Merge Sort

•Merge Sort works on the idea of divide and

conquer.

•Divide: Repeatedly divide the list of elements in

half size until the size becomes 1.

•Size 1 array is always considered as sorted array

as there is only one element in the array.

•Conquer and combine: Solve the sub problems

and combine them.
 Example
14 7 3 12 9 11 6 2
Divide
step
14 7 3 12 9 11 6 2

14 7 3 12 9 11 6 2 Divide
step
14 7 3 12 9 11 6 2

7 14 3 12 9 11 2 6

Conquer
3 7 12 14 and
2 6 9 11
combine
step

2 3 6 7 9 11 12 14
 Simple_merge(K,FIRST,SECOND,THIRD)
1.[Initialize] 3. [Copy remaining unprocessed
i=FIRST elements]
j=SECOND If i>=SECOND
l=0
Then Repeat while j<=THIRD
2.[Compare]
l=l+1
Repeat while i<SECOND and
Temp[l]=k[j]
j<=THIRD
j=j+1
If K[i]<=K[j]
Else Repeat while i<SECOND
Then l=l+1
l=l+1
Temp[l]=k[i]
Temp[l]=k[i]
i=i+1
i=i+1
Else
4.Repeat for i=1,2,3…l
l=l+1
K[FIRST-1+l]=Temp[l]
Temp[l]=k[j]
j=j+1 5. return
Two_way_merge_sort_r(K,START,FINISH)
1.[Compare the size of the current subtable]
SIZE=FINISH-START+1
2.[Test base condition]
If SIZE<=1
Return
3.[Calculate midpoint]
Middle=START+ ΓSIZE/2˥-1
4. [Recursively sort first subtable]
Call Two_way_merge_sort_r(K,START,Middle)
5. [Recursively sort second subtable]
Call Two_way_merge_sort_r(K,Middle+1, FINISH)
6.[Merge two ordered subtables]
Call Simple_merge(K, START,Middle+1,FINISH)
7. return
 Example
T1: 5,12,25 T2: 6,15

Simple_merge(K,FIRST,SECOND,THIRD K= 5 12 25 6 15
) FIRST=1 SECOND=4 THIRD=5
1.[Initialize]
i=FIRST
j=SECOND
1. i=1
l=0 j=4
2.[Compare]
Repeat while i<SECOND and
l=0
j<=THIRD
If K[i]<=K[j]
Then l=l+1
Temp[l]=k[i]
i=i+1
Else
l=l+1
Temp[l]=k[j]
j=j+1
while 3<4 &
i<SECOND & 1<4 & Y 2<4 & 2<4 & 3<4 &
Y Y 6<=5
j<=THIRD 4<=5 4<=5 5<=5 Y 5<=5 False
If K[i]<=K[j] cond
K[1]5<=K[4]6 K[2]12<=k[4]6 K[2]12<=k[5]15 K[3]25<=k[5]15
Yes: l=l+1
l=1 l=3
Temp[l]=k[i]
Temp[1]=k[1] Temp[3]=k[2]
i=i+1
i=2 i=3
No: l=l+1
l=2 l=4
Temp[l]=k[j]
Temp[2]=k[4] Temp[4]=k[5]
j=j+1
j=5 j=6

i j l
Temp[1]=5
1 4 0 index 1 2 3 4 5
Temp[2]=6 value 5 12 25 6 15
2 1 Temp[3]=12
5 2
3 3 Temp[4]=15 FIRST=1
6 4 SECOND=4
THIRD=5
3. [Copy remaining i=3 j=6 l=4
unprocessed elements] 3. If (3)>=4
If i>=SECOND
While(3<4)
Then Repeat while j<=THIRD
l=5
l=l+1
Temp[5]=k[3]=25
Temp[l]=k[j]
j=j+1
i=4
Else Repeat while i<SECOND While(4<4) false
l=l+1 Now,1 temp
Temp 2 array
3 is:4 5
val 5 6 12 15 25
Temp[l]=k[i]
i=i+1 4. k[1]=Temp[1]=5
4.Repeat for i=1,2,3…l k[2]=Temp[2]=6
K[FIRST-1+i]=Temp[i] k[3]=Temp[3]=12
k[4]=Temp[4]=15
5. return
k[5]=Temp[5]=25
T_w_m_s(K,Start,Finish)
Recursion explanation for: 6 5 25 12 15
1.SIZE=FINISH-START+1
T_w_m_s(K,1,5)
2.If SIZE<=1 Return 6 5 25 12 15 T_w_m_s(K,1,3)
6 5 25
3. Middle=START+ ΓSIZE/2˥-1 1.SIZE=5-1+1=5
2.If SIZE<=1 no 1.SIZE=3-1+1=3
4. Call
Two_way_merge_sort_r(K,START,Mid 3. Middle=1+ Γ5/2˥-1= 3 2.If SIZE<=1 no
dle) 3. Middle=1+ Γ3/2˥-1= 2
4. Call T_w_m_s(K,1,3)
5.Call 5 6
Two_way_merge_sort_r(K,Middle+1, 4. Call T_w_m_s(K,1,2)
FINISH) 5. Call T_w_m_s(K,4,5)
6. Call Simple_merge(K, 5. Call T_w_m_s(K,3,3) 25
6. Call S_m(K,1,4,5)
START,Middle+1,FINISH)
7. Return 6. Call S_m(K,1,3,3)
7. Return
7. Return

T_w_m_s(K,3,3) 25 T_w_m_s(K,1,1) 6 T_w_m_s(K,1,2)

Call T_w_m_s(K,2,2) 5 6 5
1.SIZE=3-3+1=1 1.SIZE=1-1+1=1
1.SIZE=2-2+1=1 1.SIZE=2-1+1=2
2.If SIZE<=1 yes 2.If SIZE<=1 yes
2.If SIZE<=1 yes 2.If SIZE<=1 no
3. return 3. return
3. return 3. Middle=1+ Γ2/2˥-1= 1
4. Call T_w_m_s(K,1,1)
In last, two tables which
contains 5 6 and 25 resp. are 5. Call T_w_m_s(K,2,2)
sorted and output is 5 6 25 6. Call S_m(K,1,2,2)
In 6th step, two tables which
contains 6 and 5 resp. are 7. Return
sorted and output is 5 6
• For function call T_w_m_s(K,1,5) [6 5 25 12

15] , one recursive call T_w_m_s(K,1,3) which

sort the half list and output of that function

call is : 5 6 25 is shown in previous slide.

• Do the same for remaining function call

T_w_m_s(K,4,5) Call S_m(K,1,4,5) for
complete trace of algorithm.
 Analysis
• Complexity is : O(n logn)
• Merge function runs in O(n ) time when merging n
elements.
• Divide step: constant time
• Conquer Step: recursively sort two subarrays of n/2
size. O(logn)
• Combine step: O(n)time
• Total : : O(n logn)