺎﻓﺘﻦ ﺩﺭﻭﻧﻴﺎﺏ ﻣﻌﻜﻮﺱ ﻳﻚ ﺗﺎﺑﻊ ﻣﺎﻧﻨﺪ )‪ f (x‬ﺑﻪ ﺳﺎﺩﮔﻲ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ‬
‫ﺻﻮﺭﺕ ﻛﻪ ﻧﻘﺎﻁ ‪ xi , i = 0, · · · , n‬ﺭﺍ ﺍﻧﺘﺨﺎﺏ ﻣﻴﻜﻨﻴﻢ ﻭ ﺳﭙﺲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﻣﻌﻜﻮﺱ ﻫﻤﺎﻥ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ‬
‫ﻧﻘﺎﻁ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫‪(f (xi ), xi ) , i = 0, · · · , n‬‬
‫ﻛﺎﺭﺑﺮﺩ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻌﻜﻮﺱ ﺯﻣﺎﻧﻲ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺭﺍﺣﺘﻲ ﻧﻤﻲﺗﻮﺍﻧﻴﻢ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ﻣﻌﻜﻮﺱ ﺭﺍ ﺑﻴﺎﺑﻴﻢ ﻭﻟﻲ ﺍﺯ ﻭﺟﻮﺩ ﻣﻌﻜﻮﺱ ﺗﺎﺑﻊ ﺑﺎ ﺧﺒﺮ‬
‫ﻫﺴﺘﻴﻢ‪.‬‬

‫ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻨﺤﻨﻴﻬﺎ‬ ‫‪۸.۷‬‬

‫ﺭﻭﺷﻬﺎﻱ ﺍﺭﺍﺋﻪ ﺷﺪﻩ ﺑﺮﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﺗﻮﺍﺑﻊ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺩﺭﻭﻧﻴﺎﺑﻲ ﻣﻨﺤﻨﻴﻬﺎ ﻧﻴﺰ ﺑﺴﻂ ﺩﺍﺩ‪ .‬ﺑﺎ ﺍﻳﻦ ﺷﺮﻁ ﻛﻪ ﺑﺘﻮﺍﻧﻴﻢ ﻣﻨﺤﻨﻲ ﺭﺍ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﭘﺎﺭﺍﻣﺘﺮﻱ ﺗﻮﺻﻴﻒ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﻣﻨﺤﻨﻲ ‪ f (x, y) = 0‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ ﻛﻪ ﺩﺍﺭﺍﻱ ﻓﺮﻡ ﭘﺎﺭﺍﻣﺘﺮﻱ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫(‬
‫)‪x = fx (t‬‬
‫)‪y = fy (t‬‬
‫ﺣﺎﻝ ﺍﮔﺮ )‪ pn,x (t‬ﻭ )‪ pn,y (t‬ﺑﻪ ﺗﺮﺗﻴﺐ ﭼﻨﺪﺟﻤﻠﻪﺍﻳﻬﺎﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﻮﺍﺑﻊ )‪ fx (t‬ﻭ )‪ fy (t‬ﺑﺎﺷﻨﺪ ﺁﻥﮔﺎﻩ ﻣﻨﺤﻨﻲ ﭘﺎﺭﺍﻣﺘﺮﻱ‬
‫(‬
‫)‪x = pn,x (t‬‬
‫)‪y = pn,y (t‬‬
‫ﺭﺍ ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﻨﺤﻨﻲ ﺩﺭﻭﻧﻴﺎﺏ ﻣﻨﺤﻨﻲ ‪ f (x, y) = 0‬ﺩﺭ ﻧﻈﺮ ﻣﻴﮕﻴﺮﻳﻢ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۹۴‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۸‬‬
‫ﺑﺮﺍﺯﺵ‬
‫ﻓﺼﻞ ﻫﺸﺘﻢ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۱۹۶‬‬ ‫ﺍﻧﮕﻴﺰﻩ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۸‬‬
‫‪۱۹۷‬‬ ‫ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۸‬‬
‫‪۲۰۰‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۲.۸‬‬
‫‪۲۰۱‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۲.۸‬‬
‫‪۲۰۳‬‬ ‫∗ ﺑﺮﺍﺯﺵ ﻳﻚﮔﺎﻧﻪ ﻭ ﭼﻨﺪﮔﺎﻧﻪ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۲.۸‬‬
‫‪۲۰۵‬‬ ‫ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻭ ﻏﻴﺮﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۸‬‬
‫‪۲۰۶‬‬ ‫ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺧﻲ ﺑﺮﺍﺯﺵﻫﺎﻱ ﻏﻴﺮﺧﻄﻲ ‪. . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۳.۸‬‬
‫‪۲۰۹‬‬ ‫∗ ﻧﻜﻮﻳﻲ ﺑﺮﺍﺯﺵ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۸‬‬

‫‪۱۹۵‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۹۶‬‬

‫ﻣﺒﺤﺚ ﺑﺮﺍﺯﺵ ﻳﻚ ﻣﺒﺤﺚ ﺑﻴﻦ ﺭﺷﺘﻪﺍﻱ ﺍﺳﺖ ﻛﻪ ﺩﺍﺭﺍﻱ ﻛﺎﺭﺑﺮﺩﻫﺎﻱ ﺯﻳﺎﺩﻱ ﺩﺭ ﻋﻠﻮﻡ ﻣﻬﻨﺪﺳﻲ‪ ،‬ﺁﻣﺎﺭ‪ ،‬ﻋﻠﻮﻡ ﺭﻳﺎﺿﻲ ﻭ ﻏﻴﺮﻩ‬
‫ﺩﺍﺭﺩ‪ .‬ﺩﺭ ﺭﺷﺘﻪﻫﺎﻱ ﻣﺘﻔﺎﻭﺕ ﺍﺯ ﻭﺍﮊﻩﻫﺎﻱ‬
‫ﺭﮔﺮﺳﻴﻮﻥ ‪۱‬‬ ‫•‬
‫ﻣﺮﺑﻌﺎﺕ ‪۲‬‬ ‫• ﺗﻘﺮﻳﺐ ﻛﻤﺘﺮﻳﻦ‬
‫ﺗﻘﺮﻳﺐ ‪۳‬‬ ‫• ﺑﻬﺘﺮﻳﻦ‬
‫ﺑﻪ ﺟﺎﻱ ﻭﺍﮊﻩ ﺑﺮﺍﺯﺵ ﻣﻨﺤﻨﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﺍﻧﮕﻴﺰﻩ‬ ‫‪۱.۸‬‬
‫ﺍﺑﺘﺪﺍ ﺑﺎ ﻳﻚ ﻣﺜﺎﻝ ﺍﻧﮕﻴﺰﻩ ﺑﺮﺍﺯﺵ ﻣﻨﺤﻨﻲ ﺭﺍ ﺗﻮﺿﻴﺢ ﻣﻲﺩﻫﻴﻢ‪ .‬ﻓﺮﺽ ﻛﻨﻴﻢ ﻳﻚ ﮔﺎﺯ ﺩﺍﺭﻳﻢ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺛﺎﺑﺖ ﻓﺸﺎﺭ ﮔﺎﺯ ﺭﺍ ﺩﺭ‬
‫ﺁﺯﻣﺎﻳﺸﮕﺎﻩ ﺑﻪ ﺩﺳﺖ ﺑﻴﺂﻭﺭﻳﻢ‪ .‬ﺍﺯ ﻓﻴﺰﻳﻚ ﻣﻲﺩﺍﻧﻴﻢ ﻛﻪ ﻓﺸﺎﺭ ﻳﻚ ﮔﺎﺯ ﺗﺎﺑﻌﻲ ﺍﺯ ﺩﻣﺎﻱ ﮔﺎﺯ ﺍﺳﺖ ﻭ ﺍﻳﻦ ﺗﺎﻳﻊ ﻳﻚ ﺗﺎﺑﻊ ﺧﻄﻲ ﺍﺳﺖ‪.‬‬
‫ﺑﻪ ﺯﺑﺎﻥ ﺭﻳﺎﺿﻲ‪ ،‬ﺍﮔﺮ )‪ p(t‬ﻓﺸﺎﺭ ﮔﺎﺯ ﺩﺭ ﺩﻣﺎﻱ ‪ t‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‬
‫‪p(t) = kt + c‬‬
‫ﻛﻪ ‪ k‬ﻭ ‪ c‬ﺍﻋﺪﺍﺩﻱ ﺣﻘﻴﻘﻲ ﻫﺴﺘﻨﺪ ﻭ ‪ k‬ﺑﻪ ﺛﺎﺑﺖ ﻓﺸﺎﺭ ﮔﺎﺯ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ‪ k‬ﺿﺮﻳﺐ ﺯﺍﻭﻳﻪ ﺧﻂ ﻓﻮﻕ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﮔﺎﺯﻫﺎﻱ‬
‫ﻣﺘﻔﺎﻭﺕ‪ ،‬ﻣﻘﺪﺍﺭ ‪ k‬ﻣﺘﻔﺎﻭﺕ ﺍﺳﺖ‪.‬‬
‫ﺑﺮﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩﻥ ﺛﺎﺑﺖ ﻓﺸﺎﺭ ﺩﺭ ﺁﺯﻣﺎﻳﺸﮕﺎﻩ ﺑﺎ ﻛﻤﻚ ﺁﺯﻣﺎﻳﺶ ﻓﺸﺎﺭ ﮔﺎﺯ ﺩﺍﺩﻩ ﺷﺪﻩ ﺭﺍ ﺩﺭ ﺩﻭ ﺩﻣﺎﻱ ‪ t1‬ﻭ ‪ t2‬ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺩﻭ ﺩﻣﺎ‪ ،‬ﻓﺸﺎﺭ ﮔﺎﺯ ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﺮﺍﺑﺮ ) ‪ p1 := p(t1‬ﻭ ) ‪ p2 := p(t2‬ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ‬
‫ﻗﺮﺍﺭ ﻣﻲ ﺩﻫﻴﻢ‪:‬‬
‫(‬
‫‪p1 = kt1 + c‬‬
‫‪p2 = kt2 + c‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺑﺮ ﺣﺴﺐ ‪ k‬ﻭ ‪ c‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﺎ ﺣﻞ ﺁﻥ ﻣﻘﺎﺩﻳﺮ ‪ k‬ﻭ ‪ c‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﺭﻭﺍﻝ ﻓﻮﻕ ﺩﺭ ﻭﺍﻗﻊ ﺍﺯ ﺍﻳﺪﻩ‬
‫ﺩﺭﻭﻥ ﻳﺎﺑﻲ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩﻳﻢ ﻭ ﺑﻪ ﺭﻭﺵ ﻣﺴﺘﻘﻴﻢ ﺿﺮﺍﻳﺐ ﺧﻂ ﺩﺭﻭﻧﻴﺎﺏ ﺩﻭ ﻧﻘﻄﻪ ) ‪ (t1 , p1‬ﻭ ) ‪ (t2 , p2‬ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩﻩﺍﻳﻢ‪.‬‬
‫ﺍﻣﺎ ﺭﻭﻳﻪ ﻓﻮﻕ ﻳﻚ ﺍﻳﺮﺍﺩ ﺩﺍﺭﺩ ﻭ ﺁﻥ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﺑﻪ ﻋﻠﺖ ﺧﻄﺎﻳﻲ ﻛﻪ ﺩﺭ ﻭﺳﺎﻳﻞ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﺁﺯﻣﺎﻳﺸﮕﺎﻩ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻣﻘﺪﺍﺭ‬
‫ﻛﻤﻴﺖﻫﺎﻱ ‪ t1‬ﻭ ‪ t2‬ﻭ ﻫﻤﭽﻨﻴﻦ ‪ t1‬ﻭ ‪ t2‬ﺑﺎ ﺧﻄﺎﻱ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻫﻤﺮﺍﻩ ﻫﺴﺘﻨﺪ ﻭ ﺧﻄﺎﻱ ﻣﻮﺟﻮﺩ ﺩﺭ ﺍﻳﻦ ﺩﺍﺩﻩﻫﺎ ﺑﻪ ﺛﺎﺑﺖ ﻓﺸﺎﺭ‬
‫ﻣﺤﺎﺳﺒﻪ ﺷﺪﻩ ﻧﻴﺰ ﺍﻧﺘﺸﺎﺭ ﻣﻲﻳﺎﺑﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺛﺎﺑﺖ ﻓﺸﺎﺭ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺩﺭ ﺩﺭ ﺭﻭﺍﻝ ﻓﻮﻕ ﺑﺎ ﺧﻄﺎ ﻫﻤﺮﺍﻩ ﻣﻲﺑﺎﺷﺪ ﻭ ﻣﻤﻜﻦ‬
‫ﺍﺳﺖ ﺍﻳﻦ ﺧﻄﺎ ﭼﺸﻤﮕﻴﺮ ﺑﺎﺷﺪ‪.‬‬
‫ﺑﺮﺍﻱ ﺁﻥ ﻛﻪ ﺧﻄﺎﻱ ﻓﺮﺍﻳﻨﺪ ﻓﻮﻕ ﺭﺍ ﻛﺎﻫﺶ ﺩﻫﻴﻢ ﻭ ﻣﻘﺪﺍﺭ ﺛﺎﺑﺖ ﻓﺸﺎﺭ ﺭﺍ ﺑﺎ ﺩﻗﺖ ﺑﻴﺸﺘﺮﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺩﻭ‬
‫ﻃﺮﻳﻖ ﺯﻳﺮ ﻋﻤﻞ ﻛﺮﺩ‪:‬‬
‫‪ .۱‬ﺍﺯ ﻭﺳﺎﻳﻞ ﺁﺯﻣﺎﻳﺸﮕﺎﻫﻲ ﺩﻗﻴﻖﺗﺮﻱ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪،‬‬
‫‪ .۲‬ﺍﺯ ﺍﻳﺪﻩ ﺑﺮﺍﺯﺵ ﻛﻤﻚ ﮔﺮﻓﺖ‪.‬‬
‫ﺍﻳﺪﻩ ﺍﻭﻝ ﻣﻘﺮﻭﻥ ﺑﻪ ﺻﺮﻓﻪ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﻏﻴﺮ ﻣﻤﻜﻦ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺍﻳﺪﻩ ﺑﺮﺍﺯﺵ ﺳﺎﺩﻩﺗﺮ ﻭ ﻋﻤﻠﻲﺗﺮ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ‬
‫ﺍﻳﻦ ﺍﻳﺪﻩ ﺑﻪ ﺟﺎﻱ ﺍﻧﺠﺎﻡ ﺩﻭ ﺁﺯﻣﺎﻳﺶ ﻭ ﮔﺮﻓﺘﻦ ﺩﻭ ﻧﻤﻮﻧﻪ‪ ،‬ﺁﺯﻣﺎﻳﺶﻫﺎﻱ ﺑﻴﺸﺘﺮﻱ ﺍﻧﺠﺎﻡ ﻣﻲﺷﻮﺩ ﻭ ﻧﻤﻮﻧﻪﻫﺎﻱ ﺑﻴﺸﺘﺮ‬
‫) ‪(t1 , p1 ), (t2 , p2 ), . . . , (tn , pn‬‬
‫‪1 Regression‬‬ ‫‪3 Best‬‬ ‫‪approximation‬‬
‫‪2 Least‬‬ ‫‪square approximation‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫ﺍﺧﺬ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺳﭙﺲ ﻳﻚ ﺧﻂ ‪ p(t) = ct + c0‬ﺑﺮ ﻧﻘﺎﻁ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺮﺍﺯﺵ )ﻓﻴﺖ( ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺒﺎﻧ آﻧﺎﻟﯿﺰ ﻋﺪدی‬ ‫‪4‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪pm‬‬ ‫‪pm‬‬
‫‪i‬‬ ‫‪i‬‬

‫ﺑﺮازش‬
‫‪p3‬‬ ‫‪p3‬‬
‫‪p۱۹۷‬‬ ‫‪p2‬‬
‫‪2‬‬
‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬
‫‪p1‬‬ ‫‪p1‬‬
‫ﺍﺧﺬ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺳﭙﺲ ﻳﻚ ﺧﻂ ‪ p(t) = ct + c0‬ﺑﺮ ﻧﻘﺎﻁ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺮﺍﺯﺵ )ﻓﻴﺖ( ﻣﻲﺷﻮﺩ‪.‬‬
‫‪t1‬‬ ‫‪t2‬‬ ‫‪t3‬‬ ‫‪tm‬‬ ‫‪t1‬‬ ‫‪t2‬‬ ‫‪t3‬‬ ‫‪tm‬‬

‫‪pn‬‬ ‫‪pn‬‬

‫ﺑﺮﺍﺯﺵ‬
‫‪p3‬‬ ‫‪p3‬‬
‫‪p2‬‬ ‫‪p2‬‬

‫‪p1‬‬ ‫‪p1‬‬

‫‪t1‬‬ ‫‪t2‬‬ ‫‪t3‬‬ ‫‪tn‬‬ ‫‪t1‬‬ ‫‪t2‬‬ ‫‪t3‬‬ ‫‪tn‬‬

‫ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﻣﻨﻈﻮﺭ ﺍﺯ ﺑﺮﺍﺯﺵ ﺩﺍﺩﻥ ﺧﻂ ﺑﺮ ﻧﻘﺎﻁ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻌﻴﻴﻦ ﻣﻘﺪﺍﺭ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c‬ﻭ ‪ c0‬ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻛﻪ ﺧﻂ‬
‫ﻧﻘﺎﻁﻛﻪ ﺧﻂ‬
‫ﺩﺍﺷﺘﻪ‬ ‫ﮔﻮﻧﻪﺍﻱ‬
‫ﻛﻤﺘﺮﻱ‪ c‬ﺍﺯﺑﻪ ﺍﻳﻦ‬
‫ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﻣﻨﻈﻮﺭ ﺍﺯ ﺑﺮﺍﺯﺵ ﺩﺍﺩﻥ ﺧﻂ ﺑﺮ ﻧﻘﺎﻁ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﺗﻌﻴﻴﻦ ﻣﻘﺪﺍﺭ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c‬ﻭ ‪0‬‬
‫ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ‪ p(t) = ct + c0‬ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻪ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ﻧﺰﺩﻳﻚ ﺑﺎﺷﺪ ﻭ ﺍﻧﺤﺮﺍﻑ‬
‫ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ‪ p(t) = ct + c0‬ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺑﻪ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ﻧﺰﺩﻳﻚ ﺑﺎﺷﺪ ﻭ ﺍﻧﺤﺮﺍﻑ ﻛﻤﺘﺮﻱ ﺍﺯ ﺍﻳﻦ ﻧﻘﺎﻁ‬
‫ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺑﻪ ﻟﺤﺎﻅ ﺭﻳﺎﺿﻲ ﺑﺎﻳﺪ ﻣﻌﻴﺎﺭ ﻣﺸﺨﺼﻲ ﺑﺮﺍﻱ ﺗﻌﻴﻦ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪.‬‬
‫ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﺑﻪ ﻟﺤﺎﻅ ﺭﻳﺎﺿﻲ ﺑﺎﻳﺪ ﻣﻌﻴﺎﺭ ﻣﺸﺨﺼﻲ ﺑﺮﺍﻱ ﺗﻌﻴﻦ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ‬
‫ﺑﺮﺍﺯﺵ ‪ (ti , pi ), i = 1, . . .‬ﺑﺎﺷﺪ‪ .‬ﻃﺒﻖ ﺁﺯﻣﺎﻳﺸﺎﺕ‪ ،‬ﻓﺸﺎﺭ ﮔﺎﺯ ﺩﺭ ﺩﻣﺎﻱ ‪ti‬‬
‫ﺧﻂﻫﺎﻱ ‪, n‬‬‫)‪p(t‬ﻧﻤﻮﻧﻪ‬
‫=‪:‬ﻛﻨﻨﺪﻩ‬
‫ﺑﺮﺍﺯﺵ‬
‫‪ct +‬‬ ‫ﻓﺮﺽ=‪:‬ﻛﻨﻴﺪ‬
‫)‪c0p(t‬ﺧﻂ‬ ‫ﻣﻨﻈﻮﺭ‪ct +،‬‬
‫ﺑﺮﺍﻱ ﺍﻳﻦ ‪c0‬‬
‫ﮔﺎﺯ ﺩﺭ ﺩﻣﺎﻱ ‪ ti‬ﺑﺮﺍﺑﺮ ‪ p(ti ) = cti + c0‬ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ‪ ti‬ﺑﻪ ﺍﻧﺪﺍﺯﻩ‬
‫ﻃﺒﻖ‬‫ﺑﺎﺷﺪ‪ .‬ﻓﺸﺎﺭ‬
‫ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ‪،‬‬
‫ﺧﻂ‪(ti , p‬‬
‫ﻣﻄﺎﺑﻖ ﺑﺎ‪i ), i‬‬
‫‪= 1,‬‬ ‫ﺍﺳﺖ‪..‬ﺍﻣﺎ‬ ‫ﻛﻨﻨﺪﻩ ﻧﻤﻮﻧﻪ ﺑﺮﺍﺑﺮ‬
‫ﻫﺎﻱ ‪. . , npi‬‬

‫‪pn‬‬ ‫ﻣﻄﺎﺑﻖ‪eni +‬‬

‫ﺁﺯﻣﺎﻳﺸﺎﺕ‪ ،‬ﻓﺸﺎﺭ ﮔﺎﺯ ﺩﺭ ﺩﻣﺎﻱ ‪ ti‬ﺑﺮﺍﺑﺮ ‪ pi‬ﺍﺳﺖ‪ .‬ﺍﻣﺎ‬
‫‪ei := pi − (ct‬‬ ‫) ‪c0‬‬
‫)‪e2 = p2 − (kt2 + c‬‬

‫‪c‬‬
‫ﺑﺮﺍﺑﺮ =‬
‫ﻛﻨﻨﺪﻩ )ﺭﺍ ‪i‬ﺑﻪ‪p(t‬ﮔﻮﻧﻪﺍﻱ ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ‪kt +‬ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎﻫﺎﻱ ﻓﻮﻕ ﻣﻲﻧﻴﻤﻢ ﮔﺮﺩﺩ‪ .‬ﺑﻪ‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﺑﺮﺍﺯﺵﺩﻣﺎﻱ‬
‫ﺧﻂ ‪ti‬‬ ‫ﮔﺎﺯ ﺩﺭ‬ ‫ﺩﺍﺭﻳﻢ‪.‬ﻓﺸﺎﺭ‬
‫ﺣﺎﻝ ﺩﺭ‬ ‫ﺑﺎ ﺧﻂ ﺑﺮﺍﺯﺵﺧﻄﺎﻛﻨﻨﺪﻩ‪،‬‬
‫ﺍﻧﺪﺍﺯﻩ‪ c0‬ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ ﻣﻘﺪﺍﺭ ﺯﻳﺮ ﻣﻴﻨﻴﻤﻢ ‪t‬‬ ‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺭ ‪ ti‬ﺑﻪ‬ ‫‪ cti + c0‬ﺍﺳﺖ‪.‬‬
‫=‬
‫ﮔﺮﺩﺩ‪:‬‬
‫(‪p‬‬
‫)‬ ‫ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c‬ﻭ‬ ‫ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪،‬‬
‫∑‬
‫‪m‬‬
‫]) ‪[pi − (cti + c0‬‬
‫‪2‬‬ ‫) ‪ei := pi − (cti + c0‬‬
‫ﺧﻄﺎ ﺩﺍﺭﻳﻢ‪ .‬ﺣﺎﻝ ﺩﺭ ﺑﺮﺍﺯﺵ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ‬
‫‪i=1‬‬
‫‪p3‬‬ ‫‪e3‬‬ ‫ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫‪p2‬‬ ‫ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎﻫﺎﻱ ﻓﻮﻕ ﻣﻲﻧﻴﻤﻢ ﮔﺮﺩﺩ‪ .‬ﺑﻪ‬
‫ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c‬ﻭ ‪ c0‬ﺭﺍ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﻲﻳﺎﺑﻴﻢ ﻛﻪ‬
‫)‪e1 = p1 − (kt1 + c‬‬
‫‪p1‬‬ ‫ﻣﻘﺪﺍﺭ ﺯﻳﺮ ﻣﻴﻨﻴﻤﻢ ﮔﺮﺩﺩ‪:‬‬
‫‪i‬‬ ‫‪X‬‬
‫‪m‬‬ ‫‪i‬‬
‫‪i‬‬ ‫‪t1‬‬ ‫‪t2‬‬ ‫‪t3‬‬ ‫‪tn‬‬ ‫]) ‪[pi − (cti + c0‬‬
‫‪2‬‬ ‫‪i‬‬
‫‪i=1‬‬
‫‪ii‬‬ ‫‪i‬‬ ‫‪i‬‬

‫ﻣﻌﺮﻓﻲ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‬ ‫‪۲.۸‬‬

‫ﺩﺭ ﺳﺎﺩﻩﺗﺮﻳﻦ ﺣﺎﻟﺖ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﺗﻌﺪﺍﺩﻱ ﻧﻘﻄﻪ ﺑﻪ ﺻﻮﺭﺕ‬
‫) ‪(x1 , y1 ), (x2 , y2 ), . . . , (xn , yn‬‬
‫ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻋﻤﻮﻣﺎ‪ ،‬ﺍﻳﻦ ﻧﻘﺎﻁ ﺩﺍﺩﻩﻫﺎﻱ ﺁﺯﻣﺎﻳﺶﮔﺎﻫﻲ ﻭ ﻳﺎ ﺩﺍﺩﻩﻫﺎﻱ ﺁﻣﺎﺭﮔﻴﺮﻱ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ ﻏﺎﻟﺒﺎ ﺍﺯ ﻭﺍﮊﻩ ﻧﻤﻮﻧﻪ ‪ ۴‬ﺑﺮﺍﻱ ﺍﺷﺎﺭﻩ‬
‫ﺑﻪ ﺍﻳﻦ ﻧﻘﺎﻁ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﻋﻼﻭﻩ ﺑﺮ ﻧﻘﺎﻁ ﻳﺎ ﻧﻤﻮﻧﻪﻫﺎ‪ ،‬ﻳﻚ ﺩﺳﺘﻪ ﺍﺯ ﺗﻮﺍﺑﻊ ﻧﻴﺰ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﻦ ﺩﺳﺘﻪ ﺗﻮﺍﺑﻊ‬
‫ﺭﺍ ﺑﺎ ‪ F‬ﻧﺸﺎﻥ ﻣﻲﺩﻫﻴﻢ ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺩﺳﺘﻪ ﺗﻮﺍﺑﻊ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻳﺎ ﺩﺳﺘﻪ ﺗﻮﺍﺑﻊ ﻧﻤﺎﻳﻲ‪ .‬ﻓﺮﺽ ﻣﻲﻛﻨﻴﻢ ﻫﺮ ﻋﻀﻮ ﺍﻳﻦ ﺩﺳﺘﻪ‬
‫‪4 Sample‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۱۹۸‬‬

‫ﺗﻮﺍﺑﻊ ﺑﻪ ﺻﻮﺭﺕ‬
‫) ‪y = ϕ(x; c1 , . . . , cn‬‬
‫ﺑﺎﺷﺪ‪ .‬ﺗﺎﺑﻊ ‪ ϕ‬ﺑﻪ ﺗﺎﺑﻊ ﻣﺪﻝ ‪ ۵‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ ﻭ ‪ x‬ﻣﺘﻐﻴﺮ ﻭﺍﺑﺴﺘﻪ ﺁﻥ ﻣﻲﺑﺎﺷﺪ ﻭ ‪ c1 , . . . , cm‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻳﻲ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﺿﺎﺑﻄﻪ‬
‫ﺗﺎﺑﻊ ﻣﺪﻝ ﻭﺟﻮﺩ ﺩﺍﺭﻧﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺑﻪ ﺍﺯﺍﻱ ﻣﻘﺎﺩﻳﺮ ﻣﺘﻔﺎﻭﺕ ﺍﺯ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c1 , . . . , cm‬ﺍﻋﻀﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺍﺯ ﺩﺳﺘﻪ ﺗﻮﺍﺑﻊ ‪ F‬ﺣﺎﺻﻞ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺍﺯ ﻳﻚ ﺩﻳﺪﮔﺎﻩ‪ ،‬ﻣﺪﻝ ‪ ϕ‬ﺿﺎﺑﻄﻪﺍﻱ ﺍﺳﺖ ﻛﻪ ﺑﺮ ﻧﻤﻮﻧﻪﻫﺎ ﺣﺎﻛﻢ ﺍﺳﺖ ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﺪﻝ ‪ ϕ‬ﺍﺭﺗﺒﺎﻁ ﺑﻴﻦ ﻣﻮﻟﻔﻪﻫﺎﻱ‬
‫ﻧﻤﻮﻧﻪﻫﺎ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ ﻣﺪﻝ ﻛﺎﻣﻼ ﻣﺸﺨﺺ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﺩﺭ ﺁﻥ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻟﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬
‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﻫﺪﻑ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﺗﺎﺑﻌﻲ ﺍﺯ ﺩﺳﺘﻪ ﺗﻮﺍﺑﻊ ‪ F‬ﺭﺍ ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ ﻛﻪ ﺑﻴﺸﺘﺮﻳﻦ ﺗﻄﺎﺑﻖ ﺑﺎ ﻧﻤﻮﻧﻪﻫﺎ ﺩﺍﺷﺘﻪ ﺑﺎﺷﺪ‪.‬‬
‫ﺑﻪ ﺍﺻﻄﻼﺡ‪ ،‬ﻫﺪﻑ ﺍﻳﻦ ﺍﺳﺖ ﻛﻪ ﻳﻚ ﻋﺼﻮ ﺍﺯ ‪ F‬ﺭﺍ ﺑﺮ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪ ﺑﺮﺍﺯﺵ ﺩﻫﻴﻢ‪.‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ‬

‫» ﻣﻴﺰﺍﻥ ﺗﻄﺎﺑﻖ ﻳﻚ ﻋﻀﻮ ﺍﺯ ‪ F‬ﺑﺎ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ﺑﺎ ﭼﻪ ﻣﻌﻴﺎﺭ ﻭ ﻗﺎﻋﺪﻩﺍﻱ ﺗﻌﺮﻳﻒ ﺷﻮﺩ؟ «‬

‫ﺩﺭ ﭘﺎﺳﺦ ﻣﻲﺗﻮﺍﻥ ﮔﻔﺖ ﻛﻪ ﺗﻌﺮﻳﻒﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﺮﺍﻱ ﻣﻴﺰﺍﻥ ﺗﻄﺎﺑﻖ ﻣﻲﺗﻮﺍﻥ ﺍﺭﺍﻳﻪ ﺩﺍﺩ‪ .‬ﺍﻣﺎ ﻏﺎﻟﺒﺎ‪ ،‬ﺍﻧﺪﺍﺯﻩ ﺗﻄﺎﺑﻖ ﻳﻚ ﻋﻀﻮ‬
‫) ‪ y = ϕ(x; c1 , . . . , cm‬ﺑﺎ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ) ‪ (x1 , y1 ), . . . , (xn , yn‬ﺑﺮﺍﺑﺮ ﻣﻘﺪﺍﺭ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪X‬‬
‫‪n‬‬
‫‬ ‫‪2‬‬
‫=‪SSR :‬‬ ‫‪ϕ(xi ; c1 , . . . , cm ) − yi‬‬
‫‪i=1‬‬
‫ﺑﻪ ﻃﻮﺭ ﺧﻼﺻﻪ‪ ،‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c1 , . . . , cm‬ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺗﻌﻴﻴﻦ ﻣﻲﺷﻮﻧﺪ ﻛﻪ ﻣﻘﺪﺍﺭ ‪ SSR‬ﻣﻴﻨﻴﻤﻢ ﺷﻮﺩ‪.‬‬
‫ﻛﻤﻴﺖ ‪ SSR‬ﺑﻪ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﻣﺎﻧﺪﻩ ‪ ۶‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺩﺭ ﺑﺮﺧﻲ ﻣﻨﺎﺑﻊ ﺍﺭ ﻭﺍﮊﻩ ﻭ ﻳﺎ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﻣﺎﻧﺪﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪۷‬‬

‫ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﻭ ﻫﻤﻴﻨﻄﻮﺭ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎ‪ ،‬ﻣﺜﺎﻝ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫ﺗﺎﺑﻊ ﻧﺴﺒﺖ ﺑﻪ ﺗﻌﺪﺍﺩﯼ ﻧﻤﻮﻧﻪ(‬ ‫ﻣﺜﺎﻝ ‪)۱−۸‬ﻣﺤﺎﺳﺒﻪ ﮐﻤﺘﺮﯾﻦ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎﯼ ﯾ‬

‫ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ‪ (xi , yi ), i = 1, . . . , 4‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪:‬‬

‫)‪(−2, −1.5), (0, 0.1), (1, −0.6), (2, −1.0), (4, −2.0‬‬
‫ﻫﻤﭽﻨﻴﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺳﻪ‬
‫‪y = 0.08x3 − 0.2x2 − 0.6x − 0.4‬‬
‫ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪ .‬ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎﻱ ﺍﻳﻦ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺭﺍ ﻧﺴﺒﺖ ﺑﻪ ﻧﻘﺎﻁ ﺩﺍﺩﻩ ﺷﺪﻩ ﺣﺴﺎﺏ ﻛﻨﻴﻢ‪.‬‬
‫ﺍﮔﺮ ﻗﺮﺍﺭ ﺩﻫﻴﻢ‬
‫)‪ei := yi − (0.08x3i − 0.2x2i − 0.6xi − 0.4‬‬
‫ﺁﻥﮔﺎﻩ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪e1 = −0.86, e2 = 0.50, e3 = 0.52, e4 = 0.76, e5 = −1.12.‬‬
‫ﻭ ﻟﺬﺍ ﻣﻘﺪﺍﺭ ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎ ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ‬
‫= ‪SSR‬‬ ‫‪e21‬‬ ‫‪+‬‬ ‫‪e22‬‬ ‫‪+ ··· +‬‬ ‫‪e27‬‬ ‫‪= 3.092.‬‬

‫‪5 Model‬‬ ‫‪7 Residual‬‬ ‫)‪Sum of Squares (RSS‬‬

‫‪6 Sum‬‬ ‫)‪of Squared Residual (SSR‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۱۹۹‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫‪0‬‬
‫‪e2 = 0.50‬‬

‫‪e3 = 0.52‬‬
‫‪−1‬‬ ‫‪e1 = −0.86‬‬
‫‪e4 = 0.76‬‬ ‫‪e5 = −1.12‬‬

‫‪−2‬‬

‫‪SSR = (−0.86)2 + (0.50)2 + (0.52)2 + (0.76)2 + (−1.12)2 = 3.092‬‬

‫‪−2‬‬ ‫‪−1‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺳﻪ ﺩﺍﺩﻩ ﺷﺪﻩ‪ ،‬ﻣﺠﻤﻮﻉ ﻣﺮﺑﻌﺎﺕ ﺧﻄﺎ ﺑﺮﺍﺑﺮ ‪ 3.092‬ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‬
‫ﻧﻘﺎﻁ ﻓﻮﻕ ﺑﺎ ﻳﻚ ﺟﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺳﻪ ﺑﻪ ﺩﻧﺒﺎﻝ ﺁﻥ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ ﺳﻪ ﻫﺴﺘﻴﻢ ﻛﻪ ﻣﻘﺪﺍﺭ ‪ SSR‬ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺁﻥ ﻛﻤﺘﺮﻳﻦ‬
‫ﻣﻘﺪﺍﺭ ﻣﻤﻜﻦ ﺑﺎﺷﺪ‪.‬‬

‫ﻣﻄﺎﻟﺐ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺩﺭ ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺟﻤﻊﺑﻨﺪﻱ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‬ ‫ﺗﻌﺮﯾﻒ ‪۱−۸‬‬
‫ﻣﺴﺌﻠﻪ ﺑﺮﺍﺯﺵ ﻣﻨﺤﻨﻲ‪ ،‬ﻣﺴﺎﻟﻪ ﺍﻱ ﺍﺳﺖ ﻛﻪ ﺩﺍﺭﺍﻱ ﺩﻭ ﻭﺭﻭﺩﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬

‫• ﺑﻪ ﺗﻌﺪﺍﺩ ‪ n‬ﻧﻤﻮﻧﻪ ﺑﻪ ﺻﻮﺭﺕ‪(x1 , y1 ), (x2 , y2 ), . . . , (xn , yn ) :‬‬

‫• ﻳﻚ ﻣﺪﻝ ﺭﻳﺎﺿﻲ ﺑﺎ ‪ m‬ﭘﺎﺭﺍﻣﺘﺮ ﻣﺠﻬﻮﻝ ‪ c1 , . . . , cm‬ﺑﻪ ﺻﻮﺭﺕ ) ‪ y = ϕ(x; c1 , . . . , cm‬ﻛﻪ ﺍﺭﺗﺒﺎﻁ ﺑﻴﻦ‬
‫ﻣﻮﻟﻔﻪﻫﺎﻱ ﻧﻤﻮﻧﻪﻫﺎ ﺭﺍ ﻣﺸﺨﺺ ﻣﻲﻛﻨﺪ‬

‫ﻫﺪﻑ ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﻳﺎﻓﺘﻦ ﻣﻘﺎﺩﻳﺮﻱ ﺑﺮﺍﻱ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ‪ c1 , . . . , cm‬ﺍﺳﺖ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻣﻘﺪﺍﺭ ﺯﻳﺮ ﻣﻴﻨﻴﻤﻢ ﮔﺮﺩﺩ‬
‫‪X‬‬
‫‪n‬‬
‫‬ ‫‪2‬‬
‫=‪SSR :‬‬ ‫‪ϕ(xi ; c1 , . . . , cm ) − yi‬‬
‫‪i=1‬‬

‫ﺩﺭ ﺧﺼﻮﺹ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﻣﻄﺎﻟﺐ ﺯﻳﺮ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺷﻤﺮﺩ‪.‬‬

‫‪ .۱‬ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﻫﻴﭻ ﺷﺮﻃﻲ ﺭﻭﻱ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ‪ .‬ﻻﺯﻡ ﻧﻴﺴﺴﺖ ﻛﻪ ‪xi‬ﻫﺎ ﺩﺭ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪﺍﻱ ﻣﺮﺗﺐ ﺑﺎﺷﻨﺪ‪.‬‬
‫ﺣﺘﻲ ﻻﺯﻡ ﻧﻴﺴﺖ ﻛﻪ ‪xi‬ﻫﺎ ﻣﺠﺰﺍ ﺑﺎﺷﻨﺪ‪.‬‬

‫‪ .۲‬ﺩﺭ ﻣﺴﺌﻠﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﺗﻌﺪﺍﺩ ﭘﺎﺭﺍﻣﺘﺮﻫﺎ )‪ (m‬ﺍﺯ ﺗﻌﺪﺍﺩ ﻧﻤﻮﻧﻪﻫﺎ )‪ (n‬ﻛﻤﺘﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﻳﻌﻨﻲ ‪.m ≤ n‬‬

‫‪ .۳‬ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ‪ ،m = n‬ﺁﻥﮔﺎﻩ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺑﻪ ﻣﺴﺎﻟﻪ ﺩﺭﻭﻥﻳﺎﺑﻲ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪ .۴‬ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﻳﻚ ﻣﺴﺎﻟﻪ ﺑﻬﻴﻨﻪﺳﺎﺯﻱ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﻪ ﻣﻴﻨﻴﻤﻢ ﻛﺮﺩﻥ ﻳﻚ ﺗﺎﺑﻊ ﭼﻨﺪﻣﺘﻐﻴﺮﻩ ﻣﻲﭘﺮﺩﺍﺯﺩ‪.‬‬

‫ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ‪ ،‬ﺑﺮﺧﻲ ﺍﺯ ﻣﺪﻝﻫﺎﻱ ﺭﺍﻳﺞ ﺩﺭ ﺑﺮﺍﺯﺵ ﺩﺍﺩﻩﻫﺎ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰۰‬‬

‫ﺿﺎﺑﻄﻪ ﻣﺪﻝ‬ ‫ﻧﺎﻡ ﺑﺮﺍﺯﺵ‬ ‫‪#‬‬

‫‪ϕ(x; a, b) := ax + b‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ‬ ‫‪۱‬‬
‫‪2‬‬
‫‪ϕ(x; a, b, c) := ax + bx + c‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺳﻬﻤﻲ‬ ‫‪۲‬‬
‫‪ϕ(x; a0 , . . . , an ) := an xn + a1 x + · · · + an xn‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬ ‫‪۳‬‬
‫‪ϕ(x; a, b) := abx‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﻧﻤﺎﻳﻲ ‪۸‬‬ ‫‪۴‬‬
‫‪ϕ(x; a, b) := axb‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﺗﻮﺍﻧﻲ ‪۹‬‬ ‫‪۶‬‬
‫‪ϕ(x; a, b, c) := abx + c‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﻧﻤﺎﻳﻲ ﺗﻌﻤﻴﻢ ﻳﺎﻓﺘﻪ ‪۱۰‬‬ ‫‪۷‬‬
‫‪ϕ(x; a, b, c) := axb + c‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﺗﻮﺍﻧﻲ ﺗﻌﻤﻴﻢ ﻳﺎﻓﺘﻪ ‪۱۱‬‬ ‫‪۸‬‬
‫‪ϕ(x; a, b) := aebx‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﻧﻤﺎﻳﻲ ﻃﺒﻴﻌﻲ ‪۱۲‬‬ ‫‪۹‬‬
‫=‪ϕ(x; a, b, c) :‬‬ ‫‪1‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﻣﻨﺤﻨﻲ ﻟﺠﺴﺘﻴﻚ ‪۱۳‬‬ ‫‪۱۰‬‬
‫‪abx +c‬‬
‫=‪ϕ(x; a, b) :‬‬ ‫‪1‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﻣﻨﺤﻨﻲ ﻣﻌﻜﻮﺳﻪ ‪۱۴‬‬ ‫‪۱۱‬‬
‫‪ax+b‬‬
‫=‪ϕ(x; a, b) :‬‬ ‫‪ax‬‬ ‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﻧﺮﺥ ﺭﺷﺪ ﺍﺷﺒﺎﻉ ‪۱۵‬‬ ‫‪۱۲‬‬
‫‪b+x‬‬

‫ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ‬ ‫�‬

‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ‪ ،‬ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﻪ ﺻﻮﺭﺕ ‪ y = ϕ(x; a, b) = ax + b‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﮔﺮ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪ (xi , yi ), i = 1, . . . , n‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪X‬‬
‫‪n‬‬
‫‬ ‫‪2‬‬ ‫‪X‬‬
‫‪n‬‬
‫‬ ‫‪2‬‬
‫= ‪SSR‬‬ ‫‪ϕ(xi ; a, b) − yi‬‬ ‫=‬ ‫‪axi + b − yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬
‫ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺗﺎﺑﻌﻲ ﺍﺯ ‪ b‬ﻭ ‪ a‬ﺍﺳﺖ ﻭ ﺑﺎﻳﺪ ‪ b‬ﻭ ‪ a‬ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺑﺎﺷﺪ ﻛﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﺭﺍ ﻣﻴﻨﻴﻤﻢ ﻛﻨﻨﺪ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻣﺸﺘﻘﺎﺕ‬
‫ﻧﺴﺒﻲ ﺭﺍ ﻣﺴﺎﻭﻱ ﺻﻔﺮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‬
‫‪‬‬
‫ ‪∂SSR X‬‬ ‫‬
‫‪n‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫=‬ ‫‪2 axi + b − yi = 0‬‬
‫‪ ∂b‬‬
‫‪i=1‬‬
‫‪‬‬
‫‪‬‬ ‫‪∂SSR X‬‬
‫‪n‬‬
‫‬ ‫‬
‫‪‬‬
‫‪‬‬ ‫‪2xi axi + b − yi = 0‬‬
‫‪ ∂a‬‬ ‫=‬
‫‪i=1‬‬
‫ﺑﻌﺪ ﺍﺯ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺟﺒﺮﻱ ﻣﻌﺎﺩﻻﺕ ﻓﻮﻕ ﺩﺍﺭﻳﻢ‪:‬‬
‫! ‪ n‬‬ ‫!‬
‫‪‬‬ ‫‪X‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬
‫‪‬‬
‫‪‬‬ ‫‪1‬‬ ‫‪b‬‬ ‫‪+‬‬ ‫‪x‬‬ ‫‪a‬‬ ‫=‬ ‫‪yi‬‬
‫‪‬‬
‫‪‬‬ ‫‪i‬‬
‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬
‫!‬ ‫!‬
‫‪‬‬
‫‪‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬ ‫‪n‬‬
‫‪‬‬
‫‪‬‬ ‫‪2‬‬
‫‪‬‬ ‫‪xi b +‬‬ ‫= ‪xi a‬‬ ‫‪x i yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺩﻭ ﻣﻌﺎﺩﻟﻪ ﺧﻄﻲ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﻪ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺯﻳﺮ ﻗﺎﺑﻞ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻣﻲﺑﺎﺷﺪ‬
‫"‬ ‫‪Pn‬‬ ‫‪#" #‬‬ ‫‪" Pn‬‬ ‫‪#‬‬
‫‪n‬‬ ‫‪i=1 xi‬‬ ‫‪b‬‬ ‫‪i=1 yi‬‬
‫‪Pn‬‬ ‫‪Pn‬‬ ‫‪2‬‬
‫=‬ ‫‪Pn‬‬
‫‪i=1‬‬ ‫‪xi‬‬ ‫‪i=1 xi‬‬ ‫‪a‬‬ ‫‪i=1 xi yi‬‬
‫ﺑﺎ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ ،‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ‪ b‬ﻭ ‪ a‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ ﻭ ﺿﺎﺑﻄﻪ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻧﻘﺎﻁ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۰۱‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫ﺧﻂ (‬ ‫ﻣﺜﺎﻝ ‪)۲−۸‬ﺑﺮﺍﺯﺵ ﻧﻘﺎﻁ ﺑﺎ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻁ ﺯﻳﺮ ﺭﺍ ﺑﺎ ﻳﻚ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‬
‫)‪(1, 0.5), (2, 2.5), (3, 2), (4, 4), (5, 3.5), (6, 6), (7, 5.5‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺍﮔﺮ ‪ y = ax + b‬ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻧﻘﺎﻁ ﻓﻮﻕ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ b‬ﻭ ‪ a‬ﺍﺯ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﻧﺪ‬
‫‪" P7‬‬ ‫‪P7‬‬ ‫‪#" #‬‬ ‫‪" P7‬‬ ‫‪#‬‬
‫‪i=1 1‬‬ ‫‪i=1 xi‬‬ ‫‪b‬‬ ‫‪i=1 yi‬‬
‫‪P7‬‬ ‫‪P7‬‬ ‫‪= P7‬‬
‫‪i=1‬‬ ‫‪xi‬‬ ‫‪i=1‬‬ ‫‪x2i‬‬ ‫‪a‬‬ ‫‪i=1‬‬ ‫‪x i yi‬‬

‫‪i‬‬ ‫‪xi‬‬ ‫‪yi‬‬ ‫‪x i yi‬‬ ‫‪x2i‬‬ ‫ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻣﻮﻟﻔﻪﻫﺎﻱ ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﻭ ﺑﺮﺩﺍﺭ ﺳﻤﺖ ﺭﺍﺳﺖ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪،‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪0.5‬‬ ‫‪0.5‬‬ ‫‪1‬‬ ‫ﺟﺪﻭﻝ ﺭﻭﺑﺮﻭ ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪ .‬ﻣﻘﺎﺩﻳﺮ ﻣﻮﺭﺩ ﻧﻴﺎﺯ ﺑﺮﺍﻱ ﺗﺸﻜﻴﻞ ﺩﺳﺘﮕﺎﻩ ﺩﺭ‬
‫‪2‬‬ ‫‪2‬‬ ‫‪2.5‬‬ ‫‪5.0‬‬ ‫‪4‬‬ ‫ﺳﻄﺮ ﺁﺧﺮ ﺟﺪﻭﻝ ﻗﺮﺍﺭ ﺩﺍﺭﺩ‪ .‬ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‬
‫‪3‬‬ ‫‪3‬‬ ‫‪2.0‬‬ ‫‪6.0‬‬ ‫‪9‬‬ ‫"‬ ‫‪#" #‬‬ ‫"‬ ‫‪#‬‬
‫‪7‬‬ ‫‪28‬‬ ‫‪b‬‬ ‫‪24‬‬
‫‪4‬‬ ‫‪4‬‬ ‫‪4.0‬‬ ‫‪16‬‬ ‫‪16‬‬ ‫=‬
‫‪5‬‬ ‫‪5‬‬ ‫‪3.5‬‬ ‫‪17.5‬‬ ‫‪25‬‬ ‫‪28‬‬ ‫‪140‬‬ ‫‪a‬‬ ‫‪119.5‬‬
‫‪6‬‬ ‫‪6‬‬ ‫‪6.0‬‬ ‫‪36.0‬‬ ‫‪36‬‬ ‫ﺑﺎ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ ،‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪7‬‬
‫‪P‬‬ ‫‪7‬‬ ‫‪5.5‬‬ ‫‪38.5‬‬ ‫‪49‬‬ ‫‪1‬‬ ‫‪47‬‬
‫‪28‬‬ ‫‪24‬‬ ‫‪119.5‬‬ ‫‪140‬‬ ‫=‪b‬‬ ‫‪,‬‬ ‫=‪a‬‬
‫‪14‬‬ ‫‪56‬‬
‫ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‬
‫‪1‬‬ ‫‪47‬‬
‫=‪y‬‬ ‫‪+ x‬‬
‫‪14 56‬‬

‫ﺑﺮﺍﺯﺵ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ‬ ‫�‬

‫ﮔﺎﻫﻲ ﻣﻮﺍﻗﻊ ﻣﺪﻝ ﻧﻘﺎﻁ ﺩﺍﺩﻩﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺧﻂ ﻣﺴﺘﻘﻴﻢ ﻧﻴﺴﺖ ﻭ ﺑﻠﻜﻪ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻓﺮﺽ ﻛﻨﻴﺪ‬
‫ﻛﻪ ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪ ﺑﻪ ﺻﻮﺭﺕ‬
‫‪(xi , yi ), i = 1, . . . , n‬‬
‫ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﻧﺪ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺑﺎ ﻳﻚ ﺳﻬﻤﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﻣﺪﻝ ﺑﺮﺍﺯﺵ ﺑﻪ ﺻﻮﺭﺕ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﺟﻪ‬
‫ﺩﻭﻱ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫‪y = ϕ(x; c0 , c1 , c2 ) = c0 + c1 x + c2 x2 .‬‬
‫ﺣﺎﻝ ‪ SSR‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫‪X‬‬
‫‪n‬‬
‫‬ ‫‬
‫‪2 2‬‬
‫= ‪SSR‬‬ ‫‪yi − c 0 − c 1 x i − c 2 x i‬‬
‫‪i=1‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰۲‬‬

‫ﺑﺮﺍﻱ ﻣﻲﻧﻴﻤﻢ ﻛﺮﺩﻥ ﻋﺒﺎﺭﺕ ﻓﻮﻕ‪ ،‬ﻣﺸﺘﻘﺎﺕ ﻧﺴﺒﻲ ﺁﻥ ﻧﺴﺒﺖ ﺑﻪ ‪ c1 ،c0‬ﻭ ‪ c2‬ﺭﺍ ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‬
‫‪∂SSR‬‬ ‫‪X‬‬‫‪n‬‬
‫‬ ‫‬
‫‪= −2‬‬ ‫‪yi − c0 − c1 xi − c2 x2i = 0‬‬
‫‪∂c0‬‬ ‫‪i=1‬‬

‫‪∂SSR‬‬ ‫‪X‬‬‫‪n‬‬
‫‬ ‫‬
‫‪= −2‬‬ ‫‪xi yi − c0 − c1 xi − c2 x2i = 0‬‬
‫‪∂c1‬‬ ‫‪i=1‬‬

‫‪∂SSR‬‬ ‫‪X‬‬‫‪n‬‬
‫‬ ‫‬
‫‪= −2‬‬ ‫‪x2i yi − c0 − c1 xi − c2 x2i = 0‬‬
‫‪∂c2‬‬ ‫‪i=1‬‬
‫ﻣﻌﺎﺩﻻﺕ ﻓﻮﻕ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻧﻤﻮﺩ‬
‫!‬ ‫!‬ ‫!‬
‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬
‫‪n‬‬
‫‪1 c0 +‬‬ ‫‪xi‬‬ ‫‪c1 +‬‬ ‫‪x2i‬‬ ‫= ‪c2‬‬ ‫‪yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬
‫!‬ ‫!‬ ‫!‬
‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬
‫‪xi‬‬ ‫‪c0 +‬‬ ‫‪x2i‬‬ ‫‪c1 +‬‬ ‫‪x3i‬‬ ‫= ‪c2‬‬ ‫‪x i yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬
‫!‬ ‫!‬ ‫!‬
‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬ ‫‪X‬‬‫‪n‬‬
‫‪x2i‬‬ ‫‪c0 +‬‬ ‫‪x3i‬‬ ‫‪c1 +‬‬ ‫‪x4i‬‬ ‫= ‪c2‬‬ ‫‪x2i yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬ ‫‪i=1‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺳﻪﺗﺎﻳﻲ ﺍﺳﺖ ﻛﻪ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﺁﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‬
‫‪‬‬ ‫‪Pn‬‬ ‫‪Pn‬‬ ‫‪   P‬‬ ‫‪‬‬
‫‪n‬‬
‫‪n‬‬ ‫‪i=1 xi‬‬ ‫‪x2i‬‬ ‫‪c0‬‬ ‫‪yi‬‬
‫‪ Pn‬‬ ‫‪Pn‬‬
‫‪i=1‬‬
‫‪Pn‬‬ ‫‪    Pn‬‬ ‫‪i+1‬‬ ‫‪‬‬
‫‪‬‬ ‫‪2‬‬
‫‪x3i ‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪P i=1 xi‬‬ ‫‪i=1 xi‬‬
‫‪Pn‬‬
‫‪i=1‬‬
‫‪Pn‬‬ ‫‪ c1  = P i+1 xi yi ‬‬
‫‪n‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪n‬‬ ‫‪2‬‬
‫‪i=1 xi‬‬ ‫‪i=1 xi‬‬ ‫‪i=1‬‬ ‫‪xi‬‬ ‫‪c2‬‬ ‫‪i+1 xi yi‬‬

‫ﺑﺎ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ ،‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۳−۸‬ﺑﺮﺍﺯﺵ ﺑﺎ ﺳﻬﻤ (‬

‫ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻁ‬
‫)‪(0, 2.1), (1, 7.7), (2, 13.6), (3, 27.2), (4, 40.9), (5, 61.1‬‬
‫ﺭﺍ ﺑﺎ ﻳﻚ ﺳﻬﻤﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪ .‬ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﻪ ﺻﻮﺭﺕ ‪ y = ϕ(x; c0 , c1 , c2 ) = c0 + c1 x + c2 x2‬ﺍﺳﺖ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ‬
‫ﻣﻨﻈﻮﺭ ﻣﺤﺎﺳﺒﺎﺕ ﺯﻳﺮ ﺭﺍ ﺍﻧﺠﺎﻡ ﻣﻲﺩﻫﻴﻢ‬
‫‪P‬‬ ‫‪P‬‬ ‫‪P‬‬ ‫‪P‬‬
‫‪xi = 15,‬‬ ‫‪x2i = 55,‬‬ ‫‪x3i = 225,‬‬ ‫‪x4i = 979,‬‬
‫‪P‬‬ ‫‪P‬‬ ‫‪P‬‬
‫‪yi = 152.6,‬‬ ‫‪xi yi = 585.6,‬‬ ‫‪x2i yi = 2488.8.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۰۳‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫‪61.1‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺍﻳﻦ ﺑﺮﺍﺯﺵ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺧﻮﺍﻫﺪ ﺑﻮﺩ‬
‫‪2‬‬
‫‪‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪2.47857 + 2.35929x + 1.86071x‬‬
‫‪6‬‬ ‫‪15‬‬ ‫‪55‬‬ ‫‪c0‬‬ ‫‪152.6‬‬
‫‪‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪15‬‬ ‫‪225‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪40.9‬‬ ‫‪‬‬ ‫‪55‬‬ ‫‪ c1  =  585.6 ‬‬
‫‪55‬‬ ‫‪225‬‬ ‫‪979‬‬ ‫‪c2‬‬ ‫‪2488.8‬‬
‫‪27.2‬‬ ‫ﺑﺎ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬
‫‪c0 = 2.47857, c1 = 2.35929, c2 = 1.86071‬‬
‫‪13.6‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺳﻬﻤﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪.‬‬
‫‪7.7‬‬
‫‪2.1‬‬ ‫‪y = 2.47857 + 2.35929x + 1.86071x2‬‬
‫‪1‬‬ ‫‪2‬‬ ‫‪3‬‬ ‫‪4‬‬ ‫‪5‬‬

‫∗ ﺑﺮﺍﺯﺵ ﻳﻚﮔﺎﻧﻪ ﻭ ﭼﻨﺪﮔﺎﻧﻪ‬ ‫�‬

‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺗﻌﺮﻳﻒ ﺷﺪﻩ‪ ،‬ﺗﺎﺑﻊ ﻣﺪﻝ ﻳﻚ ﺗﺎﺑﻊ ﻳﻚ ﻣﺘﻐﻴﺮﻩ ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺑﻮﺩ‪ .‬ﺗﺒﻌﺎ ﺩﺍﺩﻩﻫﺎ ﻧﻴﺰ ﺩﻭﺗﺎﻳﻲﻫﺎﻱ ﻣﺮﺗﺐ ﺑﻮﺩﻧﺪ‬
‫ﻭ ﺑﻪ ﻋﺒﺎﺭﺗﻲ ﺩﺍﺩﻩﻫﺎ ﻧﻘﺎﻃﻲ ﺩﺭ ﺻﻔﺤﻪ ﺑﻮﺩﻧﺪ‪ .‬ﺍﻣﺎ ﮔﺎﻫﻲ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻃﻲ ﺩﺭ ﻓﻈﺎﻱ ﺳﻪ ﻳﺎ ﭼﻨﺪﺑﻌﺪﻱ ﺭﺍ ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ‬
‫ﺣﺎﻟﺖ ﺩﺍﺩﻩﻫﺎ ﺳﻪﺗﺎﻳﻲ ﻭ ﻳﺎ ﭼﻨﺪﺗﺎﻳﻲﻫﺎﻱ ﻣﺮﺗﺐ ﻫﺴﺘﻨﺪ ﻭ ﻫﻤﭽﻨﻴﻦ ﺗﺎﺑﻊ ﻣﺪﻝ ﻳﻦ ﺗﺎﺑﻊ ﺩﻭ ﻳﺎ ﭼﻨﺪ ﻣﺘﻐﻴﺮﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺩﺭ ﭼﻨﻴﻦ‬
‫ﻣﻮﺍﺭﺩﻱ ﺍﺯ ﻭﺍﮊﻩ ﺑﺮﺍﺯﺵ ﭼﻨﺪﮔﺎﻧﻪ ﻭ ﻳﺎ ﺭﮔﺮﺳﻴﻮﻥ ﭼﻨﺪﮔﺎﻧﻪ ‪ ۱۶‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﮔﺮﺩﺩ‪.‬‬
‫ﺑﺮﺍﺯﺵ ﭼﻨﺪﮔﺎﻧﻪ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۸‬‬
‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﺍﮔﺮ ﺗﺎﺑﻊ ﻣﺪﻝ ﻳﻚ ﺗﺎﺑﻊ ﺩﻭ ﻳﺎ ﭼﻨﺪ ﻣﺘﻐﻴﺮﻩ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺮﺍﺯﺵ ﺭﺍ ﺑﺮﺍﺯﺵ ﭼﻨﺪﮔﺎﻧﻪ ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺴﺎﻳﻞ‬
‫ﺩﺍﺩﻩﻫﺎ ﺳﻪﺗﺎﻳﻲ ﻭ ﻳﺎ ﭼﻨﺪﺗﺎﻳﻲﻫﺎﻱ ﻣﺮﺗﺐ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺩﺭ ﻳﻚ ﺑﺮﺍﺯﺵ ‪-p‬ﮔﺎﻧﻪ‪ ،‬ﺩﺍﺩﻩﻫﺎ )‪-(p + 1‬ﺗﺎﻳﻲﻫﺎﻱ ﻣﺮﺗﺐ ﺑﻪ ﺻﻮﺭﺕ‬
‫ﺯﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‬
‫
‬
‫‪x1i , . . . , xpi , yi , i = 1, . . . , m‬‬
‫ﻭ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‬
‫
‬
‫‪y = ϕ x1 , x 2 , . . . , x p ; c 1 , . . . , c m‬‬

‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻟﻲ ﺍﺯ ﺑﺮﺍﺯﺵ ﭼﻨﺪﮔﺎﻧﻪ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺑﺮﺍﺯﺵ ﭼﻨﺪ ﻧﻘﻄﻪ ﺩﺭ ﻓﻀﺎﻱ ﺳﻪ ﺑﻌﺪﻱ ﺑﺎ ﻳﻚ ﺻﻔﺤﻪ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪.‬‬

‫ﺻﻔﺤﻪ(‬ ‫ﻣﺜﺎﻝ ‪)۴−۸‬ﺑﺮﺍﺯﺵ ﻧﻘﺎﻁ ﺩﺭ ﻓﻀﺎ ﺑﺎ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻧﻘﺎﻁ‬
‫
‬
‫‪x1i , x2i , yi , i = 1, . . . , n‬‬
‫ﺩﺭ ﻓﻀﺎ ﺩﺍﺩﻩ ﺷﺪﻩﺍﻧﺪ ﻭ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺑﺎ ﻳﻚ ﺻﻔﺤﻪ ﺑﻪ ﺻﻮﺭﺕ‬
‫
‬
‫‪y = ϕ x1 , x 2 , c 0 , c 1 , c 2 = c 0 + c 1 x1 + c 2 x2‬‬
‫ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪.‬‬
‫‪16 multiple‬‬ ‫‪regression‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰۴‬‬

‫ﻣﺸﺎﺑﻪ ﺑﺎ ﺑﺮﺍﺯﺵ ﻳﻚ ﻣﺘﻐﻴﺮﻩ‪ SSR ،‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﻣﻲﻛﻨﻴﻢ‬

‫‪X‬‬
‫‪n‬‬
‫‬ ‫
‬ ‫‪2 X‬‬
‫‪n‬‬
‫‬ ‫‪2‬‬
‫= ‪SSR‬‬ ‫= ‪ϕ x1i , x2i , c0 , c1 , c2 − yi‬‬ ‫‪c0 + c1 x1i + c2 x2i − yi‬‬
‫‪i=1‬‬ ‫‪i=1‬‬
‫ﺍﻛﻨﻮﻥ ﺩﺍﺭﻳﻢ‪:‬‬
‫ ‪∂SSR X‬‬ ‫‬
‫‪n‬‬
‫=‬ ‫‪2 c0 + c1 x1i + c2 x2i − yi = 0,‬‬
‫‪∂c0‬‬ ‫‪i=1‬‬

‫‪∂SSR X‬‬ ‫‬ ‫‬

‫‪n‬‬
‫=‬ ‫‪2x1i c0 + c1 x1i + c2 x2i − yi = 0,‬‬
‫‪∂c0‬‬ ‫‪i=1‬‬

‫‪∂SSR X‬‬ ‫‬ ‫‬

‫‪n‬‬
‫=‬ ‫‪2x2,i c0 + c1 x1i + c2 x2i − yi = 0,‬‬
‫‪∂c0‬‬ ‫‪i=1‬‬

‫ﺑﻌﺪ ﺍﺯ ﺳﺎﺩﻩﺳﺎﺯﻱﻫﺎﻱ ﺟﺒﺮﻱ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﻦ ﺑﻪ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺯﻳﺮ ﺧﻮﺍﻫﻴﻢ ﺭﺳﻴﺪ‬
‫‪‬‬ ‫‪Pn‬‬ ‫‪Pn‬‬ ‫‪   P‬‬ ‫‪‬‬
‫‪n‬‬
‫‪n‬‬ ‫‪i=1 x1i‬‬ ‫‪i=1 x2i‬‬ ‫‪c0‬‬ ‫‪y‬‬
‫‪Pn‬‬ ‫‪Pn‬‬ ‫‪Pn‬‬ ‫‪   Pn i+1 i ‬‬
‫‪‬‬ ‫‪2‬‬ ‫‪ c1  = ‬‬ ‫‪‬‬
‫‪Pi=1 x1i‬‬ ‫‪Pn‬‬
‫) ‪i=1 (x1i‬‬ ‫‪i=1‬‬
‫‪Pn‬‬
‫‪x‬‬ ‫‪1i‬‬ ‫‪x‬‬ ‫‪2i‬‬ ‫‪   Pi+1 x1i yi ‬‬
‫‪n‬‬ ‫‪2‬‬ ‫‪n‬‬
‫‪i=1 x2i‬‬ ‫‪i=1‬‬ ‫‪x1i x2i‬‬ ‫) ‪i=1 (x2i‬‬ ‫‪c2‬‬ ‫‪i+1 x2i yi‬‬

‫ﺑﺎ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ ﻭ ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ﺻﻔﺤﻪ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻣﺸﺨﺺ‬
‫ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺧﻂ(‬ ‫ﻣﺜﺎﻝ ‪)۵−۸‬ﺑﺮﺍﺯﺵ ﺷﺶ ﻧﻘﻄﻪ ﺩﺭ ﻓﻀﺎ ﺑﺎ ﯾ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻁ ﺯﻳﺮ ﺭﺍ ﺑﺎ ﻳﻚ ﺻﻔﺤﻪ ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‬
‫)‪(0, 0, 5), (2, 1, 10), (2.5, 2, 9), (1, 3, 0), (4, 6, 3), (7, 2, 27‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﺜﺎﻝ ﻗﺒﻞ‪ ،‬ﺑﺮﺍﻱ ﻓﺮﺍﻫﻢ ﻛﺮﺩﻥ ﻣﻮﻟﻔﻪﻫﺎﻱ ﺩﺳﺘﮕﺎﻩ ﻣﺘﻨﺎﻇﺮ‪ ،‬ﺟﺪﻭﻟﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‬

‫‪i‬‬ ‫‪x1i‬‬ ‫‪x2i‬‬ ‫‪yi‬‬ ‫‪x21i‬‬ ‫‪x22i‬‬ ‫‪x1i x2i‬‬ ‫‪x1i yi‬‬ ‫‪x2i yi‬‬
‫‪1‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪5‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬ ‫‪0‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪10‬‬ ‫‪4‬‬ ‫‪1‬‬ ‫‪2‬‬ ‫‪20‬‬ ‫‪10‬‬
‫‪3‬‬ ‫‪2.5‬‬ ‫‪2‬‬ ‫‪9‬‬ ‫‪6.25‬‬ ‫‪4‬‬ ‫‪5‬‬ ‫‪22.5‬‬ ‫‪18‬‬
‫‪4‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪0‬‬ ‫‪1‬‬ ‫‪9‬‬ ‫‪3‬‬ ‫‪0‬‬ ‫‪0‬‬
‫‪5‬‬ ‫‪4‬‬ ‫‪6‬‬ ‫‪3‬‬ ‫‪16‬‬ ‫‪36‬‬ ‫‪24‬‬ ‫‪12‬‬ ‫‪18‬‬
‫‪6‬‬ ‫‪7‬‬ ‫‪2‬‬ ‫‪27‬‬ ‫‪49‬‬ ‫‪4‬‬ ‫‪14‬‬ ‫‪189‬‬ ‫‪56‬‬
‫‪P‬‬
‫‪54‬‬ ‫‪16.5‬‬ ‫‪14‬‬ ‫‪76.25‬‬ ‫‪54‬‬ ‫‪48‬‬ ‫‪243.5‬‬ ‫‪100‬‬

‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺟﺪﻭﻝ ﻓﻮﻕ‪ ،‬ﺩﺳﺘﮕﺎﻩ ﺯﻳﺮ ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‬

‫‪‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪6‬‬ ‫‪16.5 14‬‬ ‫‪c0‬‬ ‫‪54‬‬
‫‪‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪16.5 76.25 48 c1  = 243.5‬‬
‫‪‬‬ ‫‪  ‬‬ ‫‪‬‬
‫‪14‬‬ ‫‪48‬‬ ‫‪54‬‬ ‫‪c2‬‬ ‫‪100‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۰۵‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫ﺑﺎ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬

‫‪c0 = 5, c1 = 4, c2 = −3.‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺻﻔﺤﻪ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪y = 5 + 4x1 − 3x3 .‬‬

‫ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻭ ﻏﻴﺮﺧﻄﻲ‬ ‫‪۳.۸‬‬

‫ﻣﺴﺎﻳﻞ ﺑﺮﺍﺯﺵ ﺍﺯ ﻳﻚ ﺩﻳﺪﮔﺎﻩ ﺑﻪ ﻣﺴﺎﻳﻞ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻭ ﻏﻴﺮﺧﻄﻲ ﺗﻘﺴﻴﻢﺑﻨﺪﻱ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫ﺑﺮﺍﺯﺵ ﺧﻄ ﻭ ﻏﯿﺮﺧﻄ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۸‬‬

‫ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﺍﮔﺮ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﺮﺣﺴﺐ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﻳﻚ ﺗﺎﺑﻊ ﺧﻄﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺴﺎﻟﻪ ﺭﺍ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‬
‫ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‪ ،‬ﻓﺮﻡ ﻛﻠﻲ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫)‪y = ϕ(x; c1 , . . . , cm ) = c1 ψ1 (x) + c2 ψ2 (x) + · · · + cm ψm (x) + ψ0 (x‬‬
‫ﺩﺭﺻﻮﺭﺗﻲ ﻛﻪ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﺮﺣﺴﺐ ﺣﺪﺍﻗﻞ ﻳﻜﻲ ﺍﺯ ﭘﺎﺭﺍﻣﺘﺮﻫﺎ ﻏﻴﺮﺧﻄﻲ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﺮﺍﺯﺵ ﺭﺍ ﻏﻴﺮﺧﻄﻲ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬

‫ﺗﻤﺎﻡ ﻧﻤﻮﻧﻪﻫﺎﻱ ﺑﺮﺍﺯﺵ ﻛﻪ ﺗﺎﻛﻨﻮﻥ ﺍﺭﺍﻳﻪ ﺷﺪ)ﺑﺮﺍﺯﺵ ﺑﺎ ﺧﻂ‪ ،‬ﺑﺮﺍﺯﺵ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎ‪ ،‬ﺑﺮﺍﺯﺵ ﺑﺎ ﺻﻔﺤﻪ( ﺍﺯ ﻧﻮﻉ ﺧﻄﻲ‬
‫ﺑﻮﺩﻧﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﮔﺮﺩﻳﺪ‪ ،‬ﺣﻞ ﻣﺴﺎﻳﻞ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﻣﻨﺘﻬﻲ ﻣﻲﮔﺮﺩﺩ‪ .‬ﺍﻣﺎ ﺩﺭ ﻣﺴﺎﻳﻞ ﺑﺮﺍﺯﺵ‬
‫ﻏﻴﺮﺧﻄﻲ ﺩﺭ ﺁﺧﺮ ﺑﺎ ﻳﻚ ﻣﺴﺎﻟﻪ ﻏﻴﺮﺧﻄﻲ ﺳﺮﻭﻛﺎﺭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪ .‬ﺍﺯ ﺍﻳﻦ ﺭﻭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﻏﻴﺮﺧﻄﻲ ﭘﻴﭽﻴﺪﻩﺗﺮ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻟﺒﺘﻪ‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺍﺷﺎﺭﻩ ﺧﻮﺍﻫﻴﻢ ﻛﺮﺩ‪ ،‬ﺩﺭ ﻣﻮﺍﺭﺩﻱ ﺧﺎﺹ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﺯﺵ ﻏﻴﺮﺧﻄﻲ ﺭﺍ ﺑﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻧﺒﺪﻳﻞ ﻧﻤﻮﺩ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۶−۸‬ﺑﺮﺍﺯﺵ ﻏﯿﺮﺧﻄ (‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻁ‬
‫)‪(0.25, 0.28), (0.75, 0.57), (1.25, 0.68), (1.75, 0.74), (2.25, 0.79‬‬
‫ﺭﺍ ﺑﺎ ﺗﺎﺑﻊ ﻣﺪﻝ‬
‫‪−c1 x‬‬
‫
‬
‫‪y = ϕ(x; c0 , c1 ) = c0 1 − e‬‬
‫ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﺮﺣﺴﺐ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻳﺶ ﻏﻴﺮﺧﻄﻲ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﺑﺮﺍﺯﺵ ﺍﺯ ﻧﻮﻉ ﻏﻴﺮﺧﻄﻲ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺮﺍﻱ‬
‫ﺍﻧﺠﺎﻡ ﺑﺮﺍﺯﺵ‪ ،‬ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‬
‫‪X‬‬
‫‪5‬‬
‫‬ ‫
‬ ‫‪2‬‬
‫= ‪SSR‬‬ ‫‪c0 1 − e−c1 xi − yi‬‬
‫‪i=1‬‬
‫ﺍﻛﻨﻮﻥ ﺑﺎﻳﺪ ﻣﻘﺎﺩﻳﺮ ‪c0‬ﻭ ‪ c1‬ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻳﺎﻓﺘﻪ ﺷﻮﻧﺪ ﻛﻪ ﻋﺒﺎﺭﺕ ﻓﻮﻕ ﻣﻴﻨﻴﻤﻢ ﮔﺮﺩﺩ‪ .‬ﻣﻲﺗﻮﺍﻥ ﻣﺎﻧﻨﺪ ﻗﺒﻞ ﻣﺸﺘﻘﺎﺕ ﻧﺴﺒﻲ ﺭﺍ ﻣﺴﺎﻭﻱ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰۶‬‬

‫ﺻﻔﺮ ﻗﺮﺍﺭ ﺩﺍﺩ‬

‫‪∂SSR‬‬ ‫‪X‬‬
‫‪5‬‬
‫
‬ ‫
‬ ‫‬
‫=‬ ‫‪2 1 − e−c1 xi‬‬ ‫‪c0 1 − e−c1 xi − yi = 0‬‬
‫‪∂c0‬‬ ‫‪i=1‬‬

‫‪∂SSR‬‬ ‫‪X‬‬
‫‪5‬‬
‫‬ ‫
‬ ‫‬
‫=‬ ‫‪2c0 xi e−c1 xi c0 1 − e−c1 xi − yi = 0‬‬
‫‪∂c1‬‬ ‫‪i=1‬‬

‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﺩﻭ ﻣﻌﺎﺩﻟﻪ ﻭ ﺩﻭ ﻣﺠﻬﻮﻝ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﺍﻳﻦ ﻣﻌﺎﺩﻻﺕ ﺍﺯ ﻧﻮﻉ ﻏﻴﺮﺧﻄﻲﺍﻧﺪ ﻭ ﺗﺒﻌﺎ ﻧﻤﻲﺗﻮﺍﻥ ﺍﻳﻦ ﻣﻌﺎﺩﻻﺕ‬
‫ﺭﺍ ﺑﻪ ﻓﺮﻡ ﻣﺎﺗﺮﻳﺴﻲ ﻫﻢ ﻧﻮﺷﺖ‪ .‬ﺑﺎ ﺣﻞ ﺍﻳﻦ ﺩﺳﺘﮕﺎﻩ ﻏﻴﺮﺧﻄﻲ‪ ،‬ﻣﻘﺎﺩﻳﺮﻱ ﺑﺮﺍﻱ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺣﺎﺻﻞ ﻣﻲﮔﺮﺩﺩ‪.‬‬

‫ﺍﻟﺒﺘﻪ ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻣﻴﻨﻴﻤﻢ ﻛﻨﻨﺪﻩﻱ ﻳﻚ ﺗﺎﺑﻊ ﭼﻨﺪﻣﺘﻐﻴﺮﻩ‪ ،‬ﺑﻪ ﺟﺎﻱ ﻣﺴﺎﻭﻱ ﺻﻔﺮ ﻗﺮﺍﺭ ﺩﺍﺩﻥ ﻣﺸﺘﻘﺎﺕ ﻧﺴﺒﻲ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ‬
‫ﺭﻭﺵﻫﺎﻱ ﻛﺎﺭﺍﺗﺮ ﻭ ﻣﻨﺎﺳﺐﺗﺮ ﺩﻳﮕﺮﻱ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‬

‫ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺧﻲ ﺑﺮﺍﺯﺵﻫﺎﻱ ﻏﻴﺮﺧﻄﻲ‬ ‫�‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﺩﺭ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‪ ،‬ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﺮﺣﺴﺐ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺧﻄﻲ ﻣﻲﺑﺎﺷﺪ ﻭ ﻟﺬﺍ ﺩﺳﺘﮕﺎﻩ ﺣﺎﺻﻞ‬
‫ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺮﺍﺯﺵ ﻏﻴﺮﺧﻄﻲ ﺗﺎﺑﻊ ﻣﺪﻝ ﺑﺮﺣﺴﺐ ﭘﺎﺭﺍﻣﺘﺮﻫﺎ ﻏﻴﺮﺧﻄﻲ ﺑﺎﺷﺪ ﻭ ﺗﺒﻌﺎ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻏﻴﺮﺧﻄﻲ‬
‫ﺑﺮﻣﻲﺧﻮﺭﻳﻢ‪ .‬ﻭﺍﺿﺢ ﺍﺳﺖ ﻛﻪ ﺣﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻏﻴﺮﺧﻄﻲ ﺑﻪ ﻣﺮﺍﺗﺐ ﭘﻴﭽﻴﺪﻩﺗﺮ ﺍﺯ ﺣﻞ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﺧﻄﻲ ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ‬
‫ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ‬

‫» ﺁﻳﺎ ﻣﻲﺗﻮﺍﻥ ﺑﺎ ﺗﺮﻓﻨﺪﻫﺎﻳﻲ ﻣﺎﻧﻨﺪ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﻏﻴﺮﺧﻄﻲ ﺭﺍ ﺑﻪ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﺗﺒﺪﻳﻞ ﻛﺮﺩ؟ «‬

‫ﺩﺭ ﺣﺎﻟﺖ ﻛﻠﻲ‪ ،‬ﻳﺎﺳﺦ ﺳﻮﺍﻝ ﻣﻨﻔﻲ ﺍﺳﺖ‪ .‬ﺍﻣﺎ ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﺭﺩ ﺧﺎﺹ‪ ،‬ﻫﻤﺎﻧﻨﺪ ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧﻲ ﻭ ﻧﻤﺎﻳﻲ ﻛﻪ ﺍﺯ ﻧﻮﻉ ﻣﺴﺎﻳﻞ ﺑﺮﺍﺯﺵ‬
‫ﻏﻴﺮﺧﻄﻲ ﻫﺴﺘﻨﺪ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻣﺴﺎﻟﻪ ﺭﺍ ﺑﻪ ﻳﻚ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﻣﻌﺎﺩﻝ ﺗﺒﺪﻳﻞ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﻧﻤﻮﻧﻪﻫﺎﻳﻲ ﺍﺯ ﺧﻄﻲﺳﺎﺯﻱ ﺍﺭﺍﻳﻪ‬
‫ﻣﻲﺷﻮﺩ‪.‬‬

‫■ ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧﻲ‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺩﺍﺩﻩﻫﺎﻱ ‪ (xi , yi ), i = 1, . . . , n‬ﺭﺍ ﺑﺎ ﺗﺎﺑﻊ ﻣﺪﻝ ﻧﻤﺎﻳﻲ ‪ y = ϕ(x; a, b) = ax‬ﺑﺮﺍﺯﺵ‬
‫‪b‬‬

‫ﻛﻨﻴﻢ‪ .‬ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧﻲ ﺑﻪ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻏﻴﺮﺧﻄﻲ ﻣﻨﺘﻬﻲ ﻣﻲﮔﺮﺩﺩ‪ ،‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧﻲ ﺭﺍ ﺧﻄﻲﺳﺎﺯﻱ ﻧﻤﻮﺩ‪ .‬ﺑﺮﺍﻱ ﺧﻄﻲ‬
‫ﺳﺎﺯﻱ ﺍﻳﻦ ﻣﺴﺎﻟﻪ ﺑﺮﺍﺯﺵ‪ ،‬ﺗﺎﺑﻊ ﻟﮕﺎﺭﻳﺘﻢ ﺩﺭ ﻣﺒﻨﺎﻱ ﺩﻩ ﺭﺍ ﺭﻭﻱ ﺗﺎﺑﻊ ﻣﺪﻝ ﻧﻤﺎﻳﻲ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ ﻭ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‬
‫‪log y = log a + b log x‬‬
‫ﺑﺎ ﻓﺮﺽ ‪ ŷ = log y‬ﻭ ‪ x̂ = log x‬ﺗﺎﺑﻊ ﻣﺪﻝ ﻓﻮﻕ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻣﻲﮔﺮﺩﺩ‪:‬‬
‫̃‪ỹ = log a + bx‬‬
‫ﻫﻤﭽﻨﻴﻦ ﺑﺎ ﻓﺮﺽ ‪ ã = log a‬ﻭ ‪ b̃ = b‬ﻣﺪﻝ ﺩﺭ ﻧﻬﺎﻳﺖ ﺑﻪ ﻓﺮﻡ ﺧﻄﻲ ﺯﻳﺮ ﺗﺒﺪﻳﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫̃‪ỹ = ã + b̃x‬‬
‫ﺗﺎ ﺍﻳﻦﺟﺎ ﻣﺪﻝ ﺧﻄﻲ ﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻣﺪﻝ ﻧﺸﺎﻥ ﺩﻫﻨﺪﻩ ﺍﺭﺗﺒﺎﻁ ﺧﻄﻲ ﺑﻴﻦ ̃‪ x‬ﻭ ̃‪ y‬ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺑﺎﻳﺪ ﻧﻤﻮﻧﻪﻫﺎﻱ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ̃‪ x‬ﻭ ̃‪ y‬ﺭﺍ‬
‫ﺗﺸﻜﻴﻞ ﺩﻫﻴﻢ‪ .‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ x̂ = log x‬ﻭ ‪ ،ŷ = log y‬ﻟﺬﺍ ﻧﻤﻮﻧﻪﻫﺎﻱ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻣﺪﻝ ﺧﻄﻲ ﺷﺪﻩ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۰۷‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪x̂i := log(xi ),‬‬ ‫‪ŷi := log(yi ).‬‬
‫ﺍﻛﻨﻮﻥ ﻣﺪﻝ ﺧﻄﻲ ﺷﺪﻩ ̃‪ ỹ = ã + b̃x‬ﺭﺍ ﺑﺎ ﻧﻘﺎﻁ ‪ (x̂i , ŷi ), i = 1, . . . , n‬ﺑﺮﺍﺯﺵ ﻣﻲﺩﻫﻴﻢ ﻭ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ̂‪ a‬ﻭ ̂‪ b‬ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ‬
‫ﻣﻲﻛﻨﻴﻢ‪ .‬ﺣﺎﻝ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ ã = log a‬ﻭ ‪ b̃ = b‬ﻣﻲﺗﻮﺍﻥ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺪﻝ ﻧﻤﺎﻳﻲ ‪ y = axb‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ‪:‬‬
‫‪a = 10ã ,‬‬ ‫‪b = b̃.‬‬
‫ﺭﻭﺍﻝ ﺧﻄﻲﺳﺎﺯﻱ ﻓﻮﻕ ﺩﺭ ﺯﻳﺮ ﺧﻼﺻﻪ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪  log y = log a + b log x‬ﺍﺯ ﻃﺮﻓﻴﻦ ﻣﺪﻝ ‪ log‬ﻣﻲﮔﻴﺮﻳﻢ ‪y = axb‬‬ ‫̂‪ŷ = log a + bx‬‬
‫‪x̂:=log x, ŷ:=log y‬‬
‫→‪−−−−−−−−−−−−‬‬ ‫→‪−−−−−−−−−−−−−‬‬
‫) ‪(xi , yi‬‬ ‫) ‪(xi , yi‬‬ ‫) ̂‪x̂i :=log xi , ŷi :=log yi (x̂ , y‬‬
‫‪i i‬‬
‫‪‬‬ ‫‪‬‬
‫̂‪ŷ = â + b̂x‬‬ ‫ﺑﺎ ﺍﻧﺠﺎﻡ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‬
‫̂‪a = 10a‬‬
‫‪â:=log a, b̂=b‬‬
‫→‪−−−−−−−−‬‬ ‫→‪−−−−−−−−−−−−−−‬‬
‫) ‪(x̂i , ŷi‬‬ ‫ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ̂‪ a‬ﻭ ̂‪ b‬ﻣﺸﺨﺺ ﻣﻲﺷﻮﻧﺪ‬ ‫̂‪b = b‬‬

‫ﺩﺭ ﻣﺜﺎﻝ ﺯﻳﺮ ﺗﻜﻨﻴﻚ ﺧﻄﻲﺳﺎﺯﻱ ﻓﻮﻕ ﺑﺮﺍﻱ ﺍﻧﺠﺎﻡ ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧﻲ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﺩ‬

‫ﻣﺜﺎﻝ ‪)۷−۸‬ﺧﻄ ﺳﺎﺯﯼ ﺑﺮﺍﺯﺵ ﺗﻮﺍﻧ (‬

‫ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‬
‫‪(1, 0.5), (2, 1.7), (3, 3.4), (4, 5.7), (5, 8.4).‬‬
‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﺑﺎ ﻣﺪﻝ ﺗﻮﺍﻧﻲ ‪ y = axb‬ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪ .‬ﺩﺭ ﺯﻳﺮ ﻓﺮﺍﻳﻨﺪ ﺧﻄﻲﺳﺎﺯﻱ ﻭ ﻳﺎﻓﺘﻦ ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ﻣﺠﻬﻮﻝ ﺁﻭﺭﺩﻩ‬
‫ﺷﺪﻩ ﺍﺳﺖ‪:‬‬

‫‪xi‬‬ ‫‪yi‬‬ ‫‪x̂i‬‬ ‫‪ŷi‬‬

‫‪1‬‬ ‫‪0.5‬‬ ‫‪0.000‬‬ ‫‪-0.301‬‬ ‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪ŷ = â + b̂x̂,‬‬ ‫‪‬‬ ‫‪b‬‬
‫‪y = ax ,‬‬
‫‪2‬‬ ‫‪1.7‬‬ ‫‪0.301‬‬ ‫‪0.226‬‬ ‫‪‬‬ ‫‪‬‬
‫‪x̂i :=log xi‬‬ ‫ﺥﻁﯼ ﺏﺭﺍﺯﺵ‬ ‫̂‪a=10â , b=b‬‬
‫→‪−−−−−−‬‬ ‫→‪−−−−−−‬‬ ‫→‪â = 0.300, −−−−−−−‬‬ ‫‪a = 0.501,‬‬
‫‪3‬‬ ‫‪3.4‬‬ ‫‪ŷi :=log yi‬‬ ‫‪0.477‬‬ ‫‪0.534‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬
‫‪4‬‬ ‫‪5.7‬‬ ‫‪0.602‬‬ ‫‪0.753‬‬ ‫‪b̂ = 1.75‬‬ ‫‪b = 1.75‬‬

‫‪5‬‬ ‫‪8.4‬‬ ‫‪0.699‬‬ ‫‪0.922‬‬

‫ﺩﺭ ﻧﻬﺎﻳﺖ ﺗﺎﺑﻊ ﻧﻤﺎﻳﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻧﻘﺎﻁ ﺑﻪ ﺻﻮﺭﺕ ‪ y = 0.501x1.75‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪.‬‬
‫ﺑﺮﺍﻱ ﺩﺭﻙ ﺑﻬﺘﺮ‪ ،‬ﺩﺭ ﺷﻜﻞ ﺳﻤﺖ ﭼﭗ ﺯﻳﺮ‪ ،‬ﻧﻘﺎﻁ ﻧﻤﻮﻧﻪ ) ‪ (xi , yi‬ﻭ ﺗﺎﺑﻊ ﺗﻮﺍﻧﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﻭ ﺩﺭ ﺷﻜﻞ ﺳﻤﺖ ﺭﺍﺳﺖ ﺯﻳﺮ‪ ،‬ﻧﻘﺎﻁ‬
‫ﻧﻤﻮﻧﻪ ) ‪ (x̂i , ŷi‬ﻭ ﺗﺎﺑﻊ ﺧﻄﻲ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺭﺳﻢ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۰۸‬‬

‫■ ﺑﺮﺍﺯﺵ ﻧﻤﺎﻳﻲ‬

‫ﻓﺮﺽ ﻛﻨﻴﻢ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 1, . . . , n‬ﺭﺍ ﺑﺎ ﻣﺪﻝ ﻧﻤﺎﻳﻲ ‪ y = ϕ(x; a, b) = aebx‬ﺑﺮﺍﺯﺵ ﻛﻨﻴﻢ‪.‬‬
‫ﺑﺮﺍﻱ ﺧﻄﻲ ﺳﺎﺯﻱ‪ ،‬ﺍﺯ ﻟﮕﺎﺭﻳﺘﻢ ﻧﭙﺮﻳﻦ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﻴﻢ‪ .‬ﻓﺮﺍﻳﻨﺪ ﺧﻄﻲﺳﺎﺯﻱ ﺩﺭ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪  ln y = ln a + bx‬ﺍﺯ ﻃﺮﻓﻴﻦ ﻣﺪﻝ ‪ ln‬ﻣﻲﮔﻴﺮﻳﻢ ‪y = aebx‬‬ ‫̂‪ŷ = ln a + bx‬‬
‫‪x̂:=x, ŷ:=ln y‬‬
‫→‪−−−−−−−−−−−‬‬ ‫→‪−−−−−−−−−−−‬‬
‫) ‪(xi , yi‬‬ ‫) ‪(xi , yi‬‬ ‫) ̂‪x̂i :=xi , ŷi :=ln yi (x̂ , y‬‬
‫‪i i‬‬
‫‪‬‬ ‫‪‬‬
‫̂‪ŷ = â + b̂x‬‬ ‫ﺑﺎ ﺍﻧﺠﺎﻡ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‬
‫̂‪a = ea‬‬
‫‪â:=ln a, b̂=b‬‬
‫→‪−−−−−−−−‬‬ ‫→‪−−−−−−−−−−−−−−‬‬
‫) ‪(x̂i , ŷi‬‬ ‫ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ̂‪ a‬ﻭ ̂‪ b‬ﻣﺸﺨﺺ ﻣﻲﺷﻮﻧﺪ‬ ‫̂‪b = b‬‬

‫■ ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺍﺯﺵ ﺑﺎ ﺗﺎﺑﻊ ﻧﺮﺥ ﺭﺷﺪ ﺍﺷﺒﺎﻉ‬

‫ﺑﺮﺍﺯﺵ ﻧﻘﺎﻁ ‪ (xi , yi ), i = 1, . . . , n‬ﺑﺎ ﺗﺎﺑﻊ ﻣﺪﻝ ﻧﺮﺥ ﺭﺷﺪ ﺍﺷﺒﺎﻉ‬

‫‪ax‬‬
‫= )‪y = ϕ(x; a, b‬‬
‫‪b+x‬‬
‫ﻳﻚ ﺑﺮﺍﺯﺵ ﻏﻴﺮﺧﻄﻲ ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺍﻣﺎ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻳﻦ ﺑﺮﺍﺯﺵ ﺭﺍ ﺑﻪ ﻳﻚ ﺑﺮﺍﺯﺵ ﺧﻄﻲ ﺗﺒﺪﻳﻞ ﻧﻤﻮﺩ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ax‬‬ ‫‪‬‬ ‫‪1‬‬ ‫‪b+x‬‬ ‫‪b1‬‬ ‫‪1‬‬ ‫‪‬‬
‫= ‪y‬‬ ‫ﻃﺮﻓﻴﻦ ﻣﺪﻝ ﺭﺍ ﻋﻜﺲ ﻣﻲﻛﻨﻴﻢ‬
‫‪‬‬ ‫=‬ ‫=‬ ‫‪+‬‬ ‫‪1‬‬
‫‪x̂:= x , ŷ:= y‬‬‫‪1‬‬ ‫‪ŷ = b x̂ + 1‬‬
‫‪b+x −‬‬ ‫→‪−−−−−−−−−−−‬‬ ‫‪y‬‬ ‫‪ax‬‬ ‫‪a‬‬ ‫‪x‬‬ ‫‪a‬‬ ‫‪−−−−−‬‬ ‫→‪−−−−−‬‬ ‫‪a‬‬ ‫‪a‬‬
‫‪‬‬ ‫‪‬‬
‫‪‬‬ ‫‪x̂i := x1 , ŷi := y1 ‬‬‫‪‬‬
‫) ‪(xi , yi‬‬ ‫) ‪(xi , yi‬‬ ‫‪i‬‬ ‫‪i‬‬ ‫) ‪(x̂i , ŷi‬‬
‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪1‬‬
‫̂‪ŷ = â + b̂x‬‬ ‫‪‬‬
‫= ‪a‬‬
‫‪1‬‬
‫‪â:= a‬‬ ‫‪b‬‬
‫‪, b̂= a‬‬ ‫ﺑﺎ ﺍﻧﺠﺎﻡ ﺑﺮﺍﺯﺵ ﺧﻄﻲ‬ ‫̂‪a‬‬
‫→‪−−−−−−−‬‬ ‫→‪−−−−−−−−−−−−−−‬‬
‫) ‪(x̂i , ŷi‬‬ ‫‪ ‬ﭘﺎﺭﺍﻣﺘﺮﻫﺎﻱ ̂‪ a‬ﻭ ̂‪ b‬ﻣﺸﺨﺺ ﻣﻲﺷﻮﻧﺪ‬ ‫‪‬‬
‫̂‪b = b‬‬
‫̂‪a‬‬

‫■ ﺧﻄﻲﺳﺎﺯﻱ ﺑﺮﺧﻲ ﺑﺮﺍﺯﺵﻫﺎﻱ ﻏﻴﺮﺧﻄﻲ‬

‫ﺩﺭ ﺍﻧﺘﻬﺎ‪ ،‬ﺑﺮﺧﻲ ﺍﺯ ﺧﻄﻲ ﺳﺎﺯﻱﻫﺎﻱ ﺑﺮﺍﺯﺵ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۰۹‬‬ ‫ﺑﺮﺍﺯﺵ‬ ‫ﻓﺼﻞ ﻫﺸﺘﻢ‪:‬‬

‫∗ ﻧﻜﻮﻳﻲ ﺑﺮﺍﺯﺵ‬ ‫‪۴.۸‬‬

‫ﻫﻨﮕﺎﻣﻲ ﻛﻪ ﺑﺮﺍﺯﺵ ﻳﻚ ﻣﺠﻤﻮﻋﻪ ﺍﺯ ﻧﻘﺎﻁ ﺑﺎ ﻳﻚ ﺗﺎﺑﻊ ﻣﺪﻝ ﺍﻧﺠﺎﻡ ﭘﺬﻳﺮﻓﺖ‪ ،‬ﻳﻚ ﺳﻮﺍﻝ ﻣﻄﺮﺡ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺗﺎ ﭼﻪ ﺣﺪ ﺗﺎﺑﻊ ﻣﺪﻝ‬
‫ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﺷﺪﻩ ﺑﺎ ﻧﻘﺎﻁ ﺍﻧﻄﺒﺎﻕ ﺩﺍﺭﺩ‪ .‬ﺑﺮﺍﻱ ﭘﺎﺳﺦ ﺑﻪ ﺍﻳﻦ ﺳﻮﺍﻝ ﺷﺎﺧﺺﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺑﺮ ﻣﺒﻨﺎﻱ ﺁﻥ ﻣﻨﺎﺳﺐ‬
‫ﺑﻮﺩﻥ ﺑﺮﺍﺯﺵ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺍﻧﺪﺍﺯﻩﮔﻴﺮﻱ ﻛﺮﺩ‪ .‬ﺍﻭﻟﻴﻦ ﺷﺎﺧﺼﻲ ﻛﻪ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ ،‬ﺷﺎﺧﺺ ‪ SSR‬ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ‬
‫ﻣﻲﮔﺮﺩﺩ‬
‫‪X‬‬
‫‪n‬‬
‫‪2‬‬
‫= ‪SSR‬‬ ‫]) ‪[yi − ϕ(x; c1 , . . . , cn‬‬
‫‪i=1‬‬
‫ﺍﻣﺎ ﺷﺎﺧﺺ ﺩﻳﮕﺮﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﺷﺎﺧﺺ ‪ R2‬ﻧﻴﺰ ﺍﻏﻠﺐ ﺑﺮﺍﻱ ﺍﺭﺯﻳﺎﺑﻲ ﺑﺮﺍﺯﺵ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﺩ‪ .‬ﺍﻳﻦ ﺷﺎﺧﺺ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺗﻌﺮﻳﻒ‬
‫ﻣﻲﮔﺮﺩﺩ‬
‫‪SST − SSR‬‬
‫= ‪R2‬‬
‫‪SST‬‬
‫ﺑﻪ ﻗﺴﻤﻲ ﻛﻪ‬
‫‪X‬‬
‫‪n‬‬
‫‪2‬‬
‫= ‪SST‬‬ ‫]̄‪[yi − y‬‬
‫‪i=1‬‬
‫ﻛﻪ ̄‪ y‬ﻫﻤﺎﻥ ﻣﻴﺎﻧﮕﻴﻦ ﻧﻘﺎﻁ ‪ yi , i = 1, . . . , n‬ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۱۰‬‬

‫ﻣﺜﺎﻝ ‪)۸−۸‬ﺑﺮﺭﺳ ﻧﮑﻮﯾﯽ ﺑﺮﺍﺯﺵ(‬

‫ﺩﺭ ﻣﺜﺎﻝ ‪ ،۲-۸‬ﻧﻘﺎﻁ‬
‫)‪(1, 0.5), (2, 2.5), (3, 2), (4, 4), (5.3.5), (6, 6), (7, 5.5‬‬
‫ﺭﺍ ﺑﺎ ﻳﻚ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﺮﺩﻳﻢ ﻭ ﺧﻂ ﺑﺮﺍﺯﺵ ﻛﻨﻨﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﮔﺮﺩﻳﺪ‬
‫‪y = 0.07142857 + 0.8392857x‬‬
‫ﺍﻛﻨﻮﻥ ﺩﺭ ﺍﻳﻦ ﻣﺜﺎﻝ ﻣﻲﺧﻮﺍﻫﻴﻢ∑ﺧﻄﺎﻱ ‪ R2‬ﺑﺮﺍﺯﺵ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ ،R2‬ﺟﺪﻭﻝ ﺯﻳﺮ ﺭﺍ ﺗﺸﻜﻴﻞ ﻣﻲﺩﻫﻴﻢ ﻭ‬
‫‪7‬‬

‫ﺗﻮﺟﻪ ﺩﺍﺭﻳﻢ ﻛﻪ ‪7‬‬

‫‪yi‬‬
‫‪ȳ = i=1‬‬‫‪7‬‬ ‫‪= 24‬‬

‫‪xi‬‬ ‫‪yi‬‬ ‫‪[xi − ȳ]2‬‬ ‫‪[yi − (c1 xi + c0 )]2‬‬

‫‪1‬‬ ‫‪0.5‬‬ ‫‪8.5765‬‬ ‫‪0.1687‬‬
‫‪2‬‬ ‫‪2.5‬‬ ‫‪0.8622‬‬ ‫‪0.5625‬‬
‫‪3‬‬ ‫‪2.0‬‬ ‫‪2.0408‬‬ ‫‪0.3473‬‬
‫‪4‬‬ ‫‪4.0‬‬ ‫‪0.3265‬‬ ‫‪0.3265‬‬
‫‪5‬‬ ‫‪3.5‬‬ ‫‪0.0051‬‬ ‫‪0.5896‬‬
‫‪6‬‬ ‫‪6.0‬‬ ‫‪6.6122‬‬ ‫‪0.7972‬‬
‫‪7‬‬ ‫‪5.5‬‬ ‫‪4.2908‬‬ ‫‪0.1993‬‬
‫‪28‬‬ ‫‪24‬‬ ‫‪22.7143‬‬ ‫‪2.9911‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ ‪ SSR = 2.9911‬ﻭ ‪ .SST = 22.7143‬ﺑﻨﺎﺑﺮﺍﻳﻦ‬

‫‪SST − SSR‬‬ ‫‪22.7143 − 2.9911‬‬
‫= ‪R2‬‬ ‫=‬ ‫‪= 0.868‬‬
‫‪SST‬‬ ‫‪22.7143‬‬

‫ﺷﻜﻞ ‪ :۱.۸‬ﻣﻘﺎﻳﺴﻪ ‪R2‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
‫ﻣﺼﻄﻔﻲ ﺷﻤﺴﻲ ‪ -‬ﺩﺍﻧﺸﮕﺎﻩ ﺻﻨﻌﺘﻲ ﺍﻣﻴﺮﻛﺒﻴﺮ ‪[email protected]‬‬ ‫‪ ۷‬ﻣﺮﺩﺍﺩ ‪۱۴۰۱‬‬

‫‪۹‬‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬
‫ﻓﺼﻞ ﻧﻬﻢ‬

‫ﻓﻬﺮﺳﺖ ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﻓﺼﻞ‬

‫‪۲۱۳‬‬ ‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ‪. . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۹‬‬
‫‪۲۱۴‬‬ ‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ‪. . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۱.۹‬‬
‫‪۲۲۲‬‬ ‫ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ‪. . . . . . . . . . . . . . . . .‬‬ ‫‪۲.۹‬‬
‫‪۲۲۳‬‬ ‫ﻣﺮﺗﺒﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۳.۹‬‬
‫‪۲۲۵‬‬ ‫∗ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۴.۹‬‬
‫‪۲۲۶‬‬ ‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۱.۴.۹‬‬
‫‪۲۲۷‬‬ ‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ ‪. . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۵.۹‬‬
‫‪۲۲۸‬‬ ‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺮﻛﺐ ‪. . . . . . . . . . . . . . . . . . . . . . . . . . . .‬‬ ‫‪۶.۹‬‬

‫‪۲۱۱‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۱۲‬‬

‫ﺍﻧﺘﮕﺮﺍﻝﻫﺎ ﺩﺭ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﻛﺎﺭﺑﺮﺩﻫﺎ ﻇﺎﻫﺮ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻃﻮﻝ ﻗﻮﺱ ‪ ۱‬ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﺯ‬
‫ﻓﺮﻣﻮﻝ ﺯﻳﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪p‬‬
‫‪1 + [f ′ (x)]2 dx‬‬
‫‪a‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﺩﻳﮕﺮ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺍﻧﺘﮕﺮﺍﻝ ﺯﻳﺮ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪ ،‬ﻛﻪ ﺩﺭ ﺍﺭﺗﺒﺎﻁ ﺑﺎ ﺗﻮﺯﻳﻊ ﻧﺮﻣﺎﻝ ﺩﺭ ﺁﻣﺎﺭ ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫‪e−z dz‬‬
‫‪2‬‬

‫‪a‬‬
‫ﺍﻧﺘﮕﺮﺍﻝﻫﺎ ﺑﻪ ﺩﻭ ﺻﻮﺭﺕ ﻧﺎﻣﻌﻴﻦ ‪ ۲‬ﻭ ﻣﻌﻴﻦ ‪ ۳‬ﺑﻪ ﻛﺎﺭ ﺑﺮﺩﻩ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫‪R‬‬
‫• ﺍﻧﺘﮕﺮﺍﻝ ﻧﺎﻣﻌﻴﻦ ﺗﺎﺑﻊ ‪ f‬ﺑﺎ ﻧﻤﺎﺩ ‪ f (x)dx‬ﻣﻌﺮﻓﻲ ﻣﻲﺷﻮﺩ ﻭ ﺑﻪ ﻣﻌﻜﻮﺱ ﻣﺸﺘﻖ ﺍﺷﺎﺭﻩ ﻣﻲﻛﻨﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺍﻧﺘﮕﺮﺍﻝ ﻧﺎﻣﻌﻴﻦ‬
‫ﻳﻚ ﺍﭘﺮﺍﺗﻮﺭ ﺍﺳﺖ ﻛﻪ ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺑﻪ ﺗﺎﺑﻌﻲ ﺩﻳﮕﺮ ﻣﺎﻧﻨﺪ ‪ F‬ﺗﺒﺪﻳﻞ ﻣﻲﻛﻨﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ )‪ .F ′ (x) = f (x‬ﺗﺎﺑﻊ ‪ F‬ﺑﻪ‬
‫ﺗﺎﺑﻊ ﺍﻭﻟﻴﻪ ﻭ ﻳﺎ ﭘﺎﺩﻣﺸﺘﻖ ‪ ۴‬ﺗﺎﺑﻊ ‪ f‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬
‫‪R‬‬
‫• ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎ ﻧﻤﺎﺩ ‪ ab f (x)dx‬ﻧﻤﺎﻳﺶ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ ﻭ ﺣﺎﺻﻞ ﺁﻥ ﻳﻚ ﻋﺪﺩ ﺍﺳﺖ‪ .‬ﺗﻌﺒﻴﺮ‬
‫ﻫﻨﺪﺳﻲ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ﻣﺴﺎﺣﺖ ﻧﺎﺣﻴﻪ ﻣﺤﺼﻮﺭ ﺑﻴﻦ ﺗﺎﺑﻊ ﻭ ﻣﺤﻮﺭ ‪x‬ﻫﺎ ﺩﺭ ﺑﺎﺯﻩ ]‪ .[a, b‬ﺍﻟﺒﺘﻪ ﺗﻌﺎﺑﻴﺮ ﻓﻴﺰﻳﻜﻲ‬
‫ﻧﻴﺰ ﺑﺮﺍﻱ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬
‫‪Rb‬‬
‫ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ ‪ f‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ‬ ‫‪a‬‬
‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ‪f (x)dx‬‬
‫ﻛﻪ ﺍﮔﺮ ‪ F‬ﭘﺎﺩﻣﺸﺘﻖ ‪ f‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫‪f (x)dx = F (b) − F (a).‬‬
‫‪a‬‬
‫‪R‬‬
‫ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ‪ ab f (x)dx‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﺑﻪ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ‪ ۵‬ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﺗﺎﻛﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ‬
‫ﻳﻚ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﺑﻪ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ‪ ،‬ﻧﻴﺎﺯ ﺑﻪ ﺍﺳﺘﺨﺮﺍﺝ ﺗﺎﺑﻊ ﭘﺎﺩﻣﺸﺘﻖ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﺩﺍﺭﺍﻱ‬
‫ﻣﺤﺪﻭﺩﻳﺖﻫﺎﻱ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪ .۱‬ﭘﺎﺩﻣﺸﺘﻖ ﺑﺴﻴﺎﺭﻱ ﺍﺯ ﺗﻮﺍﺑﻊ ﺑﻪ ﻓﺮﻡ ﻣﻘﺪﻣﺎﺗﻲ ‪ ۶‬ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﻧﻤﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﺗﺎﺑﻊ ﭘﺎﺩﻣﺸﺘﻖ ﻳﻚ ﺿﺎﺑﻄﻪ‬
‫ﺷﺎﻣﻞ ﺗﻌﺪﺍﺩ ﻣﺘﻨﺎﻫﻲ ﺟﻤﻠﻪ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ‪ ،‬ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ ‪ sin‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ ‪ − cos‬ﻛﻪ ﺑﻪ ﻓﺮﻡ ﻣﻘﺪﻣﺎﺗﻲ‬
‫
‬
‫ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻣﺎ ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ ) ‪ sin sin(−x2‬ﺭﺍ ﻧﻤﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺭﺍﻳﻪ ﻧﻤﻮﺩ‪ .‬ﻣﺘﺬﻛﺮ ﻣﻲﺷﻮﻳﻢ ﻛﻪ ﺗﺎﺑﻊ‬
‫ﺍﻭﻟﻴﻪ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﺍﻣﺎ ﻧﻤﻲﺗﻮﺍﻥ ﺁﻥ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﻣﻘﺪﻣﺎﺗﻲ ﻧﻮﺷﺖ‪ .‬ﻫﻨﮕﺎﻣﻲ ﻧﺘﻮﺍﻥ ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ ‪ f‬ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﻣﻘﺪﻣﺎﺗﻲ‬
‫ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﻧﻤﻲﺗﻮﺍﻥ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺭﺍ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﺗﺎﺑﻊ ‪ f‬ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬
‫‪p‬‬
‫‪ .۲‬ﻣﺤﺎﺳﺒﻪ ﻭ ﺍﺳﺘﺨﺮﺍﺝ ﺗﺎﺑﻊ ﭘﺎﺩﻣﺸﺘﻖ ﻣﻲﺗﻮﺍﻧﺪ ﭘﻴﭽﻴﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ )‪ tan(x‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﻣﻘﺪﻣﺎﺗﻲ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ ﻭﻟﻲ ﺍﺳﺘﺨﺮﺍﺝ ﺁﻥ ﭼﺎﻟﺸﻲ ﺍﺳﺖ ﻭ ﻧﻴﺎﺯ ﺑﻪ ﺗﻼﺵ ﻭ ﺍﺑﺘﻜﺎﺭ ﺩﺍﺭﺩ‪ .‬ﺍﻟﺒﺘﻪ ﺍﻣﺮﻭﺯﻩ ﺑﺮﻧﺎﻣﻪﻫﺎﻱ‬
‫ﻛﺎﻣﭙﻴﻮﺗﺮﻱ ﺯﻳﺎﺩﻱ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﭘﺎﺩﻣﺸﺘﻖ ﺗﻮﺍﺑﻊ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﺑﺮﻧﺎﻣﻪﻫﺎ ﻣﺒﺘﻨﻲ ﺑﺮ ﺍﻟﮕﻮﺭﻳﺘﻢﻫﺎﻱ ﭘﻴﭽﻴﺪﻩﺍﻱ ﻫﺴﺘﻨﺪ‬
‫ﻭ ﺑﻪ ﺣﺎﻓﻈﻪ ﻭ ﺯﻣﺎﻥ ﻣﺤﺎﺳﺒﺎﺗﻲ ﻧﺴﺒﺘﺎ ﺑﺎﻻﻳﻲ ﻧﻴﺎﺯ ﺩﺍﺭﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺩﻟﻴﻞ ﻧﻤﻲﺗﻮﺍﻥ ﺁﻥﻫﺎ ﺭﺍ ﺑﻪ ﺭﺍﺣﺘﻲ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬

‫‪ .۳‬ﺩﺭ ﺑﺮﺧﻲ ﻣﻮﺍﻗﻊ‪ ،‬ﺿﺎﺑﻄﻪ ﻳﻚ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﺍﺧﺘﻴﺎﺭ ﻧﺪﺍﺭﻳﻢ‪ ،‬ﺑﻠﻜﻪ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺭﺍ ﺩﺭ ﭼﻨﺪ ﻧﻘﻄﻪ ﺩﺭ ﺍﺧﺘﻴﺎﺭ ﺩﺍﺭﻳﻢ‪ .‬ﺑﻪ ﺍﺻﻼﺡ ﺗﺎﺑﻊ‬
‫‪1 Arc‬‬ ‫‪length‬‬ ‫‪4 Antiderivative‬‬
‫‪2 Indefinite‬‬ ‫‪5 Analytical‬‬ ‫‪method‬‬
‫‪3 Definite‬‬ ‫‪6 Elementary‬‬ ‫‪form‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۱۳‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺟﺪﻭﻟﻲ ﺩﺭ ﺍﺧﺘﻴﺎﺭ ﺩﺍﺭﻳﻢ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﺗﻮﺍﺑﻊ ﭘﺎﺩﻣﺸﺘﻖ ﺑﻲﻣﻌﻨﻲ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻧﻤﻲﺗﻮﺍﻥ ﺍﺯ ﺭﻭﺵ ﺗﺤﻠﻴﻠﻲ ﺑﺮﺍﻱ‬
‫ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ﺍﻳﻦ ﺗﻮﺍﺑﻊ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪.‬‬
‫ﻣﺤﺪﻭﺩﻳﺖﻫﺎﻱ ﻓﻮﻕ ﺑﺎﻋﺚ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺩﻧﺒﺎﻝ ﺭﻭﺵﻫﺎﻱ ﺟﺎﻳﮕﺰﻳﻦ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ )ﺗﻘﺮﻳﺒﻲ( ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ﻣﻌﻴﻦ ﺑﺎﺷﻴﻢ‪.‬‬
‫ﺍﺯ ﺟﻤﻠﻪ ﺭﻭﺵﻫﺎﻱ ﺟﺎﻳﮕﺰﻳﻦ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻭ ﺭﻭﺵ ﻣﻮﻧﺖﻛﺎﺭﻟﻮ ﺍﺷﺎﺭﻩ ﻛﺮﺩ‪ .‬ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺼﻞ ﺑﻪ ﺭﻭﺵﻫﺎﻱ‬
‫ﻋﺪﺩﻱ ﭘﺮﺩﺍﺧﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫‪Rb‬‬
‫ﺭﻭﺵﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ a f (x)dx‬ﺍﺯ ﭘﺎﺩﻣﺸﺘﻖ ﺗﺎﺑﻊ ‪ f‬ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻲﻛﻨﺪ‪ ،‬ﺑﻠﻜﻪ ﻓﺮﻣﻮﻟﻲ ﺍﺭﺍﻳﻪ‬
‫ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﺯ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﻳﻚ ﻳﺎ ﭼﻨﺪ ﻧﻘﻄﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ ﻭ ﻣﻘﺪﺍﺭ ﺍﻧﺘﮕﺮﺍﻝ ﺭﺍ ﺗﺨﻤﻴﻦ ﻣﻲﺯﻧﺪ‪ .‬ﻓﺮﻡ ﻛﻠﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)x ≃ w0 f (x0 ) + w1 f (x1 ) + · · · + wn f (xn‬‬ ‫)‪(۱.۹‬‬
‫‪a‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ x0 , . . . , xn‬ﻧﻘﺎﻃﻲ ﻣﺠﺰﺍ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺑﻪ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ ۷‬ﻭ ‪ w0 , . . . , wn‬ﺑﻪ ﻭﺯﻥﻫﺎﻱ ‪ ۸‬ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪.‬‬
‫ﺩﺭ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﻲﺗﻮﺍﻥ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ xi‬ﻭ ﻣﻜﺎﻥ ﺁﻥﻫﺎ ﺭﺍ ﺑﻪ ﺩﻟﺨﻮﺍﻩ ﺍﻧﺘﺨﺎﺏ ﻧﻤﻮﺩ ﻭ ﺳﭙﺲ ﻭﺯﻥﻫﺎﻱ‬
‫ﻓﺮﻣﻮﻝ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﻣﻨﺎﺳﺐ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ ﻛﻪ ﻓﺮﻣﻮﻝ ﺗﺎ ﺁﻥﺟﺎ ﻛﻪ ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﻗﻴﻖ ﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺍﻳﻦ ﻓﺼﻞ‪ ،‬ﺭﻭﺵﻫﺎﻱ ﺳﺎﺧﺖ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺗﺸﺮﻳﺢ ﻣﻲﺷﻮﺩ ﻭ ﺧﺎﻧﻮﺍﺩﻩﻫﺎﻳﻲ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪.‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬ ‫‪۱.۹‬‬

‫ﻳﻚ ﺭﻭﻳﻜﺮﺩ ﺭﺍﻳﺞ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‪ ،‬ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ‬
‫‪R‬‬
‫ﺑﺮﺍﻱ ﺍﺭﺍﻳﻪ ﻓﺮﻣﻮﻟﻲ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ ab f (x)dx‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻋﻤﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫• ﺍﺑﺘﺪﺍ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ x0 , . . . , xn‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﻧﺪ‪.‬‬
‫• ﺳﭙﺲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺗﺎﺑﻊ ‪ f‬ﻣﺒﺘﻨﻲ ﺑﺮ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ x0 , . . . , xn‬ﺑﻪ ﻋﻨﻮﺍﻥ ﺗﻘﺮﻳﺒﻲ ﺍﺯ ‪ f‬ﺩﺭ ﻧﻈﺮ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺏ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫)‪Pn (x) = f (x0 )ℓ0 (x) + f (x1 )ℓ1 (x) + · · · + f (xn )ℓn (x‬‬
‫‪Rb‬‬
‫• ﺣﺎﻝ ﺑﻪ ﺟﺎﻱ ﺗﺎﺑﻊ ‪ f‬ﺍﺯ ﺗﻘﺮﻳﺐ ﺁﻥ‪ ،‬ﻳﻌﻨﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭﻭﻧﻴﺎﺑﺶ‪ ،‬ﺍﻧﺘﮕﺮﺍﻝ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪f (x)dx ،‬‬
‫‪a‬‬
‫‪R‬‬
‫ﺭﺍ ﺑﺎ ‪ ab P(x)dx‬ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﻴﻢ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻳﻚ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺭﺍ ﻣﻲﺗﻮﺍﻧﻴﻢ ﺑﻪ ﺻﻮﺭﺕ ﺗﺤﻠﻴﻠﻲ ﻣﺤﺎﺳﺒﻪ‬
‫ﻛﻨﻴﻢ‪.‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻮﺿﻴﺤﺎﺕ ﻓﻮﻕ‪ ،‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪b‬‬
‫≃ ‪f (x)dx‬‬ ‫‪Pn (x)dx‬‬
‫‪a‬‬ ‫‪a‬‬
‫‪Z‬‬ ‫‪b‬‬
‫=‬ ‫‪[f (x0 )ℓ0 (x) + f (x1 )ℓ1 (x) + · · · + f (xn )ℓn (x)] dx‬‬
‫‪a‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪b‬‬
‫) ‪= f (x0‬‬ ‫) ‪ℓ0 (x)dx + f (x1‬‬ ‫) ‪ℓ1 (x)dx + · · · + f (xn‬‬ ‫‪ℓn (x)dx‬‬
‫‪a‬‬ ‫‪a‬‬ ‫‪a‬‬
‫‪7 Nodes‬‬ ‫‪8 Weights‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۱۴‬‬

‫‪Rb‬‬
‫=‪ wi :‬ﺩﺍﺭﻳﻢ‪:‬‬ ‫‪a‬‬
‫ﺣﺎﻝ ﺑﺎ ﻓﺮﺽ ‪ℓi (x)dx‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)dx ≃ w0 f (x0 ) + w1 f (x1 ) + · · · + wn f (xn‬‬
‫‪a‬‬
‫ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﻳﻚ ﻓﺮﻣﻮﻝ ﻳﺎ ﻗﺎﻋﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﻲﺑﺎﺷﺪ ﻛﻪ ﺑﺮﻣﺒﻨﺎﻱ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﮔﺮﺩﻳﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ‬
‫ﺩﺭ ﻓﺮﻣﻮﻝ ﻓﻮﻕ‪ ،‬ﺭﻭﺷﻲ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻭﺯﻥﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﻣﻄﺎﻟﺐ ﺍﻳﻦ ﺑﺨﺶ ﺩﺭ ﻗﺎﻟﺐ ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺍﺭﺍﻳﻪ ﻣﻲ ﺷﻮﺩ‪.‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﯼ ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻣﺒﺘﻨ ﺑﺮ ﺩﺭﻭﻧﯿﺎﺑﯽ ﭼﻨﺪﺟﻤﻠەﺍﯼ‬ ‫ﻗﻀﯿﻪ ‪۱−۹‬‬

‫‪Rb‬‬
‫ﺑﺮﺣﺴﺐ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬ ‫‪a‬‬
‫ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ )ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ( ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪f (x)dx‬‬
‫‪ x0 , . . . , xn‬ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬
‫≃ ‪f (x)dx‬‬ ‫) ‪wi f (xi‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪Y‬‬
‫‪n‬‬
‫‪b‬‬ ‫‪b‬‬
‫‪x − xj‬‬
‫=‪wi :‬‬ ‫= ‪ℓi (x)dx‬‬ ‫‪dx‬‬
‫‪a‬‬ ‫‪a j=0,j̸=i‬‬ ‫‪xi − xj‬‬

‫■ ﺩﺳﺘﻪﺑﻨﺪﻱ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬

‫ﻛﻠﻴﺖ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺩﺭ ﻗﻀﻴﻪ ‪ ۱-۹‬ﺍﺭﺍﻳﻪ ﮔﺮﺩﻳﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬
‫ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﻧﺪ ﻭ ﻭﺯﻥﻫﺎﻱ ﻓﺮﻣﻮﻝ ﺑﺮﺣﺴﺐ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﻨﺪ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ‪ ،‬ﻓﻘﻂ ﺩﺭ ﺍﻧﺘﺨﺎﺏ‬
‫ﺗﻌﺪﺍﺩ ﻭ ﻣﻜﺎﻥ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺍﺧﺘﻴﺎﺭ ﺩﺍﺭﻳﻢ‪ .‬ﺭﻭﻳﻜﺮﺩﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺩﺭ ﺍﻧﺘﺨﺎﺏ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻭﺟﻮﺩ ﺩﺍﺭﺩ ﻛﻪ ﺑﺴﺘﻪ ﺑﻪ ﻧﻮﻉ ﺍﻳﻦ ﺍﻧﺘﺨﺎﺏ‬
‫ﻳﻚ ﺩﺳﺘﻪﺑﻨﺪﻱ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪:‬‬

‫‪ .۱‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ‬

‫‪ .۲‬ﻓﺮﻣﻮﻝﻫﺎﻱ ﮔﻮﺳﻲ‬

‫‪ .۳‬ﻓﺮﻣﻮﻝﻫﺎﻳﻲ ﺑﺎ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺍﺧﺘﻴﺎﺭﻱ‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ‬ ‫�‬

‫ﺩﺭ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﻧﺪ‪ .‬ﻓﺮﺽ ﻛﻨﻴﺪ ﺑﺎﺯﻩ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﻪ ﺻﻮﺭﺕ ]‪ [a, b‬ﺑﺎﺷﺪ ﻭ ‪.n ≥ 1‬‬
‫ﻧﻘﺎﻁ ﻧﯿﻮﺗﻦ⁃ﮐﺎﺗﺲ ﺑﺴﺘﻪ‬ ‫ﺗﻌﺮﯾﻒ ‪۲−۹‬‬
‫ﺑﺎﺯﻩ ]‪ [a, b‬ﻭ ‪ n ≥ 1‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﺎ ﺗﻘﺴﻴﻢ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ‪ n‬ﻗﺴﻤﺖ ﻣﺴﺎﻭﻱ‪ ،‬ﺑﻪ ﺗﻌﺪﺍﺩ ‪ n + 1‬ﻧﻘﻄﻪ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‬
‫ﻛﻪ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﮔﺮﻩﻫﺎ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ )‪-(n + 1‬ﺗﺎﻳﻲ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ )‪-(n + 1‬ﺗﺎﻳﻲ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫‪b−a‬‬
‫‪xi := a + ih, i = 0, . . . , n‬‬ ‫=‪h :‬‬
‫‪n‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۱۵‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﻣﻌﺎﺩﻝ ﺯﻳﺮ ﻧﻴﺰ ﻧﻮﺷﺖ‪:‬‬

‫‪ib − (n − i)a‬‬
‫=‪xi :‬‬ ‫‪, i = 0, . . . , n.‬‬
‫‪n‬‬

‫ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻫﻤﺎﻧﻨﺪ ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺗﻌﺮﻳﻒ ﻣﻲ ﺷﻮﻧﺪ‪ ،‬ﺑﺎ ﺍﻳﻦ ﺗﻔﺎﻭﺕ ﻛﻪ ﻧﻘﺎﻁ ﺍﺑﺘﺪﺍ ﻭ ﺍﻧﺘﻬﺎﻱ ﺑﺎﺯﻩ ﺭﺍ‬
‫ﺷﺎﻣﻞ ﻧﻤﻲﺷﻮﻧﺪ‪ .‬ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﻧﻘﺎﻁ ﻧﯿﻮﺗﻦ⁃ﮐﺎﺗﺲ ﺑﺎﺯ‬ ‫ﺗﻌﺮﯾﻒ ‪۳−۹‬‬

‫ﺑﺎﺯﻩ ]‪ [a, b‬ﻭ ‪ n ≥ 0‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﺑﺎ ﺗﻘﺴﻴﻢ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ‪ n + 2‬ﻗﺴﻤﺖ ﻣﺴﺎﻭﻱ‪ ،‬ﺑﻪ ﺗﻌﺪﺍﺩ ‪ n + 3‬ﻧﻘﻄﻪ ﺣﺎﺻﻞ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺑﺎ ﺣﺬﻑ ﺍﻭﻟﻴﻦ ﻭ ﺁﺧﺮﻳﻦ ﻧﻘﻄﻪ‪ ،‬ﺑﻪ ﺗﻌﺪﺍﺩ ‪ n + 1‬ﻧﻘﻄﻪ ﺑﺎﻗﻲ ﻣﻲﻣﺎﻧﺪ ﻛﻪ ﺍﻳﻦ ﻧﻘﺎﻁ ﺭﺍ ﮔﺮﻩﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ‬
‫)‪-(n + 1‬ﺗﺎﻳﻲ ﻣﻲﻧﺎﻣﻴﻢ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ )‪-(n + 1‬ﺗﺎﻳﻲ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫‪b−a‬‬
‫‪xi := a + ih, i = 1, . . . , n + 1‬‬ ‫=‪h :‬‬
‫‪n+2‬‬
‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﻣﻌﺎﺩﻝ ﺯﻳﺮ ﻧﻴﺰ ﻧﻮﺷﺖ‪:‬‬
‫‪ib − (n − i)a‬‬
‫=‪xi :‬‬ ‫‪, i = 1, . . . , n + 1.‬‬
‫‪n+2‬‬

‫ﺑﺮﺍﻱ ﺗﺠﺴﻢ ﺑﻬﺘﺮ ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ‪ ،‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫نیوتن‪-‬کاتسی بسته دوتایی )‪(𝑛 = 1‬‬

‫‪a‬‬ ‫‪h‬‬ ‫‪b‬‬

‫‪x0‬‬ ‫‪x1‬‬

‫نیوتن‪-‬کاتسی بسته سهتایی )‪(𝑛 = 2‬‬ ‫نیوتن‪-‬کاتسی باز یکتایی )‪(𝑛 = 0‬‬
‫‪a b‬‬ ‫‪a b‬‬
‫‪a‬‬ ‫‪h‬‬ ‫‪2‬‬ ‫‪b‬‬ ‫‪a‬‬ ‫‪h‬‬ ‫‪2‬‬ ‫‪b‬‬

‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x‬‬ ‫‪1‬‬ ‫‪x0‬‬ ‫‪x1‬‬

‫نیوتن‪-‬کاتسی بسته چهارتایی )‪(𝑛 = 3‬‬ ‫نیوتن‪-‬کاتسی باز دوتایی )‪(𝑛 = 1‬‬
‫‪2a b‬‬ ‫‪a 2b‬‬ ‫‪2a b‬‬
‫‪a‬‬ ‫‪h‬‬ ‫‪b‬‬ ‫‪a‬‬ ‫‪a 2b‬‬
‫‪b‬‬
‫‪3‬‬ ‫‪3‬‬ ‫‪h‬‬ ‫‪3‬‬ ‫‪3‬‬

‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬

‫‪x‬‬ ‫‪1‬‬

‫نیوتن‪-‬کاتسی بسته پنجتایی )‪(𝑛 = 5‬‬ ‫نیوتن‪-‬کاتسی بسته سهتایی )‪(𝑛 = 2‬‬
‫‪3a b‬‬ ‫‪a b‬‬ ‫‪a 3b‬‬ ‫‪3a b‬‬ ‫‪a b‬‬ ‫‪a 3b‬‬
‫‪a‬‬ ‫‪h‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪b‬‬ ‫‪a‬‬ ‫‪h‬‬ ‫‪4‬‬ ‫‪2‬‬ ‫‪4‬‬ ‫‪b‬‬

‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫‪x‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬
‫‪1‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ‪-n‬ﺗﺎﻳﻲ ﻫﻤﺎﻥ ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ‪-n + 2‬ﺗﺎﻳﻲ ﺍﺳﺖ‬
‫ﻛﻪ ﻧﻘﻄﻪ ﺍﻭﻝ ﻭ ﺁﺧﺮ ﺣﺬﻑ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﺍﺩﺍﻣﻪ ﺑﺮﺧﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﻛﻨﻴﻢ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۱۶‬‬

‫■ ﻓﺮﻣﻮﻝ ﺩﻭﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ )ﻓﺮﻣﻮﻝ ﺫﻭﺯﻧﻘﻪﺍﻱ(‬

‫‪Rb‬‬
‫ﺍﺯ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺯﻳﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‬ ‫‪a‬‬
‫ﻓﺮﻣﻮﻝ ﺩﻭ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪f (x)dx‬‬
‫‪x0 := a,‬‬ ‫‪x1 := b,‬‬ ‫‪h := b − a = x1 − x0‬‬
‫ﻭ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)dx ≃ w0 f (x0 ) + w1 f (x1‬‬
‫‪a‬‬
‫ﻛﻪ ‪ w0‬ﻭ ‪ w1‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﻨﺪ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪2 x1‬‬
‫‪x1‬‬ ‫‪x1‬‬
‫‪x − x1‬‬ ‫) ‪1 (x − x1‬‬ ‫‪1‬‬ ‫‪h‬‬
‫= ‪w0‬‬ ‫= ‪ℓ0 (x)dx‬‬ ‫= ‪dx‬‬ ‫= ) ‪= − (x0 − x1‬‬
‫‪x0‬‬ ‫‪x0‬‬ ‫‪x0 − x1‬‬ ‫‪2 x0 − x1‬‬ ‫‪x0‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪2 x1‬‬
‫‪x1‬‬ ‫‪x1‬‬
‫‪x − x0‬‬ ‫) ‪1 (x − x0‬‬ ‫‪1‬‬ ‫‪h‬‬
‫= ‪w1‬‬ ‫= ‪ℓ1 (x)dx‬‬ ‫= ‪dx‬‬ ‫=‬ ‫‪(x1 − x0 ) = .‬‬
‫‪x0‬‬ ‫‪x0‬‬ ‫‪x1 − x0‬‬ ‫‪2 x1 − x0‬‬ ‫‪x0‬‬ ‫‪2‬‬ ‫‪2‬‬
‫ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﺿﺮﺍﻳﺐ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﻓﺮﻣﻮﻝ ﺩﻭﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫‪h‬‬ ‫‪h‬‬
‫≃ ‪f (x)dx‬‬
‫) ‪f (x0 ) + f (x1‬‬
‫‪a‬‬ ‫‪2‬‬ ‫‪2‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ x0 = a ،h = b − a‬ﻭ ‪ x1 = b‬ﻟﺬﺍ ﻓﺮﻣﻮﻝ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﺑﻴﺎﻥ ﻛﺮﺩ‪:‬‬
‫‪Z b‬‬
‫‪b−a‬‬
‫≃ ‪f (x)dx‬‬ ‫])‪[f (a) + f (b‬‬
‫‪a‬‬ ‫‪2‬‬
‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﻓﺮﻣﻮﻝ ﺩﻭ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺑﻪ ﻗﺎﻋﺪﻩ ﻳﺎ ﻓﺮﻣﻮﻝ ﺫﻭﺯﻧﻘﻪ ‪ ۹‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪.‬‬

‫ﺩﻟﻴﻞ ﺍﻳﻦ ﻧﺎﻣﮕﺬﺍﺭﻱ ﺑﻪ ﺧﺎﻃﺮ ﺗﻌﺒﻴﺮ ﻫﻨﺪﺳﻲ ﺍﻳﻦ ﻓﺮﻣﻮﻝ ﺍﺳﺖ‪ .‬ﺩﺭ‬
‫‪R‬‬
‫ﻭﺍﻗﻊ ‪ ab f (x)dx‬ﻣﺴﺎﺣﺖ ﻣﺤﺼﻮﺭ ﺑﻴﻦ ﺗﺎﺑﻊ ‪ f‬ﻭ ﻣﺤﻮﺭ ‪x‬ﻫﺎ‬
‫ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﺯ ﺳﻮﻱ ﺩﻳﮕﺮ ])‪2 [f (a) + f (b‬‬
‫‪b−a‬‬

‫ﻣﺴﺎﺣﺖ ﺫﻭﺯﻧﻘﻪﺍﻱ ﺍﺳﺖ ﻛﻪ ﺍﺯ ﺍﺗﺼﺎﻝ ﻧﻘﺎﻁ )‪،(b, 0) ،(a, 0‬‬

‫))‪ (b, f (b‬ﻭ ))‪ (a, f (a‬ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪ .‬ﺷﻜﻞ ﺭﻭﺑﺮﻭ ﺭﺍ‬
‫ﺑﺒﻴﻨﻴﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ‪ ،‬ﺩﺭ ﻗﺎﻋﺪﻩ ﺩﻭﻧﻘﻄﻪﺍﻱ‬
‫ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ‪ ،‬ﻣﺴﺎﺣﺖ ﺯﻳﺮ ﺗﺎﺑﻊ ﺑﺎ ﻣﺴﺎﺣﺖ ﻳﻚ ﺫﻭﺯﻧﻘﻪ ﺗﻘﺮﻳﺐ‬
‫ﺯﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫■ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺳﻪ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ )ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ(‬

‫‪R‬‬
‫ﺍﮔﺮ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ ab f (x)dx‬ﺍﺯ ﺳﻪ ﻧﻘﻄﻪ‬
‫‪a+b‬‬
‫=‪x0 := a, x1 :‬‬ ‫‪, x2 := b‬‬
‫‪2‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻓﺮﻣﻮﻝ ﺣﺎﺻﻞ ﺑﻪ ﻓﺮﻣﻮﻝ ﺳﻪ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﻭ ﻳﺎ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ‪ ۱۰‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺍﻳﻦ‬
‫ﻗﺎﻋﺪﻩ ﺑﻪ ﻓﺮﻡ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)dx ≃ w0 f (x0 ) + w1 f (x1 ) + w2 f (x2‬‬
‫‪a‬‬
‫‪9 Trapezoidal‬‬ ‫‪rule‬‬ ‫‪10 Simpson’s‬‬ ‫‪rule‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۱۷‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫=‪.h :‬‬ ‫‪b−a‬‬

‫‪2‬‬ ‫ﺗﺎﻛﻴﺪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﻘﺎﻁ ‪ x0 , x1 , x2‬ﻧﻘﺎﻁ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﻫﺴﺘﻨﺪ ﻭ ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﺩﻭ ﻧﻘﻄﻪ ﻣﺘﻮﺍﻟﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫ﻣﻄﺎﺑﻖ ﺑﺎ ﺗﻌﺮﻳﻒ ‪ ،۲-۹‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻭﺯﻥﻫﺎﻱ ‪ w1 ،w0‬ﻭ ‪ w2‬ﻗﺮﺍﺭ ﻣﻲﺩﻫﻴﻢ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬
‫‪b‬‬ ‫‪b‬‬
‫) ‪(x − x1 )(x − x2‬‬
‫= ‪w0‬‬ ‫= ‪ℓ0 (x)dx‬‬ ‫‪dx‬‬
‫‪a‬‬ ‫‪a‬‬ ‫) ‪(x0 − x1 )(x0 − x2‬‬
‫ﺑﺮﺍﻱ ﺳﺎﺩﻩ ﺷﺪﻥ ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ﻓﻮﻕ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ‪ x = th + x0‬ﺭﺍ ﺍﻋﻤﺎﻝ ﻧﻤﻮﺩ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺍﻳﻦ ﻧﻜﺘﻪ‬
‫ﻛﻪ ‪ ،xi = x0 + ih‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪Z‬‬
‫‪2‬‬
‫‪(t − 1)h(t − 2)h‬‬ ‫‪2‬‬
‫)‪(t − 1)(t − 2‬‬ ‫‪h‬‬ ‫‪2‬‬
‫= ‪w0‬‬ ‫‪hdt = h‬‬ ‫= ‪dt‬‬ ‫‪[t2 − 3t + 2]dt‬‬
‫‪0‬‬ ‫‪(0 − 1)h(0 − 2)h‬‬ ‫‪0‬‬ ‫)‪(0 − 1)(0 − 2‬‬ ‫‪2‬‬ ‫‪0‬‬
‫‪h‬‬
‫=‬
‫‪3‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ‪ ،x = x0 + th‬ﺑﺎﻋﺚ ﺳﺎﺩﻩ ﺷﺪﻥ ﺍﻧﺘﮕﺮﺍﻝ ﮔﻴﺮﻱ ﻣﻲﺷﻮﺩ‪ .‬ﻫﻤﻴﻦ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ﺭﺍ ﻣﻲﺗﻮﺍﻥ‬
‫ﺑﺮﺍﻱ ﺳﺎﺩﻩﺳﺎﺯﻱ ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ﻇﺎﻫﺮ ﺷﺪﻩ ﺩﺭ ‪ w1‬ﻭ ‪ w2‬ﻧﻴﺰ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪2‬‬ ‫‪Z‬‬ ‫‪2‬‬
‫) ‪(x−x0 )(x−x2‬‬ ‫)‪(t−0)(t−2‬‬
‫= ‪w1‬‬ ‫= ‪ℓ1 (x)dx‬‬ ‫‪(x1 −x0 )(x1 −x2 ) dx‬‬ ‫‪=h‬‬ ‫‪(1−0)(1−2) dt‬‬ ‫‪= −h‬‬ ‫‪[t2 − 2t]dt‬‬
‫‪a‬‬ ‫‪a‬‬ ‫‪0‬‬ ‫‪0‬‬
‫‪3h‬‬
‫=‬
‫‪3‬‬
‫‪Z b‬‬ ‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪2‬‬ ‫‪Z‬‬ ‫‪2‬‬
‫) ‪(x−x0 )(x−x1‬‬ ‫)‪(t−0)(t−1‬‬ ‫‪h‬‬
‫= ‪w2‬‬ ‫= ‪ℓ2 (x)dx‬‬ ‫‪(x2 −x0 )(x2 −x1 ) dx‬‬ ‫‪=h‬‬ ‫‪(2−0)(2−1) dt‬‬ ‫=‬ ‫‪[t2 − t]dt‬‬
‫‪a‬‬ ‫‪a‬‬ ‫‪0‬‬ ‫‪2‬‬ ‫‪0‬‬
‫‪h‬‬
‫=‬
‫‪3‬‬
‫ﺑﺎ ﺟﺎﻳﮕﺬﺍﺭﻱ ﻭﺯﻥﻫﺎﻱ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﻓﺮﻣﻮﻝ ﺳﻴﻤﺴﻮﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪x2‬‬
‫‪h‬‬
‫≃ ‪f (x)dx‬‬ ‫]) ‪[f (x0 ) + 4f (x1 ) + f (x2‬‬ ‫)‪(۲.۹‬‬
‫‪x0‬‬ ‫‪3‬‬
‫ﻓﺮﻡ ﺩﻳﮕﺮ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻧﻤﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪x2‬‬
‫‪b−1‬‬ ‫‪a+b‬‬
‫≃ ‪f (x)dx‬‬ ‫( ‪[f (a) + 4f‬‬ ‫])‪) + f (b‬‬ ‫)‪(۳.۹‬‬
‫‪x0‬‬ ‫‪6‬‬ ‫‪2‬‬

‫■ ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﻳﻚ ﻧﻘﻄﻪﺍﻱ ﺑﺎﺯ )ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ(‬

‫‪R‬‬
‫ﺩﺭ ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﻳﻚ ﻧﻘﻄﻪﺍﻱ ﺑﺎﺯ‪ ،‬ﺍﻧﺘﮕﺮﺍﻝ ﻣﻌﻴﻦ ‪ ab f (x)dx‬ﺗﻨﻬﺎ ﺑﺮﺣﺴﺐ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﺎﺯﻩ ]‪ [a, b‬ﺗﻘﺮﻳﺐ ﺯﺩﻩ‬
‫ﻣﻲﺷﻮﺩ‪ .‬ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﮔﺮﻩ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻳﻚﺗﺎﻳﻲ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺑﺎﺯﻩ ]‪ [a, b‬ﻫﻤﺎﻥ ﻧﻘﻄﻪ ﻭﺳﻂ ﺑﺎﺯﻩ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻓﺮﻣﻮﻝ‬
‫ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻳﻚ ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﻗﺎﻋﺪﻩ ﻣﻴﺎﻧﻲ ‪ ۱۱‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﺍﺳﺖ‪ .‬ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)dx ≃ w0 f (x0‬‬
‫‪a‬‬
‫= ‪ x0‬ﻭ‬ ‫‪a+b‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫‪Z‬‬ ‫‪x1‬‬ ‫‪Z‬‬ ‫‪x1‬‬
‫‪2‬‬

‫=‪w0 :‬‬ ‫= ‪ℓ0 (x)dx‬‬ ‫‪dx = x1 − x−1 = 2h‬‬

‫‪x−1‬‬ ‫‪x0‬‬

‫‪11 Midpoint‬‬ ‫‪rule‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۱۸‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ‪ ℓ0‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﭘﺎﻳﻪﺍﻱ ﻻﮔﺮﺍﻧﮋ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﺗﻨﻬﺎ ﻧﻘﻄﻪ ‪ x0‬ﺍﺳﺖ ﻭ ﻟﺬﺍ ‪ .ℓ0 (x) = 1‬ﺣﺎﻝ ﺑﺎ ﺟﺎﻳﮕﺰﺍﺭﻱ‪ ،‬ﻓﺮﻣﻮﻝ‬
‫ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪x1‬‬
‫) ‪f (x)dx ≃ 2hf (x0‬‬ ‫)‪(۴.۹‬‬
‫‪x−1‬‬
‫ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﺑﻴﺎﻥ ﻛﺮﺩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‬ ‫‬
‫‪a+b‬‬
‫‪f (x)dx ≃ (b − a)f‬‬ ‫)‪(۵.۹‬‬
‫‪a‬‬ ‫‪2‬‬

‫‪a+b‬‬
‫
‬
‫‪f‬‬ ‫‪2‬‬

‫ﺑﻪ ﻟﺤﺎﻅ ﻫﻨﺪﺳﻲ‪ ،‬ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺳﻄﺢ ﺯﻳﺮ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﺎ‬
‫ﻳﻚ ﻣﺴﺘﻄﻴﻞ ﺗﻘﺮﻳﺐ ﻣﻲﺯﻧﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ﻳﻚ ﻭﺟﻪ ﻣﺴﺘﻄﻴﻞ ﺭﻭﻱ ﺑﺎﺯﻩ‬
‫‪ f ( a+b‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﺷﻜﻞ ﺭﻭﺑﺮﻭ‬ ‫]‪ [a, b‬ﻣﻨﻄﺒﻖ ﺍﺳﺖ ﻭ ﻭﺟﻪ ﺩﻳﮕﺮ ﺑﻪ ﺍﻧﺪﺍﺯﻩ ) ‪2‬‬
‫ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫‪x‬‬
‫‪a‬‬ ‫‪a+b‬‬ ‫‪b‬‬
‫‪2‬‬

‫■ ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺩﻭ ﻧﻘﻄﻪﺍﻱ ﺑﺎﺯ‬

‫ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﺩﻭﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪Z‬‬ ‫‪x2‬‬
‫) ‪f (x)dx ≃ w0 f (x0 ) + w1 f (x1‬‬
‫‪x−1‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫‪2a + b‬‬ ‫‪a + 2b‬‬ ‫‪b−a‬‬
‫=‪x−1 := a, x0 :‬‬ ‫=‪, x1 :‬‬ ‫=‪, x2 = b, h :‬‬
‫‪3‬‬ ‫‪3‬‬ ‫‪3‬‬
‫ﺣﺎﻝ ﻭﺯﻥﻫﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﻧﺪ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪Z‬‬
‫‪x2‬‬ ‫‪x2‬‬
‫‪x − x1‬‬ ‫‪2‬‬
‫‪t−1‬‬ ‫‪3h‬‬
‫=‪w0 :‬‬ ‫= ‪ℓ0 (x)dx‬‬ ‫‪dx = h‬‬ ‫= ‪dt‬‬ ‫‪,‬‬
‫‪x−1‬‬ ‫‪x−1‬‬ ‫‪x0 − x1‬‬ ‫‪−1‬‬ ‫‪0−1‬‬ ‫‪2‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‪Z‬‬
‫‪x2‬‬ ‫‪x2‬‬
‫‪x − x0‬‬ ‫‪2‬‬
‫‪t−0‬‬ ‫‪3h‬‬
‫=‪w1 :‬‬ ‫= ‪ℓ1 (x)dx‬‬ ‫‪dx = h‬‬ ‫= ‪dt‬‬ ‫‪,‬‬
‫‪x−1‬‬ ‫‪x−1‬‬ ‫‪x1 − x0‬‬ ‫‪−1‬‬ ‫‪1−0‬‬ ‫‪2‬‬

‫ﺑﻨﺎﺑﺮﺍﻳﻦ‪ ،‬ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﺩﻭﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﻣﻲﺷﻮﺩ‪:‬‬

‫‪Z‬‬ ‫‪x2‬‬
‫‪3h‬‬ ‫‪3h‬‬
‫≃ ‪f (x)dx‬‬ ‫‪f (x0 ) +‬‬ ‫) ‪f (x1‬‬ ‫)‪(۶.۹‬‬
‫‪x−1‬‬ ‫‪2‬‬ ‫‪2‬‬
‫ﺍﻳﻦ ﻓﺮﻣﻮﻝ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﺑﻴﺎﻥ ﻛﺮﺩ‪:‬‬
‫‪Z‬‬ ‫ ‬ ‫‬ ‫‬ ‫‬
‫‪b‬‬
‫‪b−a‬‬ ‫‪2a + b‬‬ ‫‪a + 2b‬‬
‫≃ ‪f (x)dx‬‬ ‫‪f‬‬ ‫‪+f‬‬ ‫)‪(۷.۹‬‬
‫‪a‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۱۹‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫■ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺑﺎﻻﺗﺮ‬

‫ﺩﺭ ﻗﺴﻤﺖﻫﺎﻱ ﻗﺒﻞ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺩﻭ ﻭ ﺳﻪ ﻧﻘﻄﻪﺍﻱ ﺑﺴﺘﻪ ﻭ ﻫﻤﭽﻨﻴﻦ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻳﻚ ﻭ ﺩﻭﻧﻘﻄﻪﺍﻱ ﺑﺎﺯ ﺍﺳﺘﺨﺮﺍﺝ‬
‫ﮔﺮﺩﻳﺪ‪ .‬ﺑﻪ ﻫﻤﺎﻥ ﺭﻭﺵ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺑﺎﻻﺗﺮ ﺭﺍ ﻧﻴﺰ ﺍﺳﺘﺨﺮﺝ ﻛﺮﺩ‪ .‬ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ ﺑﺎﻳﺪ‬
‫ﻭﺯﻥﻫﺎﻱ ﻓﺮﻣﻮﻝ ﺭﺍ ﺑﺎ ﻳﻚ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺗﺤﻠﻴﻠﻲ ﺗﻌﻴﻴﻦ ﻛﺮﺩ‪ .‬ﺍﺯ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ‪ x = x0 + th‬ﺑﺮﺍﻱ ﺳﺎﺩﮔﻲ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﺗﺤﻠﻴﻠﻲ‬
‫ﺍﻧﺘﮕﺮﺍﻝﻫﺎ ﻣﻲﺗﻮﺍﻥ ﺍﺳﺘﻔﺎﺩﻩ ﻧﻤﻮﺩ‪ .‬ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ﻣﺮﺑﻮﻃﻪ ﻭ ﺗﻐﻴﻴﺮ ﻣﺘﻐﻴﺮ ﺩﺍﺩﻩ ﺷﺪﻩ ﺁﻥﻫﺎ ﺑﺮﺍﻱ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺑﻪ‬
‫ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫‪Z‬‬ ‫‪Y‬‬‫‪n‬‬ ‫‪Z‬‬ ‫‪Y‬‬
‫‪n‬‬
‫‪xn‬‬
‫‪x − xj‬‬ ‫‪h‬‬ ‫‪n‬‬
‫= ‪wi‬‬ ‫= ‪dx‬‬ ‫‪(t − j)dt‬‬
‫‪x‬‬ ‫‪− xj‬‬ ‫‪Y‬‬
‫‪n‬‬
‫‪x0‬‬ ‫‪j=0 i‬‬ ‫‪0‬‬
‫‪j̸=i‬‬
‫)‪(i − j‬‬ ‫‪j=0,j̸=i‬‬

‫‪j=0,j̸=i‬‬
‫ﻭ ﺑﺮﺍﻱ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪n+1 Y‬‬
‫‪n‬‬ ‫‪Z‬‬ ‫‪Y‬‬
‫‪n‬‬
‫‪x − xj‬‬ ‫‪h‬‬ ‫‪n+1‬‬
‫= ‪wi‬‬ ‫= ‪dx‬‬ ‫‪(t − j)dt‬‬
‫‪−1‬‬ ‫‪x‬‬ ‫‪− xj‬‬ ‫‪Y‬‬
‫‪n‬‬
‫‪−1‬‬
‫‪j=0 i‬‬
‫‪j̸=i‬‬
‫)‪(i − j‬‬ ‫‪j=0,j̸=i‬‬

‫‪j=0,j̸=i‬‬

‫ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﻓﺮﻣﻮﻝ ﭼﻬﺎﺭ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺑﻪ ﻗﺎﻋﺪﻩ ‪ ۳/۸‬ﺳﻴﻤﺴﻮﻥ ‪ ۱۲‬ﻭ ﻓﺮﻣﻮﻝ ﭘﻨﺞ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬‬
‫ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺑﻪ ﻗﺎﻋﺪﻩ ﺑﻮﻝ ‪ ۱۳‬ﻧﻴﺰ ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ ﺑﻪ ﻫﻤﺮﺍﻩ ﺑﺮﺧﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺩﻳﮕﺮ ﺩﺭ ﺍﻧﺘﻬﺎﻱ ﺍﻳﻦ ﺑﺨﺶ ﺁﻭﺭﺩﻩ‬
‫ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫■ ﺟﻤﻠﻪ ﺧﻄﺎ )ﺧﻄﺎﻱ ﺑﺮﺷﻲ( ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ‬

‫ﺑﺮﺍﻱ ﻗﺎﻋﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺯﻳﺮ‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬
‫= ‪f (x)dx‬‬ ‫) ‪wi f (xi‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫ﺟﻤﻠﻪ ﺧﻄﺎ ﻭ ﻳﺎ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ‪ ۱۴‬ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬
‫=‪E :‬‬ ‫‪f (x)dx −‬‬ ‫) ‪wi f (xi‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫ﺑﻪ ﺯﺑﺎﻥ ﺳﺎﺩﻩ‪ ،‬ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‪ ،‬ﻫﻤﺎﻥ ﺗﻔﺎﻭﺕ ﻣﻘﺪﺍﺭ ﻭﺍﻗﻌﻲ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺎ ﻣﻘﺪﺍﺭ ﺗﻘﺮﻳﺒﻲ ﺣﺎﺻﻞ ﺍﺯ‬
‫ﻓﺮﻣﻮﻝ ﺍﺳﺖ‪.‬‬
‫ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ﺩﺭ ﻗﻀﻴﻪﻫﺎﻱ ﺯﻳﺮ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺧﻄﺎﯼ ﻓﺮﻣﻮﻝﻫﺎﯼ ﻧﯿﻮﺗﻦ⁃ﮐﺎﺗﺴ ﺑﺴﺘﻪ‬ ‫ﻗﻀﯿﻪ ‪۴−۹‬‬

‫ﺍﮔﺮ ‪ x0 , . . . , xn‬ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﻨﺪ ﺁﻥﮔﺎﻩ‬

‫• ﺍﮔﺮ ‪ n ≥ 2‬ﻋﺪﺩﻱ ﺯﻭﺝ ﺑﺎﺷﺪ ﻭ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ )‪(n + 2‬ﺍﻡ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪X‬‬ ‫‪Z‬‬
‫‪b‬‬ ‫‪n‬‬
‫)‪hn+3 f (n+2) (η‬‬ ‫‪n‬‬ ‫‬ ‫‬
‫‪f (x)dx −‬‬ ‫= ) ‪wi f (xi‬‬ ‫‪t2 (t − 1) . . . (t − n) dt‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫!)‪(n + 2‬‬ ‫‪0‬‬

‫‪12 Simpson’s‬‬ ‫‪3/8 Rule‬‬ ‫‪14 Truncation‬‬ ‫‪error‬‬

‫‪13 Boole’s‬‬ ‫‪rule‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲۰‬‬

‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﺳﺖ‪.‬‬

‫• ﺍﮔﺮ ‪ n ≥ 1‬ﻋﺪﺩﻱ ﻓﺮﺩ ﺑﺎﺷﺪ ﻭ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ )‪(n + 1‬ﺍﻡ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‪Z‬‬ ‫‪n‬‬
‫)‪hn+2 f (n+1) (η‬‬
‫‪f (x)dx −‬‬ ‫= ) ‪wi f (xi‬‬ ‫‪[t(t − 1) . . . (t − n)] dt‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫!)‪(n + 1‬‬ ‫‪0‬‬

‫ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻧﻴﺰ ﻗﻀﻴﻪ ﻣﺸﺎﺑﻪ ﺯﻳﺮ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪.‬‬

‫ﺧﻄﺎﯼ ﻓﺮﻣﻮﻝﻫﺎﯼ ﻧﯿﻮﺗﻦ⁃ﮐﺎﺗﺴ ﺑﺴﺘﻪ‬ ‫ﻗﻀﯿﻪ ‪۵−۹‬‬

‫ﺍﮔﺮ ‪ x0 , . . . , xn‬ﻧﻘﺎﻁ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﻨﺪ ﺁﻥﮔﺎﻩ‬

‫• ﺍﮔﺮ ‪ n ≥ 0‬ﻋﺪﺩﻱ ﺯﻭﺝ ﺑﺎﺷﺪ ﻭ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ )‪(n + 2‬ﺍﻡ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪X‬‬ ‫‪Z‬‬
‫‪b‬‬ ‫‪n‬‬
‫)‪hn+3 f (n+2) (η‬‬ ‫‪n+1‬‬ ‫‬ ‫‬
‫‪f (x)dx −‬‬ ‫= ) ‪wi f (xi‬‬ ‫‪t2 (t − 1) . . . (t − n) dt‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫!)‪(n + 2‬‬ ‫‪−1‬‬

‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺍﺳﺖ‪.‬‬

‫• ﺍﮔﺮ ‪ n ≥ 1‬ﻋﺪﺩﻱ ﻓﺮﺩ ﺑﺎﺷﺪ ﻭ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ )‪(n + 1‬ﺍﻡ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‪Z‬‬ ‫‪n+1‬‬
‫)‪hn+2 f (n+1) (η‬‬
‫‪f (x)dx −‬‬ ‫= ) ‪wi f (xi‬‬ ‫‪[t(t − 1) . . . (t − n)] dt‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫!)‪(n + 1‬‬ ‫‪−1‬‬

‫ﻃﺒﻖ ﻗﻀﺎﻳﺎﻱ ﻓﻮﻕ‪ ،‬ﺑﺎ ﻣﺤﺎﺳﺒﻪ ﺗﺤﻠﻴﻠﻲ ﻳﻚ ﺍﻧﺘﮕﺮﺍﻝ ﻣﻲﺗﻮﺍﻥ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ‬
‫ﻧﻤﻮﺩ‪ .‬ﻣﺜﺎﻝﻫﺎﻱ ﺯﻳﺮ ﺭﺍ ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﺪ‪.‬‬

‫■ ﺧﻄﺎﻱ ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ‬

‫ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﻫﻤﺎﻥ ﻓﺮﻣﻮﻝ ﺩﻭ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺍﺳﺖ‪ .‬ﺩﻭ ﻧﻘﻄﻪ ﺑﻪ ﺻﻮﺭﺕ ‪ x0 = a‬ﻭ ‪ x1 = b‬ﺍﺳﺖ‪ .‬ﺗﻮﺟﻪ‬
‫ﺷﻮﺩ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻓﺮﻣﻮﻝ ﺩﻭﻧﻘﻄﻪﺍﻱ ﺩﺍﺭﻳﻢ‪ .n = 1 :‬ﺣﺎﻝ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﻀﻴﻪ ‪ ۴-۹‬ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦﻛﻪ ‪ n‬ﻓﺮﺩ ﺍﺳﺖ‪ ،‬ﺟﻤﻠﻪ ﺧﻄﺎﻱ‬
‫ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪Z‬‬ ‫‪1‬‬
‫)‪h3 f (2) (η‬‬ ‫‪h3 ′′‬‬
‫‪[t(t − 1)] dt = −‬‬ ‫)‪f (η‬‬
‫!‪2‬‬ ‫‪0‬‬ ‫‪12‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ ،h = b − a‬ﻟﺬﺍ ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻗﺎﺑﻞ ﺑﻴﺎﻥ ﺍﺳﺖ‪:‬‬
‫‪(b − a)3 ′′‬‬
‫‪−‬‬ ‫)‪f (η‬‬
‫‪12‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۲۱‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫■ ﺧﻄﺎﻱ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ‬

‫ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﻫﻤﺎﻥ ﻓﺮﻣﻮﻝ ﺳﻪ ﻧﻘﻄﻪﺍﻱ )‪ (n = 2‬ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﻀﻴﻪ ‪ ۴-۹‬ﻣﻲﺗﻮﺍﻥ‬
‫ﺟﻤﻠﻪ ﺧﻄﺎ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻮﺷﺖ‪:‬‬
‫‪Z‬‬
‫)‪h5 f (4) (η‬‬ ‫‪2‬‬ ‫‬ ‫‬ ‫‪h5‬‬
‫)‪t2 (t − 1)(t − 2) dt = − f (4) (η‬‬
‫!‪4‬‬ ‫‪0‬‬ ‫‪90‬‬
‫= ‪ ،h‬ﻟﺬﺍ ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻴﺰ ﻗﺎﺑﻞ ﺑﻴﺎﻥ ﺍﺳﺖ‪:‬‬ ‫‪b−a‬‬
‫‪2‬‬ ‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ‬
‫)‪(b − a) (4‬‬
‫‪5‬‬
‫‪−‬‬ ‫)‪f (η‬‬
‫‪2880‬‬

‫■ ﺧﻄﺎﻱ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‬

‫ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻓﺮﻣﻮﻝ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﺑﺎ ‪ n = 0‬ﺍﺳﺖ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﻀﻴﻪ ‪ ۵-۹‬ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬
‫)‪h3 f (2) (η‬‬ ‫‪1‬‬ ‫‪ 2‬‬ ‫‪h3 ′′‬‬ ‫‪(b − a)3 ′′‬‬
‫= ‪t dt‬‬ ‫= )‪f (η‬‬ ‫)‪f (η‬‬
‫!)‪(2‬‬ ‫‪−1‬‬ ‫‪3‬‬ ‫‪24‬‬

‫■ ﺟﺪﻭﻝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ‬

‫ﺭﻭﺵ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ﺑﺎ ﭼﻨﺪ ﻣﺜﺎﻝ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﻧﺤﻮﻩ ﺍﺳﺘﺨﺮﺍﺝ‬
‫ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝﻫﺎ ﺑﺎ ﻛﻤﻚ ﺩﻭ ﻗﻀﻴﻪ ﺗﻮﺿﻴﺢ ﺩﺍﺩﻩ ﺷﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻲﺗﻮﺍﻥ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﻭ ﺑﺴﺘﻪ ﺑﺎ‬
‫ﻫﺮ ﺗﻌﺪﺍﺩ ﻧﻘﻄﻪ ﺭﺍ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺍﻳﻦ ﻛﺎﺭ ﻭﻗﺖﮔﻴﺮ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﺩﺭ ﺍﺩﺍﻣﻪ ﺟﺪﻭﻝﻫﺎﻳﻲ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺩﺭ ﺁﻥﻫﺎ ﻓﺮﻣﻮﻝﻫﺎﻱ‬
‫ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ ﺑﻪ ﻫﻤﺮﺍﻩ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺁﻥﻫﺎ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺎﺯ‬

‫ﻧﺎﻡ ﻓﺮﻣﻮﻝ‬ ‫ﻓﺮﻣﻮﻝ‬ ‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ‬

‫‪R x1‬‬
‫ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‬ ‫‪x−1‬‬
‫]) ‪f (x)dx = 2h [f (x0‬‬ ‫‪h3 ′′‬‬
‫)‪3 f (η‬‬
‫‪Rb‬‬ ‫‬ ‫‬ ‫‪(b−a)3 ′′‬‬
‫‪ ۱‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫‪f (x)dx = (b − a) f ( a+b‬‬
‫) ‪2‬‬ ‫)‪24 f (η‬‬
‫‪R x2‬‬ ‫‪3h‬‬ ‫‪3h3 ′′‬‬
‫‪x−1‬‬
‫= ‪f (x)dx‬‬ ‫‪2‬‬ ‫]) ‪[f (x0 ) + f (x1‬‬ ‫)‪4 f (η‬‬
‫‪Rb‬‬ ‫‬ ‫‬ ‫‪(b−a)3 ′′‬‬
‫‪ ۲‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪2‬‬ ‫‪f ( 2a+b‬‬ ‫‪a+2b‬‬
‫) ‪3 ) + f( 3‬‬ ‫)‪36 f (η‬‬

‫‪R x3‬‬ ‫)‪14h5 (4‬‬

‫‪x−1‬‬
‫= ‪f (x)dx‬‬ ‫‪4h‬‬
‫‪3‬‬ ‫]) ‪[2f (x0 ) − f (x1 ) + 2f (x2‬‬ ‫‪45 f‬‬ ‫)‪(η‬‬
‫‪Rb‬‬ ‫‬ ‫‬ ‫)‪7(b−a)5 (4‬‬
‫‪ ۳‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪3‬‬ ‫) ‪4 ) − f( 2 ) + f( 4‬‬
‫‪2f ( 3a+b‬‬ ‫‪a+b‬‬ ‫‪a+3b‬‬
‫‪5760 f‬‬ ‫)‪(η‬‬
‫‪R x4‬‬ ‫‪5h‬‬ ‫)‪95h5 (4‬‬
‫‪x−1‬‬
‫= ‪f (x)dx‬‬ ‫‪24‬‬ ‫]) ‪[11f (x0 ) + f (x1 ) + f (x2 ) + 11f (x3‬‬ ‫‪144 f‬‬ ‫)‪(η‬‬
‫‪Rb‬‬ ‫‬ ‫)‪ 19(b−a)5 (4‬‬
‫‪ ۴‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪24‬‬ ‫‪11f ( 4a+b‬‬ ‫‪3a+2b‬‬ ‫‪2a+3b‬‬ ‫‪a+4b‬‬
‫) ‪5 ) + f ( 5 ) + f ( 5 ) + 11f ( 5‬‬ ‫‪90000 f‬‬ ‫)‪(η‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲۲‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ‬

‫ﻧﺎﻡ ﻓﺮﻣﻮﻝ‬ ‫ﻓﺮﻣﻮﻝ‬ ‫ﺧﻄﺎﻱ ﺑﺮﺷﻲ‬

‫‪R x1‬‬
‫ﺫﻭﺯﻧﻘﻪ‬
‫‪3‬‬

‫‪x0‬‬
‫= ‪f (x)dx‬‬ ‫‪h‬‬
‫‪2‬‬ ‫]) ‪[f (x0 ) + f (x1‬‬ ‫)‪− h12 f ′′ (η‬‬
‫‪Rb‬‬ ‫‪3‬‬
‫‪ ۲‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪2‬‬ ‫])‪[f (a) + f (b‬‬ ‫)‪− (b−a‬‬ ‫‪′′‬‬
‫)‪12 f (η‬‬
‫‪R x2‬‬
‫ﺳﻴﻤﭙﺴﻮﻥ‬
‫‪5‬‬

‫‪x0‬‬
‫= ‪f (x)dx‬‬ ‫‪h‬‬
‫‪3‬‬ ‫]) ‪[f (x0 ) + 4f (x1 ) + f (x2‬‬ ‫)‪− h90 f (4) (η‬‬
‫‪Rb‬‬ ‫‬ ‫‬ ‫‪5‬‬
‫‪ ۳‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪6‬‬ ‫‪f (a) + 4f ( a+b‬‬
‫)‪2 ) + f (b‬‬ ‫)‪− (b−a‬‬
‫‪2880 f‬‬
‫)‪(4‬‬
‫)‪(η‬‬
‫‪R x3‬‬
‫‪ 38‬ﺳﻴﻤﭙﺴﻮﻥ‬
‫‪5‬‬

‫‪x0‬‬
‫= ‪f (x)dx‬‬ ‫‪3h‬‬
‫‪8‬‬ ‫]) ‪[f (x0 ) + 3f (x1 ) + 3f (x2 ) + f (x3‬‬ ‫‪− 3h‬‬
‫‪80 f‬‬
‫)‪(4‬‬
‫)‪(η‬‬
‫‪Rb‬‬ ‫‬ ‫‬ ‫‪3‬‬
‫‪ ۴‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪8‬‬‫‪f (a) + 3f ( 2a+b‬‬ ‫‪a+2b‬‬
‫)‪3 ) + 3f ( 3 ) + f (b‬‬ ‫)‪− (b−a‬‬
‫‪6480 f‬‬
‫)‪(4‬‬
‫)‪(η‬‬
‫‬ ‫‬
‫‪R x4‬‬ ‫‪7f (x0 ) + 32f (x1 ) + 12f (x2 )+‬‬
‫ﺑﻮﻝ‬
‫‪7‬‬

‫‪x0‬‬
‫‪f‬‬ ‫‪(x)dx‬‬ ‫=‬ ‫‪2h‬‬
‫‪45‬‬ ‫‪− 8h‬‬
‫‪945 f‬‬
‫)‪(6‬‬
‫)‪(η‬‬
‫) ‪32f (x3 ) + 7f (x4‬‬
‫‪ ۵‬ﻧﻘﻄﻪﺍﻱ‬ ‫‬ ‫‬
‫‪Rb‬‬ ‫‪7f (a) + 32f ( 3a+b‬‬ ‫‪a+b‬‬
‫‪4 ) + 12f ( 2 )+‬‬
‫‪7‬‬
‫)‪(b−a‬‬
‫= ‪f (x)dx‬‬ ‫‪b−a‬‬
‫‪− 1935360‬‬ ‫)‪f (6) (η‬‬
‫‪a‬‬ ‫‪90‬‬ ‫‪32f ( a+3b‬‬
‫)‪4 ) + 7f (b‬‬
‫‪‬‬ ‫‪‬‬
‫‪R x6‬‬ ‫‪41f‬‬ ‫‪(x‬‬ ‫‪0 ) + 216f (x1 )+‬‬
‫‪h ‬‬ ‫‪‬‬
‫ﻭﺩﻝ‬
‫‪9‬‬

‫‪x0‬‬
‫‪f (x)dx = 140‬‬ ‫‪27f (x2 ) + 272f (x3 )+‬‬ ‫‪− 1400‬‬
‫‪9h‬‬
‫)‪f (8) (η‬‬
‫) ‪27f (x4 ) + 216f (x5 ) + 41f (x6‬‬
‫‪ ۷‬ﻧﻘﻄﻪﺍﻱ‬ ‫‪‬‬ ‫‪‬‬
‫‪Rb‬‬ ‫‪41f (a) + 216f ( 5a+b‬‬
‫‪6 )+‬‬
‫‪ 27f ( 4a+2b ) + 272f ( 3a+3b )+‬‬ ‫‪‬‬
‫‪9‬‬
‫)‪(b−a‬‬
‫‪a‬‬
‫‪f (x)dx = b−a‬‬
‫‪700‬‬ ‫‪6‬‬ ‫‪6‬‬ ‫‪− 15676416‬‬ ‫)‪f (8) (η‬‬
‫‪4a+2b‬‬ ‫‪5a+b‬‬
‫)‪27f ( 6 ) + f ( 6 ) + 41f (b‬‬

‫ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬ ‫‪۲.۹‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺫﻛﺮ ﺷﺪ‪ ،‬ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﻓﺮﻡ ﻛﻠﻲ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬
‫‪n‬‬
‫≃ ‪f (x)dx‬‬ ‫‪ωi f (xi ).‬‬ ‫)‪(۸.۹‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫ﻏﺎﻟﺒﺎ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ xi‬ﺩﺭ ﺍﺑﺘﺪﺍ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﻧﺪ ﻭ ﺳﭙﺲ ﻭﺯﻥﻫﺎﻱ ‪ ωi‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﻧﺪ‪ .‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ‬
‫‪R‬‬
‫ﺩﻳﺪﻳﻢ‪ ،‬ﺍﻳﻦ ﺿﺮﺍﻳﺐ ﺑﻪ ﺻﻮﺭﺕ ‪ ωi = ab ℓi (x)dx‬ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﻧﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺭﻭﺵ ﺩﻳﮕﺮﻱ ﻣﻮﺳﻮﻡ ﺑﻪ ﺭﻭﺵ ﺿﺮﺍﻳﺐ‬
‫ﻻﮔﺮﺍﻧﮋ ﺑﺮﺍﻱ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩﻥ ﺿﺮﺍﻳﺐ ﻻﮔﺮﺍﻧﮋ ﺍﺭﺍﻳﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺍﺿﺎﻓﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﻭ ﺭﻭﺵ ﻻﮔﺮﺍﻧﮋ ﻛﺎﻣﻼ‬
‫ﻣﻌﺎﺩﻝ ﻳﻚﺩﻳﮕﺮ ﻣﻲﺑﺎﺷﻨﺪ ﻭ ﻫﺮ ﺩﻭ ﺭﻭﺵ ﻣﻘﺎﺩﻳﺮ ﻳﻜﺴﺎﻧﻲ ﺑﺮﺍﻱ ﻭﺯﻥﻫﺎﻱ ‪ ωi‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻭﺭﻧﺪ‪ ،‬ﺍﻣﺎ ﺭﻭﺵ ﻣﺤﺎﺳﺒﻪ ﺁﻥﻫﺎ ﻣﺘﻔﺎﻭﺕ‬
‫ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫ﺩﺭ ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ‪ ،‬ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﻭﺯﻥﻫﺎﻱ ‪ ωi‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻋﻤﻞ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﻪ ﺟﺎﻱ ‪ f‬ﺩﺭ ﺩﻭ ﻃﺮﻑ ﻓﺮﻣﻮﻝ‬
‫)‪ (۸.۹‬ﺗﻮﺍﺑﻊ‬
‫‪1, x, x2 , . . . , xn‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۲۳‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﻗﺮﺍﺭ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪‬‬ ‫‪Z‬‬
‫‪‬‬
‫‪‬‬
‫‪b‬‬
‫‪f (x) := 1‬‬
‫‪‬‬ ‫⇒=‬ ‫)‪1dx = ω0 (1) + ω1 (1) + · · · + ωn (1‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪Z‬‬
‫‪a‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪b‬‬
‫‪f (x) := x‬‬ ‫⇒=‬ ‫) ‪xdx = ω0 (x0 ) + ω1 (x1 ) + · · · + ωn (xn‬‬
‫‪a‬‬
‫‪‬‬
‫‪‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪Z‬‬
‫‪‬‬
‫‪‬‬ ‫‪b‬‬
‫‪‬‬
‫‪f (x) := xn‬‬
‫‪‬‬ ‫⇒=‬ ‫) ‪xn dx = ω0 (xn0 ) + ω1 (xn1 ) + · · · + ωn (xnn‬‬
‫‪a‬‬
‫‪Rb‬‬
‫‪.‬‬ ‫‪a‬‬
‫= ‪xk dx‬‬ ‫‪1‬‬
‫‪k+1 (b‬‬
‫‪k+1‬‬
‫ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ﻇﺎﻫﺮ ﺷﺪﻩ ﺩﺭ ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﺳﺎﺩﮔﻲ ﻣﻲﺗﻮﺍﻥ ﻣﺤﺎﺳﺒﻪ ﻧﻤﻮﺩ‪ .‬ﺩﺍﺭﻳﻢ‪− ak+1 ) :‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻣﻲﺗﻮﺍﻥ ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﺎﺯﻧﻮﻳﺴﻲ ﻧﻤﻮﺩ‪:‬‬
‫‪‬‬
‫‪ ω0 + ω1 + · · · + ωn = b − a‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫
‬
‫‪‬‬
‫‪‬‬ ‫‪1 2‬‬
‫‪‬‬
‫‪ω0 (x0 ) + ω1 (x1 ) + · · · + ωn (xn ) = 2 b − a‬‬
‫‪2‬‬

‫‪..‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫)‪(۹.۹‬‬

‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬
‫‪‬‬ ‫
‬
‫‪ω (xn ) + ω (xn ) + · · · + ω (xn ) = 1‬‬ ‫‪bn+1 − an+1‬‬
‫‪0 0‬‬ ‫‪1 1‬‬ ‫‪n n‬‬
‫‪n+1‬‬
‫ﺭﻭﺍﺑﻂ ﻓﻮﻕ ﻳﻚ ﺩﺳﺘﮕﺎﻩ ﻣﻌﺎﺩﻻﺕ ﺧﻄﻲ )‪(n + 1‬ﺗﺎﻳﻲ ﺑﻪ ﻓﺮﻡ ‪ Aω = b‬ﻣﻲﺑﺎﺷﺪ‪ ،‬ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪‬‬
‫‪1‬‬ ‫‪1‬‬ ‫‪1 ... 1‬‬ ‫‪ω0‬‬ ‫‪b−a‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫
‬ ‫‪‬‬
‫‪ x0‬‬ ‫‪x1 x2 . . . x n ‬‬ ‫‪ ω1 ‬‬ ‫‪‬‬ ‫‪1‬‬
‫‪b2 − a 2‬‬ ‫‪‬‬
‫‪‬‬ ‫‪‬‬ ‫‪ ‬‬ ‫‪‬‬ ‫‪2‬‬
‫
‬ ‫‪‬‬
‫‪ 2‬‬ ‫‪2‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪3 b −a‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪1‬‬ ‫‪3‬‬ ‫‪3‬‬
‫‪A :=  x0‬‬ ‫‪x1 x2 . . . x n ‬‬ ‫‪ω :=  ω2 ‬‬ ‫‪b := ‬‬ ‫‪‬‬
‫‪ .‬‬ ‫‪..‬‬ ‫‪..‬‬ ‫‪.. ‬‬‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬ ‫‪‬‬
‫‪ .‬‬ ‫‪ .. ‬‬ ‫‪‬‬ ‫‪..‬‬ ‫‪‬‬
‫‪ .‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪.‬‬ ‫‪ . ‬‬ ‫‪‬‬ ‫‪.‬‬ ‫‪‬‬
‫
‬
‫‪xn0‬‬ ‫‪n‬‬ ‫‪n‬‬
‫‪x1 x2 . . . x n‬‬ ‫‪n‬‬
‫‪ωn‬‬ ‫‪1‬‬
‫‪n+1 b‬‬
‫‪n+1‬‬
‫‪−a‬‬ ‫‪n+1‬‬

‫ﻣﺎﺗﺮﻳﺲ ﺿﺮﺍﻳﺐ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ‪ ،‬ﻣﺎﺗﺮﻳﺲ ﻭﺍﻧﺪﺭﻣﻮﻧﺪ ﻣﻲﺑﺎﺷﺪ ﻭ ﭼﻮﻥ ‪ xi‬ﻫﺎ ﻣﺠﺰﺍ ﻣﻲﺑﺎﺷﻨﺪ‪ ،‬ﻟﺬﺍ ﺍﻳﻦ ﻣﺎﺗﺮﻳﺲ ﻭﺍﺭﻭﻥﭘﺬﻳﺮ ﺍﺳﺖ ﻭ‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻭﺯﻥﻫﺎﻱ ‪ ωi‬ﺍﺯ ﺣﻞ ﺩﺳﺘﮕﺎﻩ ﻓﻮﻕ ﺑﻪ ﻃﻮﺭ ﻣﻨﺤﺼﺮ ﺑﻪ ﻓﺮﺩ ﺍﺳﺘﺨﺮﺍﺝ ﻣﻲﺷﻮﻧﺪ‪.‬‬

‫ﻣﺮﺗﺒﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‬ ‫‪۳.۹‬‬

‫ﻣﺮﺗﺒﻪ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‪ ،‬ﺷﺎﺧﺼﻲ ﺍﺳﺖ ﻛﻪ ﺑﺮﺍﻱ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﻛﺎﺭ ﻣﻲﺭﻭﺩ‪ .‬ﻣﺮﺗﺒﻪ‬
‫ﺩﻗﺖ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺩﺭ ﺯﻳﺮ ﺗﻌﺮﻳﻒ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﯾ‬ ‫ﺗﻌﺮﯾﻒ ‪۶−۹‬‬

‫ﮔﻮﻳﻴﻢ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﺮﺍﺑﺮ ‪ k‬ﺍﺳﺖ‪ ،‬ﻫﺮﮔﺎﻩ ﺁﻥ ﻓﺮﻣﻮﻝ ﺑﺮﺍﻱ ﺗﻤﺎﻣﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺍﺯ ﺩﺭﺟﻪ‬
‫ﺣﺪﺍﻛﺜﺮ ‪ k‬ﺩﻗﻴﻖ ﺑﺎﺷﺪ‪ ،‬ﻭﻟﻲ ﺑﺮﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺍﺯ ﺩﺭﺟﻪ ‪ k + 1‬ﺩﻗﻴﻖ ﻧﺒﺎﺷﺪ‪.‬‬

‫ﻧﺘﻴﺠﻪ ﺯﻳﺮ ﺭﺍ ﺑﻼﻓﺎﺻﻠﻪ ﻣﻲﺗﻮﺍﻥ ﮔﺮﻓﺖ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲۴‬‬

‫ﻧﺘﯿﺠﻪ ‪۷−۹‬‬
‫ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﺮﺍﺑﺮ ‪ k‬ﺍﺳﺖ‪ ،‬ﻫﺮﮔﺎﻩ ﺁﻥ ﻓﺮﻣﻮﻝ ﺑﻪ ﺍﺯﺍﻱ‬
‫‪f (x) = 1, f (x) = x, f (x) = x2 , . . . , f (x) = xn‬‬
‫ﺩﻗﻴﻖ ﺑﺎﺷﺪ‪ ،‬ﺍﻣﺎ ﺑﺎ ﺍﺯﺍﻱ ‪ f (x) = xn+1‬ﺩﻗﻴﻖ ﻧﺒﺎﺷﺪ‪.‬‬

‫ﺑﺮﺍﻱ ﻣﺸﺨﺺ ﻛﺮﺩﻥ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‪ ،‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺗﻌﺮﻳﻒ ﻭ ﻳﺎ ﻓﺮﻣﻮﻝ ﺧﻄﺎﻱ ﺑﺮﺭﺷﻲ ﺍﺳﺘﻔﺎﺩﻩ‬
‫ﻛﺮﺩ‪.‬‬

‫■ ﺗﻌﻴﻴﻦ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺗﻌﺮﻳﻒ‬

‫ﺩﺭ ﺍﻳﻦ ﺭﻭﺵ ﺍﺯ ﻧﺘﻴﺠﻪ )‪ (۷-۹‬ﺑﺮﺍﻱ ﻳﺎﻓﺘﻦ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻛﻤﻚ ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺑﻪ ﺗﺮﺗﻴﺐ ﺑﻪ ﺟﺎﻱ ‪ f‬ﺩﺭ ﻓﺮﻣﻮﻝ‬
‫)‪ (۸.۹‬ﺑﻪ ﺗﺮﺗﻴﺐ ﺗﻮﺍﺑﻊ ‪ x2 ،x ،1‬ﻭ ‪ ...‬ﻗﺮﺍﺭ ﺩﺍﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﮔﺮ ‪ k‬ﻛﻮﭼﻚﺗﺮﻳﻦ ﻋﺪﺩ ﺻﺤﻴﺢ ﻧﺎﻣﻨﻔﻲ ﺑﺎﺷﺪ ﻛﻪ ﻗﺎﻋﺪﻩ )‪ (۸.۹‬ﺑﻪ ﺍﺯﺍﻱ‬
‫‪ f (x) = xk+1‬ﺩﺭﺳﺖ ﻧﺒﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻓﺮﻣﻮﻝ ﺩﻗﻴﻘﺎ ﺑﺮﺍﺑﺮ ‪ k‬ﺍﺳﺖ‪.‬‬

‫ﻣﺜﺎﻝ ‪)۱−۹‬ﺗﻌﯿﯿﻦ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﯿﺎﻧ (‬

‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺭﺍ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﻳﺎﺩﺁﻭﺭﻱ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﻗﺎﻋﺪﻩ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫ ‬ ‫‬
‫‪a+b‬‬
‫‪f (x)dx ≃ (b − a) f‬‬ ‫‪.‬‬
‫‪a‬‬ ‫‪2‬‬
‫ﺍﮔﺮ ‪ f (x) := 1‬ﺁﻥﮔﺎﻩ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺮﺍﺑﺮ ‪ b − a‬ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﻣﻘﺪﺍﺭ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‬
‫ﺑﺎ ‪ .(b − a)[1] = b − a‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﻘﺪﺍﺭ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺗﻮﺳﻂ ﺭﻭﺵ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﺎ ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺑﺮﺍﺑﺮ ﺍﺳﺖ‪ .‬ﻟﺬﺍ‬
‫ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺮﺗﺒﻪ ﻫﻤﮕﺮﺍﻳﻲ ﺭﻭﺵ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺣﺪﺍﻗﻞ ﺑﺮﺍﺑﺮ ‪ 0‬ﺍﺳﺖ‪.‬‬
‫ﺍﻛﻨﻮﻥ ﺗﺎﺑﻊ ‪ f (x) := x‬ﺭﺍ ﺩﺭ ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻧﺘﮕﺮﺍﻝ ﺑﻪ ﺍﺯﺍﻱ ﺍﻳﻦ ﺗﺎﺑﻊ ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ ) ‪ . 2 (b − a‬ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ‬
‫‪1 2‬‬ ‫‪2‬‬
‫‬ ‫‬
‫‪ (b − a) b−a‬ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪ .‬ﺑﻨﺎﺑﺮﺍﻳﻦ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‬ ‫‪2‬‬ ‫ﺍﺯ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻣﻘﺪﺍﺭ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺮﺍﺑﺮ ) ‪= 12 (b2 − a2‬‬
‫ﺑﺮﺍﻱ ‪ f (x) := x‬ﻧﻴﺰ ﺩﻗﻴﻖ ﺍﺳﺖ ﻭ ﻟﺬﺍ ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﺭﻭﺵ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺣﺪﺍﻗﻞ ‪ 1‬ﺍﺳﺖ‪.‬‬
‫ﻧﻈﺮ ﻣﻲﮔﻴﺮﻳﻢ‪ .‬ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺮﺍﺑﺮ ) ‪ . 13 (b3 − a3‬ﺍﺳﺖ ﻭ ﻣﻘﺪﺍﺭ ﺑﻪ ﺩﺳﺖ ﺁﻣﺪﻩ ﺍﺯ ﻗﺎﻋﺪﻩ‬ ‫ﺣﺎﻝ ‪
f2 i(x) := x‬ﺭﺍ ﺩﺭ ‪h‬‬
‫‪2‬‬

‫‪ (b − a) b−a‬ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺍﻳﻦ ﺩﻭ ﻣﻘﺪﺍﺭ ﻫﻤﻮﺍﺭﻩ ﺑﺮﻗﺮﺍﺭ ﻧﻤﻲﺑﺎﺷﻨﺪ‪ ،‬ﻳﻌﻨﻲ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ‬ ‫‪2‬‬ ‫ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﺮﺍﺑﺮ‬
‫ﻣﻴﺎﻧﻲ ﺑﺮﺍﻱ ‪ f (x) := x2‬ﺩﻗﻴﻖ ﻧﻤﻲﺑﺎﺷﺪ‪ .‬ﻟﺬﺍ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺩﻗﻴﻘﺎ ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ‪.‬‬

‫■ ﺗﻌﻴﻴﻦ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﻓﺮﻣﻮﻝ ﺧﻄﺎﻱ ﺑﺮﺷﻲ‬

‫ﺍﮔﺮ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺭﺍ ﺩﺍﺷﺘﻪ ﺑﺎﺷﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﺑﻪ ﺭﺍﺣﺘﻲ ﻣﻲﺗﻮﺍﻥ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻓﺮﻣﻮﻝ ﺭﺍ‬
‫ﻣﺸﺨﺺ ﻛﺮﺩ‪ .‬ﺑﻪ ﺍﻳﻦ ﺻﻮﺭﺕ ﻛﻪ ﺍﮔﺮ ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‪ ،‬ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ ‪ k + 1‬ﻇﺎﻫﺮ ﺷﺪ‪،‬‬
‫ﺁﻥﮔﺎﻩ ﺑﺪﻳﻬﻲ ﺍﺳﺖ ﻛﻪ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﺍﺯﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ k‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ ﺁﻥ ﻓﺮﻣﻮﻝ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﺍﺯﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺍﺯ ﺩﺭﺟﻪ ﺣﺪﺍﻛﺜﺮ ‪ k‬ﺩﻗﻴﻖ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ‬
‫ﺑﺮﺍﺑﺮ ﺍﺳﺖ ﺑﺎ ‪.k‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻣﺜﺎﻝ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻋﺒﺎﺭﺕ ﺍﺳﺖ ﺍﺯ‪ . 3 f (η) :‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻣﺸﺘﻖ ﻣﺮﺗﺒﻪ ﺩﻭ ﺗﺎﺑﻊ‬
‫‪h3 ′′‬‬

‫ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻇﺎﻫﺮ ﺷﺪﻩ‪ ،‬ﺑﻨﺎﺑﺮﻳﻦ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﻪ ﺍﺯﺍﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺣﺪﺍﻛﺜﺮ ﺍﺯ ﺩﺭﺟﻪ ‪ 1‬ﺑﺮﺍﺑﺮ ﺻﻔﺮ ﺍﺳﺖ‪ .‬ﻟﺬﺍ ﻣﺮﺗﺒﻪ‬
‫ﺩﻗﺖ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﺮﺍﺑﺮ ‪ 1‬ﺍﺳﺖ‪.‬‬
‫ﺩﺭ ﺍﻧﺘﻬﺎﻱ ﺍﻳﻦ ﺑﺨﺶ‪ ،‬ﺍﺭﺗﺒﺎﻁ ﺑﻴﻦ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺩﺭ ﻓﺮﻣﻮﻝﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻭ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﺁﻥ ﻓﺮﻣﻮﻝ ﺩﺭ ﻗﻀﻴﻪ ﺯﻳﺮ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۲۵‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﻣﺸﺨﺺ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻗﻀﯿﻪ ‪ ۸−۹‬ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﯾ ﻗﺎﻋﺪﻩ ‪ n‬ﻧﻘﻄەﺍﯼ‬
‫ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ‪ n‬ﻧﻘﻄﻪﺍﻱ‪ ،‬ﺑﺰﺭﮒﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ n − 1‬ﻭ ﻛﻮﭼﻚﺗﺮ ﻳﺎ ﻣﺴﺎﻭﻱ ‪ 2n − 1‬ﺍﺳﺖ‪.‬‬
‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻗﺎﻋﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺒﺘﻨﻲ ﺑﺮ ﺩﺭﻭﻧﻴﺎﺑﻲ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﺣﺪﺍﻗﻞ ﺑﺮﺍﺑﺮ ‪ n − 1‬ﻭ ﺣﺪﺍﻛﺜﺮ‬
‫ﺑﺮﺍﺑﺮ ‪ 2n − 1‬ﺍﺳﺖ‪.‬‬

‫∗ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ‬ ‫‪۴.۹‬‬

‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﻪ ﻋﻨﻮﺍﻥ ﻳﻚ ﺷﺎﺧﺺ ﺑﺮﺍﻱ ﺩﻗﺖ ﺁﻥ ﻓﺮﻣﻮﻝ ﺩﺭ ﻧﻈﺮ‬
‫ﮔﺮﻓﺘﻪ ﻣﻲﺷﻮﺩ ﻭ ﻫﻤﺎﻥ ﻃﻮﺭ ﻛﻪ ﺩﺭ ﻗﻀﻴﻪ ‪ ۸-۹‬ﺍﺷﺎﺭﻩ ﺷﺪ‪ ،‬ﺣﺪﺍﻛﺜﺮ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻳﻚ ﻓﺮﻣﻮﻝ ‪ n‬ﻧﻘﻄﻪﺍﻱ ﺑﺮﺍﺑﺮ ‪ 2n − 1‬ﺍﺳﺖ‪.‬‬
‫ﺣﺎﻝ ﺳﻮﺍﻟﻲ ﻛﻪ ﻣﻄﺮﺡ ﻣﻲﺷﻮﺩ ﺁﻥ ﺍﺳﺖ ﻛﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺑﺎ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻣﺎﻛﺰﻳﻤﻢ ﭼﻪ ﻣﻲﺑﺎﺷﻨﺪ؟ ﭘﺎﺳﺦ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﮔﻮﺳﻲ ﻣﻲﺑﺎﺷﻨﺪ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﺑﺨﺶ ﺑﻪ ﺁﻥﻫﺎ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬
‫‪۱۵‬‬
‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﺑﺎ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺩﺭ ﺍﺭﺗﺒﺎﻁ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﻟﺬﺍ ﻗﺒﻞ ﺍﺯ ﺍﺭﺍﻳﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﮔﻮﺳﻲ‪ ،‬ﺑﻪ ﻣﻌﺮﻓﻲ ﻣﺨﺘﺼﺮ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﻣﻲﭘﺮﺩﺍﺯﻳﻢ‪.‬‬

‫■ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ‬
‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺗﻮﺍﺑﻌﻲ ﺧﺎﺹ ﻭ ﺷﻨﺎﺧﺘﻪ ﺷﺪﻩﺍﻱ ﺩﺭ ﻳﺎﺿﻴﺎﺕ ﻭ ﻓﻴﺰﻳﻚ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺛﺎﺑﺖ ﻣﻲﺷﻮﺩ ﻛﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻟﮋﺍﻧﺪﺭ‬
‫ﺩﺭﺟﻪ ‪ n‬ﺩﺍﺭﺍﻱ ‪ n‬ﺭﻳﺸﻪ ﻣﺘﻤﺎﻳﺰ ﺩﺭ ﺩﺍﺧﻞ ﺑﺎﺯﻩ ]‪ [−1, 1‬ﻣﻲﺑﺎﺷﺪ‪ .‬ﻫﻤﭽﻨﻴﻦ ﺍﻳﻦ ﺭﻳﺸﻪﻫﺎ ﻣﺘﻘﺎﺭﻥ ﻣﻲﺑﺎﺷﻨﺪ‪ .‬ﺑﻪ ﺍﻳﻦ ﻣﻌﻨﻲ ﻛﻪ ﺍﮔﺮ‬
‫‪ ξ‬ﻳﻚ ﺭﻳﺸﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ‪ −ξ‬ﻧﻴﺰ ﻳﻚ ﺭﻳﺸﻪ ﺍﺳﺖ‪.‬‬
‫ﺑﺮﺍﻱ ﻧﻤﻮﻧﻪ‪ ،‬ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ﻳﻚ ﺍﻟﻲ ﭼﻬﺎﺭ ﻭ ﻫﻤﻴﻨﻄﻮﺭ ﺭﻳﺸﻪﻫﺎﻱ ﺁﻥﻫﺎ ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﺁﻭﺭﺩﻩ ﺷﺪﻩﺍﻧﺪ‪.‬‬

‫ﺭﻳﺸﻪﻫﺎﻱ‬ ‫ﺿﺎﺑﻄﻪ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ‬

‫‪،ξ0 := 0‬‬ ‫‪L1 (x) := x‬‬ ‫ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ‪۱‬‬
‫√‬ ‫√‬
‫‪ξ0 := −‬‬ ‫‪3‬‬
‫‪3 ,‬‬ ‫‪ξ1 := +‬‬ ‫‪3‬‬
‫)‪L2 (x) := 12 (3x2 − 1‬‬ ‫ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ‪۲‬‬
‫‪q‬‬ ‫‪3‬‬ ‫‪q‬‬
‫‪ξ0 := − 35 , ξ1‬‬ ‫=‪:= 0, ξ2 :‬‬ ‫‪3‬‬
‫)‪L3 (x) := 12 (5x3 − 3x‬‬ ‫ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ‪۳‬‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪5‬‬
‫‪q‬‬
‫‪ξ0 := − 37 + 72 65 ,‬‬ ‫‪ξ1 := −‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪−‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5,‬‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪q‬‬ ‫)‪L4 (x) := 18 (35x4 − 30x2 + 3‬‬ ‫ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ‪۴‬‬
‫‪ξ2 := 37 − 27 65 ,‬‬ ‫=‪ξ3 :‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪+‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5‬‬

‫ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺩﺭﺟﻪ ﺑﺎﻻﺗﺮ ﻟﮋﺍﻧﺪﺭ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻓﺮﻣﻮﻝ ﺑﺎﺯﮔﺸﺘﻲ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﺩ‪:‬‬
‫‪2n + 1‬‬ ‫‪n‬‬
‫=‪Ln+1 (x) :‬‬ ‫‪xLn (x) −‬‬ ‫)‪Ln−1 (x‬‬
‫‪n+1‬‬ ‫‪n+1‬‬
‫ﺍﻟﺒﺘﻪ ﻻﺯﻡ ﺑﻪ ﺫﻛﺮ ﺍﺳﺖ ﻛﻪ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺍﺯ ﻣﺮﺗﺒﻪ ﭘﺎﻳﻴﻦ ﺭﺍ‪ ،‬ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﺭ ﺟﺪﻭﻝ ﻓﻮﻕ ﻣﺸﺎﻫﺪﻩ ﻣﻲﺷﻮﺩ‪،‬‬
‫ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻡ ﺑﺴﺘﻪ ﺍﺳﺘﺨﺮﺍﺝ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﺍﺯ ﺩﺭﺟﻪ ﺑﺎﻻﺗﺮ ﺭﺍ ﻧﻤﻲﺗﻮﺍﻥ ﺑﻪ ﻓﺮﻡ ﺑﺴﺘﻪ ﺗﻮﺻﻴﻒ ﻛﺮﺩ‪ .‬ﺍﺯ‬
‫ﺍﻳﻦ ﺭﻭ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺍﺯ ﺩﺭﺟﻪ ﺑﺎﻻ ﺑﺎ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﻣﺤﺎﺳﺒﻪ ﻣﻲﺷﻮﺩ‪ .‬ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺧﺎﺻﻲ ﺍﺑﺪﺍﻉ‬
‫ﺷﺪﻩ ﻛﻪ ﺑﺎ ﺳﺮﻋﺖ ﻭ ﺩﻗﺖ ﻣﻨﺎﺳﺒﻲ ﺑﻪ ﺍﺳﺘﺨﺮﺍﺝ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﻣﻲﭘﺮﺩﺍﺯﻧﺪ‪.‬‬
‫‪15 Legendre‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲۶‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ‬ ‫�‬

‫ﺩﺭ ﻓﺮﻣﻮﻝﻫﺎﻱ ﮔﻮﺳﻲ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ xi‬ﺍﺯ ﻗﺒﻞ ﺍﻧﺘﺨﺎﺏ ﻧﻤﻲﺷﻮﻧﺪ‪ ،‬ﺑﻠﻜﻪ ﻣﻘﺪﺍﺭ ﺁﻥﻫﺎ ﺑﻪ ﮔﻮﻧﻪﺍﻱ ﺍﻧﺘﺨﺎﺏ ﻣﻲﺷﻮﺩ ﻛﻪ ﺑﺎﻻﺗﺮﻳﻦ‬
‫ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﺣﺎﺻﻞ ﮔﺮﺩﺩ‪ .‬ﻗﻀﻴﻪ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬
‫ﻗﻀﯿﻪ ﮔﻮﺱ‬ ‫ﻗﻀﯿﻪ ‪۹−۹‬‬
‫‪R +1‬‬
‫ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﺯ ﻗﺎﻋﺪﻩ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﺯﻳﺮ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪:‬‬ ‫‪−1‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺍﻧﺘﮕﺮﺍﻝ ‪f (x)dx‬‬
‫‪Z‬‬ ‫‪+1‬‬
‫‪f (x)dx ≃ ω0 f (x0 ) + · · · + ωn f (xn ).‬‬
‫‪−1‬‬
‫ﺍﮔﺮ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ‪ xi‬ﺭﺍ ﺑﺮﺍﺑﺮ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻟﮋﺍﻧﺪﺭ ﺩﺭﺟﻪ ‪ n+1‬ﺩﺭ ﻧﻈﺮ ﺑﮕﻴﺮﻳﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﺑﺮﺍﺑﺮ ‪ 2n − 1‬ﺧﻮﺍﻫﺪ ﺑﻮﺩ‪.‬‬

‫ﻗﻀﻴﻪ ﻓﻮﻕ ﻣﻲﮔﻮﻳﺪ ﻛﻪ ﺁﻥ ﻧﻘﺎﻁ ﻃﻼﻳﻲ ﻛﻪ ﻣﺮﺗﺒﻪ ﺩﻗﺖ ﻣﺎﻛﺰﻳﻤﻢ ﺭﺍ ﻧﺘﻴﺠﻪ ﻣﻲﺩﻫﺪ‪ ،‬ﻫﻤﺎﻥ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱ ﻟﮋﺍﻧﺪﺭ‬
‫ﻫﺴﺘﻨﺪ‪.‬‬
‫ﺑﻌﺪ ﺍﺯ ﺍﻧﺘﺨﺎﺏ ﺭﻳﺸﻪﻫﺎﻱ ﭼﻨﺪﺟﻤﻠﻪﺍﻱﻫﺎﻱ ﻟﮋﺍﻧﺪﺭ ﺑﻪ ﻋﻨﻮﺍﻥ ﮔﺮﻩﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‪ ،‬ﻧﻮﺑﺖ ﺑﻪ ﻣﺤﺎﺳﺒﻪ ﻭﺯﻥﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﮔﻮﺳﻲ ﻣﻲﺭﺳﺪ‪ .‬ﺍﻳﻦ ﻛﺎﺭ ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺭﻭﺵﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ ﻭ ﻳﺎ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺍﻧﺠﺎﻡ ﺩﺍﺩ‪ .‬ﺩﺭ ﺟﺪﻭﻝ ﺯﻳﺮ ﻭﺯﻥﻫﺎﻱ ﻭ ﮔﺮﻩﻫﺎﻱ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﻳﻚ ﺗﺎ ‪ ۵‬ﻧﻘﻄﻪﺍﻱ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪.‬‬

‫ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬ ‫ﻭﺯﻥﻫﺎ‬

‫‪،ξ0 = 0‬‬ ‫‪ω0 = 2‬‬ ‫ﻳﻚﻧﻘﻄﻪﺍﻱ‬
‫√‬ ‫√‬
‫‪ξ0 = − 33 , ξ1 = + 33‬‬ ‫‪ω0 = ω1 = 1‬‬ ‫ﺩﻭﻧﻘﻄﻪﺍﻱ‬
‫‪q‬‬ ‫‪q‬‬
‫‪ξ0 = − 35 , ξ1 = 0, ξ2 = 35‬‬ ‫= ‪ω0 = ω2 = 59 , ω1‬‬ ‫‪8‬‬
‫ﺳﻪﻧﻘﻄﻪﺍﻱ‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪q‬‬
‫‪9‬‬
‫√‬
‫‪ξ0 := − 37 + 27 65 , ξ1 := − 37 − 27 65 ,‬‬ ‫= ‪ω0 = ω3‬‬ ‫‪18+ 30‬‬
‫‪,‬‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪q‬‬
‫‪36‬‬
‫√‬ ‫ﭼﻬﺎﺭﻧﻘﻄﻪﺍﻱ‬
‫‪18− 30‬‬
‫= ‪ω1 = ω2‬‬
‫=‪ξ2 :‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪−‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5,‬‬ ‫=‪ξ3 :‬‬ ‫‪3‬‬
‫‪+‬‬
‫‪7‬‬
‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5‬‬
‫‪36‬‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪q‬‬
‫√‬ ‫√‬
‫‪ξ0 := − 13‬‬ ‫‪5+2‬‬ ‫‪10‬‬
‫‪7 ,‬‬ ‫‪ξ1 := − 13‬‬ ‫‪5−2‬‬ ‫‪10‬‬
‫‪7 ,‬‬ ‫=‪ω0 :‬‬ ‫‪322−13 70‬‬
‫‪,‬‬ ‫=‪ω1 :‬‬ ‫‪322+13 70‬‬
‫‪,‬‬
‫‪900‬‬ ‫‪900‬‬
‫‪ξ2 := 0,‬‬ ‫=‪ω2 :‬‬ ‫‪128‬‬
‫‪,‬‬ ‫ﭘﻨﺞﻧﻘﻄﻪﺍﻱ‬
‫‪r‬‬ ‫‪q‬‬ ‫‪r‬‬ ‫‪q‬‬
‫‪225‬‬ ‫√‬ ‫√‬
‫‪322+13 70‬‬ ‫‪322−13 70‬‬
‫=‪ω3 :‬‬ ‫‪,‬‬ ‫= ‪ω4‬‬
‫‪ξ3 := + 13‬‬ ‫‪5 − 2 10‬‬
‫‪7 ,‬‬ ‫‪ξ4 = + 13‬‬ ‫‪5+2‬‬ ‫‪10‬‬
‫‪7‬‬
‫‪900‬‬ ‫‪900‬‬

‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﺗﻘﺎﺭﻥ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﻭﺯﻥﻫﺎ ﻧﻴﺰ ﺳﺮﺍﻳﺖ ﻛﺮﺩﻩ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﺒﺎﺭﺕ ﺩﻳﮕﺮ‪ ،‬ﻭﺯﻥ ﻣﺘﻨﺎﻇﺮ ﺑﺎ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻗﺮﻳﻨﻪ‬
‫ﻳﻜﺴﺎﻥ ﺍﺳﺖ‪ .‬ﻣﻄﺎﺑﻖ ﺑﺎ ﺟﺪﻭﻝ ﻓﻮﻕ‪ ،‬ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﭼﻬﺎﺭﻧﻘﻄﻪﺍﻱ ﮔﻮﺳﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪Z‬‬ ‫‪r‬‬ ‫!‬
‫‪1‬‬ ‫√‬ ‫‪q‬‬
‫≃ ‪f (x)dx‬‬ ‫‪18+ 30‬‬
‫‪36‬‬ ‫‪f‬‬ ‫‪−‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪+‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5‬‬ ‫‪+‬‬
‫‪−1‬‬
‫‪r‬‬ ‫!‬ ‫‪r‬‬ ‫!‬ ‫‪r‬‬ ‫!‬
‫√‬ ‫‪q‬‬ ‫√‬ ‫‪q‬‬ ‫√‬ ‫‪q‬‬
‫‪2‬‬
‫‪18− 30‬‬
‫‪36‬‬ ‫‪f‬‬ ‫‪−‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪−‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5‬‬ ‫‪+‬‬ ‫‪18− 30‬‬
‫‪36‬‬ ‫‪f‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪−‬‬ ‫‪2‬‬
‫‪7‬‬
‫‪6‬‬
‫‪5‬‬ ‫‪+‬‬ ‫‪18+ 30‬‬
‫‪36‬‬ ‫‪f‬‬ ‫‪3‬‬
‫‪7‬‬ ‫‪+‬‬ ‫‪6‬‬
‫‪5‬‬ ‫)‪(۱۰.۹‬‬
‫‪7‬‬
‫ﺩﺭ ﺟﺪﻭﻝ ﻓﻮﻕ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﻭ ﻭﺯﻥﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﻳﻚ ﺗﺎ ﭘﻨﺞ ﻧﻘﻄﻪﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﺑﺴﺘﻪ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺍﺳﺖ‪ .‬ﻧﻜﺘﻪ ﺍﻳﻦ‬
‫ﺍﺳﺖ ﻛﻪ ﺑﻪ ﺍﺯﺍﻱ ‪n‬ﻫﺎﻱ ﺑﺰﺭﮒﺗﺮ‪ ،‬ﻭﺯﻥﻫﺎﻱ ﮔﻮﺳﻲ ﺩﺍﺭﺍﻱ ﻓﺮﻡ ﺑﺴﺘﻪ ﻧﻤﻲﺑﺎﺷﻨﺪ ﻭ ﻫﻤﺎﻧﻨﺪ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‪ ،‬ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻭﺯﻥﻫﺎ‬
‫ﺍﺯ ﺭﻭﺵﻫﺎﻱ ﻋﺪﺩﻱ ﺧﺎﺹ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۲۷‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫■ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﺩﺭ ﺑﺎﺯﻩﻫﺎﻱ ﺩﻟﺨﻮﺍﻩ‬

‫‪R +1‬‬
‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ −1 f (x)dx‬ﺍﺭﺍﻳﻪ ﺷﺪﻧﺪ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﺑﺎﺯﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ]‪ [−1, +1‬ﻣﻲﺑﺎﺷﺪ‪.‬‬
‫‪R‬‬
‫ﺍﻣﺎ ﭼﮕﻮﻧﻪ ﻣﻲﺗﻮﺍﻥ ﺍﻳﻦ ﻓﺮﻣﻮﻝﻫﺎ ﺭﺍ ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ‪ ab f (x)dx‬ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ؟ ﭘﺎﺳﺦ ﺍﻳﻦ ﺳﻮﺍﻝ ﺳﺎﺩﻩ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﺎ ﻳﻚ ﺗﻐﻴﻴﺮ‬
‫ﻣﺘﻐﻴﺮ ﺳﺎﺩﻩ ﻣﻲﺗﻮﺍﻥ ﺑﺎﺯﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺭﺍ ﺍﺯ ]‪ [a, b‬ﺑﻪ ]‪ [−1, +1‬ﺗﺒﺪﻳﻞ ﻛﺮﺩ ﻭ ﺳﭙﺲ ﻗﺎﻋﺪﻩﻫﺎﻱ ﮔﻮﺳﻲ ﺭﺍ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪ .‬ﺩﺍﺭﻳﻢ‪:‬‬
‫‪Z‬‬ ‫‪Z‬‬ ‫‬ ‫‬
‫‪b‬‬
‫‪b−a‬‬ ‫‪b−a‬‬ ‫‪+1‬‬
‫‪b+a‬‬
‫= ‪f (x)dx‬‬ ‫‪t+‬‬ ‫‪f‬‬ ‫‪dt‬‬
‫‪a‬‬ ‫‪−1‬‬ ‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬
‫ﺣﺎﻝ ﺑﺎ ﺑﻪ ﻛﺎﺭﮔﻴﺮﻱ ﻗﺎﻋﺪﻩ ﮔﻮﺳﻲ ‪ n + 1‬ﻧﻘﻄﻪﺍﻱ ﮔﻮﺳﻲ‪ ،‬ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺍﻧﺘﮕﺮﺍﻝ ﺳﻤﺖ ﺭﺍﺳﺖ‪ ،‬ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪Z b‬‬ ‫‬ ‫‬
‫‪b−aX‬‬
‫‪n‬‬
‫‪b−a‬‬ ‫‪b+a‬‬
‫≃ ‪f (x)dx‬‬ ‫‪ωi f‬‬ ‫‪ξi +‬‬ ‫‪.‬‬
‫‪a‬‬ ‫‪2 i=0‬‬ ‫‪2‬‬ ‫‪2‬‬
‫‪R‬‬
‫ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﮔﻮﺳﻲ ﺳﻪ ﻧﻘﻄﻪﺍﻱ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ ab f (x)dx‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻣﻲﺑﺎﺷﺪ‪:‬‬
‫‪Z b‬‬ ‫ ‬ ‫‪q‬‬ ‫‬ ‫‬ ‫‪q‬‬ ‫‬
‫‪b−a 5‬‬ ‫‪
5‬‬
‫≃ ‪f (x)dx‬‬ ‫‪9f − 2‬‬
‫‪b−a‬‬ ‫‪3‬‬ ‫‪b+a‬‬
‫‪5 + 2‬‬
‫‪8‬‬ ‫‪b+a‬‬
‫‪+ 9f 2 + 9f‬‬ ‫‪b−a‬‬
‫‪2‬‬
‫‪3‬‬
‫‪5 + 2‬‬
‫‪b+a‬‬
‫‪.‬‬
‫‪a‬‬ ‫‪2‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺮﺍﻱ ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ‬ ‫‪۵.۹‬‬

‫ﺩﺭ ﺗﻮﺍﺑﻊ ﺟﺪﻭﻟﻲ‪ ،‬ﺿﺎﺑﻄﻪ ﺗﺎﺑﻊ ‪ f‬ﺩﺭ ﺍﺧﺘﻴﺎﺭ ﻧﻤﻲﺑﺎﺷﺪ ﻭ ﻣﻘﺪﺍﺭ ﺩﺭ ﻧﻘﺎﻁ ﻣﺸﺨﺼﻲ ﺩﺭ ﺍﺧﺘﻴﺎﺭ ﺍﺳﺖ‪ .‬ﺩﺭ ﻳﻚ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﻣﻤﻜﻦ ﺍﺳﺖ‬
‫ﻛﻪ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﺎﻁ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﺳﺘﺨﺮﺍﺝ ﺷﺪﻩ‪ ،‬ﻣﺎﻧﻨﺪ ﺫﻭﺯﻧﻘﻪ‪،‬‬
‫ﺳﻴﻤﺴﻮﻥ ﻭ ‪ ...‬ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﺎﺷﺪ‪:‬‬

‫‪xi‬‬ ‫‪0.0‬‬ ‫‪0.1‬‬ ‫‪0.2‬‬ ‫‪0.3‬‬ ‫‪0.4‬‬

‫) ‪f (xi‬‬ ‫‪2.1‬‬ ‫‪3.7‬‬ ‫‪3.2‬‬ ‫‪2.9‬‬ ‫‪1.9‬‬

‫ﺑﺮﺍﻱ ﻣﺤﺎﺳﺒﻪ ﺍﻧﮕﺮﺍﻝ ﺗﺎﺑﻊ ﺟﺪﻭﻟﻲ ﻓﻮﻕ ﺩﺭ ﺑﺎﺯﻩ ]‪ [0.0, 0.5‬ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‪ ،‬ﺫﻭﺯﻧﻘﻪ‪ ،‬ﺩﻭﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬‬
‫ﻛﺎﺗﺲ ﺑﺎﺯ ﻭ ﭘﻨﺞ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺴﻲ ﺑﺴﺘﻪ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺍﻟﺒﺘﻪ ﻓﻘﻂ ﻓﺮﻣﻮﻝ ﭘﻨﺞ ﻧﻘﻄﻪﺍﻱ ﻧﻴﻮﺗﻦ‪-‬ﻛﺎﺗﺲ ﺑﺴﺘﻪ ﺍﺯ ﺗﻤﺎﻣﻲ ﻧﻘﺎﻁ‬
‫ﺩﺍﺩﻩ ﺷﺪﻩ ﺗﺎﺑﻊ ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﻛﻨﺪ‪ .‬ﺍﺻﻮﻻ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺍﻧﺘﮕﺮﺍﻝ ﺗﺎﺑﻊ ﻓﻮﻕ‪ ،‬ﻧﺒﺎﻳﺪ ﺑﻪ ﺩﻧﺒﺎﻝ ﺍﺳﺘﺨﺮﺍﺝ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺟﺪﻳﺪ‬
‫ﺑﺎﺷﻴﻢ‪ .‬ﭼﺮﺍ ﻛﻪ ﺍﺯ ﺑﻴﻦ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻣﻮﺟﻮﺩ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺧﻲ ﺭﺍ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪.‬‬
‫‪R 0.4‬‬
‫‪ 0.0‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ ﻛﻪ ‪ f‬ﺑﻪ ﺻﻮﺭﺕ ﺟﺪﻭﻟﻲ ﺯﻳﺮ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺖ‪:‬‬ ‫ﺣﺎﻝ ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ‪f (x)dx‬‬

‫‪xi‬‬ ‫‪0.0‬‬ ‫‪0.2‬‬ ‫‪0.3‬‬ ‫‪0.4‬‬

‫) ‪f (xi‬‬ ‫‪2.1‬‬ ‫‪3.2‬‬ ‫‪2.9‬‬ ‫‪1.9‬‬

‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﺑﺎﺯ ﻫﻢ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‪ ،‬ﺫﻭﺯﻧﻘﻪ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺍﻧﺘﮕﺮﺍﻝ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺍﮔﺮ ﺑﺨﻮﺍﻫﻴﻢ ﺍﺯ‬
‫ﺗﻤﺎﻣﻲ ﻧﻘﺎﻁ ﺩﺍﺩﻩ ﺷﺪﻩ ﺍﺳﺘﻔﺎﺩﻩ ﻛﻨﻴﻢ‪ ،‬ﺁﻥﮔﺎﻩ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺟﺪﻭﻟﻲ ﺩﺍﺩﻩ ﺷﺪﻩ ﺑﻪ ﻛﺎﺭ ﻧﻤﻲﺁﻳﻨﺪ ﻭ ﺑﺎﻳﺪ ﻳﻚ ﻓﺮﻣﻮﻝ ﭼﻬﺎﺭ ﻧﻘﻄﻪﺍﻱ ﺑﺎ‬
‫ﻧﻘﺎﻁ ‪ x0 = 0, x1 = 0.2, x3 = 0.3, x4 = 0.4‬ﺑﺴﺎﺯﻳﻢ‪.‬‬
‫ﻧﺤﻮﻩ ﺳﺎﺧﺖ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﻪ ﺁﺭﺍﻳﺶ ﻭ ﺗﻮﺯﻳﻊ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺭﺑﻄﻲ ﻧﺪﺍﺭﺩ‪ .‬ﺑﻨﺎﺑﺮﺍﻱ ﻭﻗﺘﻲ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ‬
‫ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﻧﺒﺎﺷﻨﺪ ﻭ ﻳﺎ ﻧﻘﺎﻁ ﮔﻮﺳﻲ ﻧﺒﺎﺷﻨﺪ‪ ،‬ﺑﺎﺯ ﻫﻢ ﻣﻲﺗﻮﺍﻥ ﺍﺯ ﺭﻭﺵﻫﺎﻱ ﻻﮔﺮﺍﻧﮋ ﻭ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ‬
‫ﻓﺮﻣﻮﻝ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۲۸‬‬

‫ﻣﺜﺎﻝ ‪)۲−۹‬ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﺑﺮﺣﺴﺐ ﻧﻘﺎﻁ ﺧﺎﺹ(‬

‫‪R‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻣﻲﺧﻮﺍﻫﻴﻢ ‪ ab f (x)dx‬ﺭﺍ ﺑﺮﺣﺴﺐ ﻣﻘﺪﺍﺭ ﺗﺎﺑﻊ ﺩﺭ ﻧﻘﺎﻁ ﺯﻳﺮ ﺑﻴﺎﺑﻴﻢ‪:‬‬
‫‪a+b‬‬ ‫‪a + 3b‬‬
‫=‪x0 := a, x1 :‬‬ ‫=‪, x2 :‬‬ ‫‪, x3 := b‬‬
‫‪2‬‬ ‫‪4‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ ﻧﻘﺎﻁ ﻣﺘﺴﺎﻭﻱﺍﻟﻔﺎﺻﻠﻪ ﻧﻴﺴﺘﻨﺪ‪ .‬ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﻳﻚ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﭘﻴﺸﻨﻬﺎﺩ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫) ‪f (x)dx ≃ w0 f (x0 ) + w1 f (x1 ) + w2 f (x2 ) + w3 f (x3‬‬
‫‪a‬‬
‫ﻛﻪ‬
‫‪Z‬‬ ‫‪b‬‬
‫= ‪wi‬‬ ‫‪ℓi (x)dx, i = 0, 1, 2, 3.‬‬
‫‪a‬‬
‫ﺑﻨﺎﺑﺮﺍﻳﻦ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝ ﻛﺎﻓﻲ ﺍﺳﺖ ﻛﻪ ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ﻓﻮﻕ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺗﺤﻠﻴﻠﻲ ﺑﻪ ﺩﺳﺖ ﺁﻭﺭﻳﻢ‪ .‬ﻛﻪ ﺍﻟﺒﺘﻪ ﻛﻤﻲ ﻭﻗﺖﮔﻴﺮ‬
‫ﺍﺳﺖ‪ .‬ﺭﻭﺵ ﺿﺮﺍﻳﺐ ﻧﺎﻣﻌﻴﻦ ﺭﺍ ﻧﻴﺰ ﻣﻲﺗﻮﺍﻥ ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻓﺮﻣﻮﻝ ﺑﻪ ﻛﺎﺭ ﺑﺮﺩ‪.‬‬

‫ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺮﻛﺐ‬ ‫‪۶.۹‬‬

‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ﺍﻧﺘﮕﺮﺍﻝ ﺯﻳﺮ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﻋﺪﺩﻱ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪:‬‬
‫‪Z‬‬ ‫‪11π‬‬
‫‪sin(x)dx‬‬
‫‪0‬‬
‫ﻣﻘﺪﺍﺭ ﺩﻗﻴﻖ ﺍﻧﺘﮕﺮﺍﻝ ﺑﺮﺍﺑﺮ ‪ 2‬ﺍﺳﺖ‪ .‬ﺩﺭ ﺯﻳﺮ ﺑﺮﺧﻲ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﺭﺍﻳﻪ ﺷﺪﻩ ﺭﺍ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ﺍﻳﻦ ﺍﻧﺘﮕﺮﺍﻝ ﺑﻪ ﻛﺎﺭ ﻣﻲﺑﺮﻳﻢ‪:‬‬
‫‪R 11π‬‬
‫ﺫﻭﺯﻧﻘﻪ‬ ‫‪0‬‬
‫≃ ‪sin(x)dx‬‬ ‫‪11π‬‬
‫‪2 [sin(0) + sin(11π)] = 0‬‬
‫‪R 11π‬‬ ‫ ‪11π‬‬ ‫‬
‫ﺳﻴﻤﺴﻮﻥ‬ ‫‪0‬‬
‫≃ ‪sin(x)dx‬‬ ‫‪2‬‬
‫‪3‬‬ ‫‪2 ) + sin(11π) = − 3 ≃ −2.0944‬‬
‫‪sin(0) + 4 sin( 11π‬‬ ‫‪2π‬‬

‫‪R 11π‬‬ ‫ ‪3 3‬‬ ‫‬

‫‪11π‬‬
‫≃ ‪ ۳/۸ 0 sin(x)dx‬ﺳﻴﻤﺴﻮﻥ‬ ‫‪3 ) + 3 sin( 3 ) + sin(11π) ≃ −22.4458‬‬
‫‪sin(0) + 3 sin( 11π‬‬ ‫‪22π‬‬
‫‪R 11π‬‬ ‫‪8‬‬
‫‬ ‫‬
‫ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‬ ‫‪0‬‬
‫≃ ‪sin(x)dx‬‬ ‫‪2 ) ≃ −34.5575‬‬
‫‪11π sin( 11π‬‬
‫ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻧﺘﺎﻳﺞ ﺑﺎ ﺧﻄﺎﻱ ﺯﻳﺎﺩﻱ ﻫﻤﺮﺍﻩ ﺍﺳﺖ‪ .‬ﺍﻳﻦ ﺧﻄﺎﻱ ﺑﺰﺭﮒ ﺭﺍ ﻣﻲﺗﻮﺍﻧﺴﺘﻴﻢ ﭘﻴﺶﺑﻴﻨﻲ ﻛﻨﻴﻢ‪ .‬ﭼﺮﺍ ﻛﻪ ﺑﺎﺯﻩ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺰﺭﮒ ﺍﺳﺖ ﻭ ﻟﺬﺍ ‪ h‬ﻧﻴﺰ ﺑﺰﺭﮒ ﺍﺳﺖ‪ .‬ﭼﻮﻥ ﺗﻮﺍﻧﻲ ﺍﺯ ‪ h‬ﺩﺭ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﻇﺎﻫﺮ ﻣﻲﺷﻮﺩ‪ ،‬ﻟﺬﺍ ﺧﻄﺎﻱ ﺑﺮﺷﻲ ﺑﺰﺭﮒ‬
‫‪ 11π‬ﺍﺳﺖ ﻭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﺑﻪ ﺻﻮﺭﺕ )‪− h90 f (4) (η‬‬ ‫ﻣﻲﺷﻮﺩ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ ‪ h‬ﺩﺭ ﺭﻭﺵ ﺳﻴﻤﺴﻮﻥ ﺑﺮﺍﺑﺮ ‪2‬‬
‫‪5‬‬

‫ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﺯﺍﻱ ﺍﻳﻦ ‪ h‬ﺟﻤﻠﻪ ﺧﻄﺎ ﺑﻪ ﺻﻮﺭﺕ )‪ −17112.76954f (4) (η‬ﺍﺳﺖ‪ .‬ﻣﻼﺣﻈﻪ ﻣﻲﺷﻮﺩ ﻛﻪ ﻋﻤﻮﻣﺎ ﭼﻨﻴﻦ ‪h‬ﻱ ﺑﻪ‬
‫ﺧﻄﺎﻱ ﺑﺰﺭﮔﻲ ﻣﻨﺠﺮ ﻣﻲﺷﻮﺩ‪ .‬ﺍﺻﻮﻻ ﺑﺮﺍﻱ ﺁﻥ ﻛﻪ ﻋﺎﻣﻞ ‪ h5‬ﺑﺎﻋﺚ ﺑﺰﺭﮒ ﺷﺪﻥ ﺟﻤﻠﻪ ﺧﻄﺎ ﻧﺸﻮﺩ‪ ،‬ﺑﺎﻳﺪ ﻛﻮﭼﻚﺗﺮ ﺍﺯ ﻳﻚ ﺑﺎﺷﺪ‬
‫ﻭ ﺑﺮﺍﻱ ﺍﻳﻦ ﻣﻨﻈﻮﺭ ﺑﺎﻳﺪ ﺍﺯ ﻳﻚ ﻓﺮﻣﻮﻝ ﺣﺪﺍﻗﻞ ‪ ۳۵‬ﻧﻘﻄﻪﺍﻱ ﺍﺳﺘﻔﺎﺩﻩ ﺷﻮﺩ‪.‬‬
‫ﻫﻤﺎﻥﮔﻮﻧﻪ ﻛﻪ ﺩﻳﺪﻳﻢ‪ ،‬ﺩﺭ ﺑﺮﺧﻲ ﺍﻧﺘﮕﺮﺍﻝﻫﺎ ﻧﻤﻲﺗﻮﺍﻥ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﻛﻢ ﺑﻪ ﺟﻮﺍﺏ ﻣﻄﻠﻮﺑﻲ ﺭﺳﻴﺪ ﻭ ﺑﺎﻳﺪ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ‬
‫ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺑﺎﻻ ﺍﺳﺘﻔﺎﺩﻩ ﻛﺮﺩ‪ .‬ﺍﻣﺎ ﺩﻭ ﻣﺸﻜﻞ ﺯﻳﺮ ﺩﺭ ﺍﻳﻦ ﺭﺍﻩ ﻭﺟﻮﺩ ﺩﺍﺭﺩ‪:‬‬
‫‪ ۱‬ﺳﺎﺧﺖ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺑﺎﻻ ﭘﺮﻫﺰﻳﻨﻪ ﺍﺳﺖ‪.‬‬
‫‪ ۲‬ﻫﺮﭼﻪ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺩﺭ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﺑﺎﻻﺗﺮ ﺑﺮﻭﺩ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻣﺎﺷﻴﻨﻲ ﺩﺭ ﻣﺤﺎﺳﺒﻪ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ‬
‫ﺑﺎﻻﺗﺮ ﻣﻲﺭﻭﺩ ﻭ ﻧﺎﭘﺎﻳﺪﺍﺭﻱ ﻣﺤﺎﺳﺒﺎﺗﻲ ﺭﺥ ﻣﻲﺩﻫﺪ‪.‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺗﻮﺿﻴﺤﺎﺕ ﻓﻮﻕ ﻭ ﺑﺮﺍﻱ ﺍﺟﺘﻨﺎﺏ ﺍﺯ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﺎ ﺗﻌﺪﺍﺩ ﻧﻘﺎﻁ ﺑﺎﻻ ﺍﺯ ﺍﻳﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻋﺪﺩﻱ ﻣﺮﻛﺐ ‪۱۶‬‬

‫‪16 Composite‬‬ ‫‪numerical integration‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۲۹‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪ .‬ﺍﻳﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﺮﻛﺐ ﺑﺴﻴﺎﺭ ﺳﺎﺩﻩ ﺍﺳﺖ‪ .‬ﺩﺭ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﺮﻛﺐ‪ ،‬ﺑﺎﺯﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﻪ ﭼﻨﺪﻳﻦ ﺯﻳﺮﺑﺎﺯﻩ‬
‫ﺗﻘﺴﻴﻢ ﻣﻲ ﺷﻮﺩ ﻭ ﺩﺭ ﻫﺮ ﺯﻳﺮﺑﺎﺯﻩ ﺍﺯ ﻳﻚ ﻗﺎﻋﺪﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﻣﺎﻧﻨﺪ ﺫﻭﺯﻧﻘﻪ‪ ،‬ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ‪ ،‬ﺳﻴﻤﺴﻮﻥ ﻭ ﻳﺎ ‪ ...‬ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲ ﺷﻮﺩ‪.‬‬
‫ﻓﺮﻣﻮﻝ ﺣﺎﺻﻞ ﺑﻪ ﻓﺮﻣﻮﻝ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ‪ ،‬ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻣﺮﻛﺐ‪ ،‬ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ ﻭ ‪ ....‬ﻣﻮﺳﻮﻡ ﻣﻲﺑﺎﺷﺪ‪.‬‬

‫■ ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ‬

‫‪Rb‬‬
‫ﻓﺮﺽ ﻛﻨﻴﺪ ﻛﻪ ﻣﻲﺧﻮﺍﻫﻴﻢ ‪ a f (x)dx‬ﺭﺍ ﺗﻘﺮﻳﺐ ﺑﺰﻧﻴﻢ‪ .‬ﺍﺑﺘﺪﺍ ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﻪ ‪ n‬ﻗﺴﻤﺖ ﻣﺴﺎﻭﻱ ﺗﻘﺴﻴﻢ ﻣﻲﻛﻨﻴﻢ ﻭ ﻧﻘﺎﻁ‬
‫‪ h := b−a‬ﺍﺳﺖ ﻭ ﻧﻘﺎﻁ ﺗﻘﺴﻴﻢ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ‬
‫ﺗﻘﺴﻴﻢ ﺭﺍ ‪ x0 , x1 , . . . , xn‬ﻣﻲﻧﺎﻣﻴﻢ‪ .‬ﺩﺭ ﺍﻳﻦ ﺻﻮﺭﺕ‪ ،‬ﻓﺎﺻﻠﻪ ﺑﻴﻦ ﻧﻘﺎﻁ ﺑﺮﺍﺑﺮ ‪n‬‬
‫ﻣﻲﺑﺎﺷﻨﺪ‪:‬‬
‫‪x0 := a, xn := b, xi := a + ih, i = 0, . . . , n‬‬
‫‪R‬‬
‫ﺑﺮﺍﻱ ﺍﺳﺘﺨﺮﺍﺝ ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ‪ ،‬ﺍﻧﺘﮕﺮﺍﻝ ‪ ab f (x)dx‬ﺭﺍ ﺑﻪ ﺍﻧﺘﮕﺮﺍﻝ ﺭﻭﻱ ﺑﺎﺯﻩﻫﺎﻱ ] ‪ [xi , xi+1‬ﻣﻲﺷﻜﻨﻴﻢ‪:‬‬
‫‪Z b‬‬ ‫‪Z x1‬‬ ‫‪Z x2‬‬ ‫‪Z xn−1‬‬ ‫‪Z xn‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x)dx +‬‬ ‫‪f (x)dx + · · · +‬‬ ‫‪f (x)dx +‬‬ ‫‪f (x)dx‬‬
‫‪a‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫‪xn−2‬‬ ‫‪xn−1‬‬
‫ﻭ ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﺭﺍ ﺭﻭﻱ ﻫﺮ ﺯﻳﺮ ﺑﺎﺯﻩ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫ﻃﺒﻖ ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﻭ ﺧﻄﺎﻱ ﺁﻥ‪ ،‬ﺍﻧﺘﮕﺮﺍﻝ ﺭﻭﻱ ﺑﺎﺯﻩ ] ‪ [xj , xj+1‬ﺭﺍ ﻣﻲﺗﻮﺍﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻧﻮﺷﺖ‪:‬‬
‫‪Z‬‬ ‫‪xj+1‬‬
‫‪h‬‬ ‫‪h3‬‬
‫= ‪f (x)dx‬‬ ‫) ‪[f (xj ) + f (xj+1 )] − f ′′ (ηj‬‬
‫‪xj‬‬ ‫‪2‬‬ ‫‪12‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ ηj‬ﻋﺪﺩﻱ ﺩﺭ ﺑﺎﺯﻩ ] ‪ [xj , xj+1‬ﺍﺳﺖ‪ .‬ﺣﺎﻝ ﺑﺎ ﺍﺳﺘﻔﺎﺩﻩ ﺍﺯ ﺩﻭ ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﺩﺍﺭﻳﻢ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‬ ‫ ‬ ‫‬
‫‪h‬‬ ‫‪h3‬‬ ‫‪h‬‬ ‫‪h3‬‬
‫= ‪f (x)dx‬‬ ‫‪[f (x0 ) + f (x1 )] − f ′′ (η0 ) +‬‬ ‫‪[f (x1 ) + f (x2 )] − f ′′ (η1 ) + · · · +‬‬
‫‪a‬‬ ‫‪2‬‬ ‫‪12‬‬ ‫‪2‬‬ ‫‪12‬‬
‫‬ ‫ ‬ ‫‬
‫‪h‬‬ ‫‪h3 ′′‬‬ ‫‪h‬‬ ‫‪h3‬‬
‫‪[f (xn−2 ) + f (xn−1 )] − f (ηn−2 ) +‬‬ ‫) ‪[f (xn−1 ) + f (xn )] − f ′′ (ηn−1‬‬
‫‪2‬‬ ‫‪12‬‬ ‫‪2‬‬ ‫‪12‬‬
‫ﺑﺎ ﺩﺳﺘﻜﺎﺭﻱ ﺟﺒﺮﻱ ﺧﻮﺍﻫﻴﻢ ﺩﺍﺷﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫‪hh‬‬ ‫‪i‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x0 ) + 2f (x1 ) + 2f (x2 ) + 2f (xn−1 ) + f (xn ) −‬‬
‫‪a‬‬ ‫‪2‬‬
‫‪h3 h ′′‬‬ ‫‪i‬‬
‫) ‪f (η0 ) + f ′′ (η1 ) + · · · + f ′′ (ηn−1‬‬ ‫)‪(۱۱.۹‬‬
‫‪12‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۳۰‬‬

‫ﺍﮔﺮ ‪ f ′′‬ﺩﺭ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ )ﺑﺪﻭﻥ ﺍﺛﺒﺎﺕ ﻗﺒﻮﻝ ﻣﻲﻛﻨﻴﻢ ﻛﻪ( ] ‪ η ∈ [ab‬ﻣﻮﺟﻮﺩ ﺍﺳﺖ ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‬
‫)‪f ′′ (η0 ) + f ′′ (η1 ) + · · · + f ′′ (ηn−1 ) = nf ′′ (η‬‬
‫ﺍﺯ ﺩﻭ ﺭﺍﺑﻄﻪ ﻗﺒﻞ ﻭ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺍﻳﻦ ﻛﻪ ‪ ،nh = b − a‬ﻧﺘﻴﺠﻪ ﻣﻲﮔﻴﺮﻳﻢ ﻛﻪ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬
‫‪hh‬‬ ‫‪i b−a‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x0 ) + 2f (x1 ) + 2f (x2 ) + 2f (xn−1 ) + f (xn ) −‬‬ ‫)‪h2 f ′′ (η‬‬
‫‪a‬‬ ‫‪2‬‬ ‫‪12‬‬
‫ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺭﺍﺑﻄﻪ ﻓﻮﻕ‪ ،‬ﻗﺎﻋﺪﻩ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ ﺭﺍ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﻭ ﺟﻤﻠﻪ ﺧﻄﺎﻱ ﺁﻥ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻴﺎﻥ ﻣﻲﺷﻮﺩ‪:‬‬
‫‪Z‬‬ ‫"‬ ‫‪#‬‬
‫‪b‬‬
‫‪h‬‬ ‫‪X‬‬
‫‪n−1‬‬
‫‪b − a 2 ′′‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x0 ) + 2‬‬ ‫‪f (xi ) + f (xn ) ,‬‬ ‫‪E := −‬‬ ‫)‪h f (η‬‬
‫‪a‬‬ ‫‪2‬‬ ‫‪i=1‬‬
‫‪12‬‬

‫■ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻣﺮﻛﺐ‬

‫‪Rb‬‬
‫ﺩﺭ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻣﺮﻛﺐ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪ ، a f (x)dx‬ﺑﺎﺯﻩ ]‪ [a, b‬ﺭﺍ ﺑﻪ ‪ n + 1‬ﻗﺴﻤﺖ ﻣﺴﺎﻭﻱ ﺗﻘﺴﻴﻢ ﻣﻲﻛﻨﻴﻢ ﺗﺎ ﻧﻘﺎﻁ‬
‫ﺗﻘﺴﻴﻢ ‪ x0 , . . . , xn , xn+1‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺣﺎﺻﻞ ﺷﻮﺩ‪:‬‬
‫‪b−a‬‬
‫‪xi := a + i‬‬ ‫‪, i = 0, . . . , n + 1‬‬
‫‪n+1‬‬
‫ﻭ ﺳﭙﺲ ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺭﺍ ﺭﻭﻱ ﻫﺮ ﺯﻳﺮﺑﺎﺯﻩ ﺍﻋﻤﺎﻝ ﻣﻲﻛﻨﻴﻢ‪ .‬ﺷﻜﻞ ﺯﻳﺮ ﺭﺍ ﺑﺒﻴﻨﻴﺪ‪.‬‬

‫‪h‬‬

‫‪a‬‬ ‫‪x0 x0‬‬ ‫‪x1‬‬

‫‪x1‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x2‬‬ ‫‪xn xn‬‬ ‫‪xn‬‬ ‫‪1‬‬ ‫‪xn‬‬ ‫‪1‬‬ ‫‪b‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬

‫ﺑﻪ ﻋﺒﺎﺭﺕ ﺭﻳﺎﺿﻲ ﺩﺍﺭﻳﻢ‪:‬‬

‫‪Z‬‬ ‫‪b‬‬ ‫‪Z‬‬ ‫‪x1‬‬ ‫‪Z‬‬ ‫‪xn+1‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x)dx + · · · +‬‬‫‪f (x)dx‬‬
‫‪a‬‬ ‫‪x0‬‬ ‫‪xn‬‬
‫‬ ‫‬ ‫‬ ‫‬ ‫‬ ‫‬
‫
‬ ‫‪h3 ′′‬‬ ‫‪xn +xn+1‬‬ ‫‪h3 ′′‬‬
‫‪= 2hf‬‬ ‫‪x0 +x1‬‬
‫‪2‬‬ ‫‪] + f (η0 ) + · · · + 2hf‬‬ ‫‪2‬‬ ‫) ‪] + f (ηn‬‬
‫‪3‬‬ ‫‪3‬‬
‫‪X‬‬ ‫‬ ‫‬
‫‪h X ′′‬‬
‫‪n‬‬ ‫‪3‬‬ ‫‪n‬‬
‫‪xi + xi+1‬‬
‫‪=2h‬‬ ‫‪f‬‬ ‫‪+‬‬ ‫) ‪f (ηi‬‬
‫‪i=0‬‬
‫‪2‬‬ ‫‪3 i=0‬‬
‫‪xi −xi−1‬‬
‫= ‪.h‬‬
‫ﺗﻮﺟﻪ ﺷﻮﺩ ﻛﻪ‬ ‫‪2‬‬ ‫=‬ ‫‪b−a‬‬
‫)‪2(n+1‬‬
‫‪Pn‬‬
‫ﺩﺭ ﺣﺎﻟﺘﻲ ﻛﻪ ﻣﺸﺘﻖ ﺩﻭﻡ ‪ f‬ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﭘﻴﻮﺳﺘﻪ ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪، i=0 f ′′ (ηi ) = (n + 1)f ′′ (η) :‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ‪ η‬ﻋﺪﺩﻱ ﺩﺭ ]‪ [a, b‬ﺍﺳﺖ‪ .‬ﺑﻪ ﺍﻳﻦ ﺗﺮﺗﻴﺐ ﺍﺯ ﺭﺍﺑﻄﻪ ﻓﻮﻕ ﻭ ﺍﻳﻦ ﻧﻜﺘﻪ ﻛﻪ ‪ ،(n + 1)2h = b − a‬ﻧﺘﻴﺠﻪ ﺧﻮﺍﻫﻴﻢ‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪۲۳۱‬‬ ‫ﺍﻧﺘﮕﺮﺍﻝﮔﯿﺮﯼ ﻋﺪﺩﯼ‬ ‫ﻓﺼﻞ ﻧﻬﻢ‪:‬‬

‫ﮔﺮﻓﺖ ﻛﻪ‪:‬‬
‫‪Z‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‬ ‫‬
‫‪b‬‬
‫‪xi + xi+1‬‬ ‫‪b − a 2 ′′‬‬
‫‪f (x)dx = 2h‬‬ ‫‪f‬‬ ‫‪+‬‬ ‫)‪h f (η‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫‪2‬‬ ‫‪6‬‬
‫ﻣﻄﺎﻟﺐ ﻓﻮﻕ ﺩﺭ ﺗﻌﺮﻳﻒ ﺯﻳﺮ ﺧﻼﺻﻪ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﻗﺎﻋﺪﻩ ﻧﻘﻄﻪ ﻣﯿﺎﻧ ﻣﺮﮐﺐ‬ ‫ﺗﻌﺮﯾﻒ ‪۱۰−۹‬‬
‫‪Rb‬‬
‫ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬ ‫ﻓﺮﻣﻮﻝ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﻣﺮﻛﺐ ‪ n‬ﻧﻘﻄﻪﺍﻱ ﺑﺮﺍﻱ ﺗﻘﺮﻳﺐ ‪f (x)dx‬‬
‫‪a‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬‫‪n‬‬ ‫‬ ‫‬
‫‪xi + xi+1‬‬
‫‪f (x)dx ≃ 2h‬‬ ‫‪f‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫‪2‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ‪:‬‬
‫‪b−a‬‬
‫=‪h :‬‬ ‫‪, xi := a + 2ih, i = 0, . . . , n + 1.‬‬
‫)‪2(n + 1‬‬
‫ﺍﮔﺮ ‪ f‬ﺩﺍﺭﺍﻱ ﻣﺸﺘﻖ ﭘﻴﻮﺳﺘﻪ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﺎﺷﺪ‪ ،‬ﺁﻥﮔﺎﻩ ﺧﻄﺎﻱ ﻓﺮﻣﻮﻝ ﻧﻘﻄﻪ ﻣﻴﺎﻧﻲ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬
‫‪b − a 2 ′′‬‬
‫=‪E‬‬ ‫)‪h f (η‬‬
‫‪6‬‬
‫ﺑﻪ ﻃﻮﺭﻱ ﻛﻪ ]‪.η ∈ [a, b‬‬

‫ﻣﻤﻜﻦ ﺍﺳﺖ ﺩﺭ ﺑﺮﺧﻲ ﻣﻨﺎﺑﻊ ﺩﻳﮕﺮ‪ ،‬ﻓﺮﻣﻮﻝ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ ﺑﺎ ﻇﺎﻫﺮﻱ ﻣﺘﻔﺎﻭﺕ ﺍﺭﺍﻳﻪ ﺷﻮﺩ‪ .‬ﺩﺭ ﻭﺍﻗﻊ ﻓﺮﻣﻮﻝﻫﺎﻱ ﺩﻳﮕﺮﻱ‬
‫ﺑﺮﺍﻱ ﺫﻭﺯﻧﻘﻪ ﻣﺮﻛﺐ ﻭﺟﻮﺩ ﻧﺪﺍﺭﺩ ﻭ ﻋﻠﺖ ﺗﻔﺎﻭﺕ ﻇﺎﻫﺮﻱ ﻓﺮﻣﻮﻝﻫﺎ ﺗﻔﺎﻭﺕ ﺩﺭ ﻧﺎﻣﮕﺬﺍﺭﻱ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺍﺳﺖ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‬
‫ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ‪ ،‬ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﻓﺮﻡ ﺩﻳﮕﺮﻱ ﻧﺎﻣﮕﺬﺍﺭﻱ ﺷﺪﻩﺍﻧﺪ ﻭ ﺗﺒﻌﺎ ﻓﺮﻡ ﻓﺮﻣﻮﻝ ﻧﻴﺰ ﺑﻪ ﻃﻮﺭ ﻣﺘﻔﺎﻭﺗﻲ ﺍﺳﺘﺨﺮﺍﺝ ﺧﻮﺍﻫﺪ ﺷﺪ‪.‬‬

‫‪Z‬‬ ‫‪b‬‬ ‫‪X‬‬‫‪n‬‬ ‫‬ ‫‬

‫‪xi + xi+1‬‬
‫‪f (x)dx ≃ 2h‬‬ ‫‪f‬‬
‫‪a‬‬ ‫‪i=0‬‬
‫‪2‬‬
‫‪b−a‬‬
‫=‪h :‬‬
‫)‪2(n + 1‬‬
‫‪h‬‬
‫‪xi = a + (2i + 1)h, i = 0, . . . , n‬‬
‫‪a‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪xn‬‬ ‫‪1‬‬ ‫‪xn‬‬ ‫‪b‬‬

‫■ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ‬

‫ﺩﺭ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ‪ ،‬ﺑﺎﺯﻩ ﺍﻧﺘﮕﺮﺍﻝﮔﻴﺮﻱ ﺑﻪ ﺯﻳﺮﺑﺎﺯﻩﻫﺎﻳﻲ ﺗﻘﺴﻴﻢ ﻣﻲﺷﻮﺩ ﻭ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﺭﻭﻱ ﺯﻳﺮﺑﺎﺯﻩﻫﺎ ﺍﻋﻤﺎﻝ ﻣﻲﺷﻮﺩ‪.‬‬
‫ﺍﮔﺮ ﻧﻘﺎﻁ ‪ x0 , . . . , xn‬ﺭﺍ ﺩﺭ ﺑﺎﺯﻩ ]‪ [a, b‬ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺍﻧﺘﺨﺎﺏ ﻛﻨﻴﻢ‪:‬‬
‫‪b−a‬‬
‫‪xi = a + i‬‬ ‫‪, i = 0, . . . , n‬‬
‫‪n‬‬
‫ﺁﻥﮔﺎﻩ ﻣﻲﺗﻮﺍﻥ ﻧﻮﺷﺖ‪:‬‬
‫‪Z‬‬ ‫‪b‬‬ ‫‪X Z xi+1‬‬
‫‪n−1‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x)dx‬‬
‫‪a‬‬ ‫‪i=0‬‬ ‫‪xi‬‬
‫‪R xi−1‬‬
‫‪xi , xi +x‬‬
‫‪2‬‬
‫‪i+1‬‬
‫ﺭﺍ ﺑﺎ ﻗﺎﻋﺪﻩ ﺳﻴﻤﺴﻮﻥ ﻣﺤﺎﺳﺒﻪ ﻣﻲﻛﻨﻴﻢ ﻛﻪ ﺩﺭ ﺍﻳﻦ ﻗﺎﻋﺪﻩ ﺍﺯ ﻧﻘﺎﻁ ‪, xi+1‬‬ ‫‪xi‬‬
‫ﺣﺎﻝ ﺍﻧﺘﮕﺮﺍﻝﻫﺎﻱ ‪f (x)dx‬‬
‫ﺍﺳﺘﻔﺎﺩﻩ ﻣﻲﺷﻮﺩ‪.‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬
 ‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬

‫ﻣﺤﺎﺳﺒﺎﺕ ﻋﺪﺩﯼ‬ ‫‪۲۳۲‬‬

‫‪h‬‬

‫‪x0‬‬ ‫‪x1‬‬ ‫‪x1‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x2‬‬ ‫‪xn‬‬ ‫‪xn‬‬ ‫‪xn‬‬ ‫‪xn‬‬ ‫‪b‬‬
‫‪a‬‬ ‫‪x0‬‬ ‫‪1‬‬
‫‪1‬‬
‫‪2‬‬ ‫‪2‬‬ ‫‪2‬‬

‫ﺑﺎ ﺗﻮﺿﻴﺤﺎﺕ ﻓﻮﻕ‪ ،‬ﺩﺭ ﻧﻬﺎﻳﺖ ﻓﺮﻣﻮﻝ ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ ﺑﻪ ﺻﻮﺭﺕ ﺯﻳﺮ ﺑﻪ ﺩﺳﺖ ﻣﻲﺁﻳﺪ‪:‬‬
‫‪‬‬ ‫‪‬‬
‫‪Z‬‬ ‫‪X‬‬
‫‪n−1‬‬ ‫‬ ‫‬ ‫‪X‬‬
‫‪n−1‬‬
‫‪xn‬‬
‫‪h‬‬ ‫)‪b − a 4 (4‬‬
‫= ‪f (x)dx‬‬ ‫‪f (x0 ) + 4‬‬ ‫‪f‬‬ ‫‪xi +xi+1‬‬
‫‪2‬‬ ‫‪+2‬‬ ‫‪f (xi ) + f (xn ) −‬‬ ‫‪h f (η),‬‬
‫‪x0‬‬ ‫‪3‬‬ ‫‪i=0‬‬ ‫‪j=1‬‬
‫‪180‬‬
‫‪b−a‬‬ ‫‪b−a‬‬
‫=‪h :‬‬ ‫‪, xi := a + i‬‬ ‫‪, i = 0, . . . , n‬‬
‫‪2n‬‬ ‫‪2‬‬
‫ﻓﺮﻣﻮﻝ ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ ﻧﻴﺰ ﺑﻪ ﻓﺮﻡﻫﺎﻱ ﻣﺘﻔﺎﻭﺗﻲ ﺑﻴﺎﻥ ﻣﻲﺷﻮﺩ‪ .‬ﺗﻔﺎﻭﺕ ﻓﺮﻡﻫﺎﻱ ﻓﺮﻣﻮﻝ ﺳﻴﻤﺴﻮﻥ ﺗﻨﻬﺎ ﺩﺭ ﻧﺎﻣﮕﺬﺍﺭﻱ‬
‫ﻣﺘﻐﻴﺮﻫﺎﻱ ﮔﺮﻩﺍﻱ ﻣﻲﺑﺎﺷﺪ‪ .‬ﺑﻪ ﻋﻨﻮﺍﻥ ﻧﻤﻮﻧﻪ‪ ،‬ﺍﮔﺮ ﻧﺎﻣﮕﺬﺍﺭﻱ ﻧﻘﺎﻁ ﮔﺮﻩﺍﻱ ﺑﻪ ﺻﻮﺭﺕ ﻣﻨﺪﺭﺝ ﺩﺭ ﺷﻜﻞ ﺯﻳﺮ ﺑﺎﺷﺪ‬

‫‪h‬‬

‫‪a‬‬ ‫‪x0‬‬ ‫‪x1‬‬ ‫‪x2‬‬ ‫‪x3‬‬ ‫‪x4‬‬ ‫‪x 2n‬‬ ‫‪2‬‬ ‫‪x 2n‬‬ ‫‪1‬‬
‫‪x 2n‬‬ ‫‪b‬‬

‫ﺁﻥﮔﺎﻩ ﻓﺮﻣﻮﻝ ﺳﻴﻤﺴﻮﻥ ﻣﺮﻛﺐ ﺑﻪ ﻗﺮﺍﺭ ﺯﻳﺮ ﺍﺳﺖ‪:‬‬

‫‪‬‬ ‫‪‬‬
‫‪Z‬‬ ‫‪x2n‬‬ ‫‪X‬‬
‫‪n‬‬ ‫‪X‬‬
‫‪n−1‬‬
‫‪h‬‬
‫≃ ‪f (x)dx‬‬ ‫‪f (x0 ) + 4‬‬ ‫‪f (x2i−1 ) + 2‬‬ ‫‪f (x2j ) + f (x2n ) ,‬‬
‫‪x0‬‬ ‫‪3‬‬ ‫‪i=1‬‬ ‫‪j=1‬‬

‫‪b−a‬‬
‫=‪h :‬‬ ‫‪, xi := a + ih, i = 0, . . . , 2n‬‬
‫‪2n‬‬

‫‪i‬‬ ‫‪i‬‬

‫‪i‬‬ ‫‪i‬‬