0% found this document useful (0 votes)
26 views32 pages2nd Chapter
Fun
Uploaded by
shampavi53Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
26 views32 pages2nd Chapter
Fun
Uploaded by
shampavi53Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 32
Pe
Unite!
Ect
Funders ot ste Alt & of Agile’ 4
duis Frame ots The puoi on writen
an algorithm.
ficidd by meannirg tH porfarmenes dan ap
We can’ meaune te perferavance & an abpoathon Pip
compubrs. two fact’
p Amount of Hime saqpoud boy an algonithon 2 enceudt
2 Amount 5 rbosvage poyuitud by an algontthm:
This i popularly raven ar time conplity and .
cere Lalani
: n
‘space Mor open mpetng ji
comply TE Cant ele ae
a of apy thon
ng Meaning
‘me eenni
SEY eile ra) Gee
of algorithms,
Fig Rralyas & alganithons
Spoce Complexity
The npace complerity Lan be oS
Sf memory ve veyuired bg an algonithon to suum: -
To compute the space complexity wt ve Ae
factor : Contant ard jnstorce chanactensbica,»
The space suquisement scp) cam be given or
SCP) =Ct+Sp
where CV a Contant ey Pied. pat amd
j4 denotes the apace & inputs aoral eutpeit”
® This xpac Yan amount by opoce taken by
jnitructin, Variables ard, jdemtifuens.
And Sp ira 2rpoce depenctrt upon invtames
characteovtica, This 2a variahle pant exhort
pace requisrerent lepers on particular problem
jnstomee.
Connider three exampla & algorithers 40 compute
the space complerity.
Ex Ie
t
ALGORETHM Add (a,b,c)
[1 Problem Dencription: This olgomithen computer
// fhe addition 5 Ahoce Hemant.
U/ Inpvtia,b ade ore g floating type
I) output: The addition is meturned
wturn atbte
The space requirement $e alganition given
in the above example
SCA)=C @ (5,) =O
Th we arnme that a,b and c sccupy me
wond, rige then total wae tomas $ be 3.
ramp 2!
ALGORITHM AddkCX,n)
UW Problem Dercription: The algorithm performs addition
W of alt the cementy in an array, Array ir pleating type.
W Cepit fen axvoy x amd nan total number if elements
in armw,
MV GPa or oeeeroes which is bf dele type Heat.
@. 4 Sum <0.0
fox it ton do
gum — sum + xLid
zeturn sum
The sequirement fer the above given
allgonithem a
Cf) > Cn+3)
The ‘py’ apace suguiud fot x0 I, ene writ
Apace fon, one unmet for 7 and one wnit tn
Sum.
Example 3:
Algorithm Add (x7)
1 Problem clercription: Thin ina sectinrive algonithen
Wvsith compul addition § al the elmerti inan
Mosroy x03.
IL Trput: XCi7 2 of floating type totod nuumber
I demerds in am ovray.
I) oudput: returns addition 1 element Fe an array
yeturn Add (x, n-1) +xCn]
The rpace poqtuisement JA
SCP) > 3(n+1)
The internal stock ured fo recusoion includes
apace gor formal parameters, Locel variables
and. petuan addres, The space reyuired by
cach cath function Add woquirr atleast
Piorae words CH FO n Valiies + xpaca fot
pel adler + pointer tp xL1).-
® The dlepth G secwrion as nti (nr times call to
function and ome sehuig calh), The secursion
stack space wtll be > 3Cn+1).
ch etme “Comp dea ip
The Hime cempleritiy 3 am algonitiom bs the
ammount GB computer time seguisud by an
algosuttem ue sum Le tonplotion.
Tt as difhielt te comput the time comphaitiy
in dorama & phepscalle clockid fire.
Fox instance in melbiurcr system, executing
fime olpends en mary factors puch on -
¥ Sytem load
x Number of often programa xumning
¥ Instruction set wre.
¥ Speed & underlying pardumre.
The time complaity is Herefete given in terms
G peyueny coin.
Frrequei count iva court olemoting number
& timer & execution Fp Haterment
Fox example -
Lh we wk a code for caladoting rum fn
numbers in am array them oe com irk 6,
Fine Lomphyify wary preyumey count.
This prequeneg tourk dlenoter bour many timer
we particular slatement ur eaccuted. fos Ciz0, len, iH)
t
Sum = sum+ ald
}
Sdtotement Prequaney cout
yo t
ign Thin rkatement enceulr fot (nti) Himes.
when condition is true jes ikN i true
She enccution happert de be n timer, ard
the rhatement- executes once more eshen
jen a tole
ith n dime
Sumzsumteli]] mm times
Total
3nt+s
Ther we got frwqueney eunk $0 be a
The time Aompbenity as nosvmably demoted in
Fours & oh notation (0). Lene -if we reglect
the contort thon Let got thy time complexity
Ue be Oln).
Another examph
Let ur Lorvider smother enampe .
The follerasing cole is stpically wed. to perfor
matrix addition
for Ci=0; ian; 1+)
f Por f=; jen-gtt)
eCiIljJ = aCIIG? +e ICI; The freqyueney, count Lan he computed a follows:
» 1 =0 executes once + Prequency count =1
2 15 enxeceter fort ntl times
s itt enecater for n times.
+ [0 executes for
NXCI)= mn times
for outer Pov iniatialization
loop of j
+ § balled, average cae! pire, Comp lea ty. ALGORITHM SeqpuentialSearch( A Co. n- 17, keg)
I] Seanchu fora given value in a given avrag by
Uf vequerrbled, peach.
[[ LnputAn arog ALO.0-1] ond a pearch key k
1) output: The index Sy the first clement & A
I Heat matcher k ot ~1 af Phere ove ro matching
[1 oforrents
i<-o
cahil, 16N and ACiI¢k do
jeitl
iP ign yeturn i
ese yturn -|
If the element 'k! is present ab first Location
in the Lot (ALo--n-1J) then algorithm puma
fot a very short Sime pond Ploreby coe will
gat He bet care #ime compen ly.
we com olencle the bet coe time Lompleaity
Cort =!
worst coe time mpl
Wort Lane firme Lonphiai Bat bo!
In the above sanchirg. alganithin Lhe key oh f
kin sanched. prom the Lot Bon elements. Oo
Tt the beg genenl k dn preset abn" Iyray,
then the algorithm oll sun toh Aorgey
fime and, terceg cot will get the coor Care
Hime Lompharty We can denote the worst cose
Hime Lomphenity as
Casost =F
Arnage fare tire coop hat
This type complexity gives infommation about
puquivad or compute, vonage core fire compl
Let the abgonithm 2 fei sequortial search and
pha probability ¥ getting succvped search
n is te totol rucriber b elements in the. Hit.
The pint match 9 the elernent wrth ator
at 1% Location Horer pribabilile Gy occuning
frirot match av fe hr wo jth ehmmont.
The probability q getting umstecerpul search
4 (1-P).
Now), exe can find sauerage cast time compli ty
Covg 7 4
cg fl eck
Lion in the Thur we can obtain the gonoral formule. for
sempetng arage cet Bie empty
Suppose if pro that sneanh there i no seaccovfad,
pear je, we Jat tanned the ortize Lie & n
gbrunts and oth we alo rot pind, the derrnd
ebrunt in Phe Lot Shen im yuh a fuation,
Cong 6") = OCnt V/g + HC1-8)
Conger) = 7
Thurs the average care Puunnitg. dine compheri ly
becomes equal 1.
Suppor if P=) ies coe get a rucceenful search ten
Cong Cn) = L-Cnti/2+ n-Ct~1)
Cogn) = met
Zz
That greark the algonithm scam abot
i
|
| Asymptotic Notations
Oe Gren
Te choose the bet abganithm, we reid tx
shock afpicionce & each algoniflin. The epficiong,
wan he eared, by compuding time comploniTy
& cach algorithm. Arympltic notation 4 a
hort bard vy te papponnt The me compart
Using -eonynp lotic notations use can give time
fom plerity ar “fastest possible”, “slowest possibh’
et "average time”,
vaniouwr notation yich a4 0,0 and O raed
are cavlod, oyrprotic notations
Bit oh notation
The Big oh notation ir sheroted bg '0'. tha
a muthod GB reproprting., the upper bound
Aabgonithas furs time. Using big oh nosestion
We Lan give Aergok arnsunt & time taken
boy the algosithen te Aorphete. v O-nototion
Debinition: A unction t(n) is said to be
in O96) , denoted 40) ECOG), IF
1g bounded above by some Constant muttifle
of go for all farge n, veyit there exist
Some positive Constant c and some pon negetit
integer Me suck that
ten) £ cgin) or all nme
gen)
c
we tin)
goo! 8
ee
lati Big -oh notation ; ttn) €0G™)
Example ce
loon+s5 € O(nt)
Proof loont5< joontn Chrall N25)
1oonts S$ loin
lor ns loi n®
Thus, for te constants, C= 101 ,M0=5
we obtain joonts £ joi nt
Hence — joont 5 E O Cn)
Ele :-
loonts £ loont Sn (for alt nz) = 1057 < josn®
Hence for 02105 and No=zl we can
Soy toont+s € O(n)
@ 0L~ notation
fn
Definition : A function £0) js sail t© be in
(gen), denoted tome O49), i? oi
1S bounded below bog some positive one
multiple of go” for all dang od ol
theve exist sorte positive Const
Some non negative integer No S
LON 2 C90) dovall nM
ton)
Fg. Big-omega notation ; ten) € (gem)
Example 36 9 (92)
nen* dor all nd, je, woe can
Sele# C=1 and ry <0.
then zn" hence n?EQCn*)
é -notation
Dedinition A function t¢n) 35 said to be
in Oem) , denoted EME B (Gen), if tm)
1S bounded both above and below by some
@ [~
Los yy
positive constant multiples of gin) br all
large n, ie, if there exist come positive
Constants Ci and Cx and seme nonnegative
integer Ne such that
Cagem LOIS Cr gor fr all nz Mo.
cu gen)
Loe
Cage
Ne 7
FigeBig- theta notation : &nE O (gem)
Example ©
Let ur prove that fonln-!) 600),
First , we Prove the right inequality
(the upper bound):
nord) = gets fn < ae fu all nyo .
Second, we Preve the left inequality
Ctte lower loounel ):
batret) bot bn > bat bo}
& Amie ge nici 47 Chor al n72)
: Ip > xy nt
ie, Bem) 2 Y, ne 4
Hi c
ence we can select ca= 4, crzh | No =e. Some Ex hte Order
D bog n is fen) then
log on EOcn) “Leg in & OCA), the order P grmusth of
login is slower than vn.
beg. €0¢nt) vlog £Oln*), the order & grert!
ob beg 8 stewer Atian nt as coll
Log nd ctn) log ng cen) and iPa contin |
. function Pin js belonging do-0.C) |
Ve should ratify the condbidion
Br) > cw gen). |
&® similerly lean fa (nt) ov Cn
y) Let fin) = n(a-s)[ 0
Then
n(n-i)/2 €O0n) ~~ #0) >00”) ese gt £10) = nlo-Die
= oten
; : z
pty manimum order is 1 eshich
is >On)
7 Hence £/n)do0¢n)
But nlm le €OCn*Y As £ln)< On")
and. nln-i)/2 €.0¢n%)
Si mrilanhy ,
n(n-/2 Ent +. fr) > 0 Cr)
yvitn-i/s En (n*). 2 $ONS _o.(n")
n(n-Nled (m9) +0 £m) f-2(n")
th £m) ondr & gi(r) ard Bln) 2 oder Ken,
“then filni+ her) € OC man gicr), get)
2. Polynomials & degus m € O(n")
hak meer maximum degree i congidred for
the polpnomiad
For orample: agi Phanttasnte har the order Gp ypok
O(n) .
3. O61 < Oller) < Oln) KOC) < O(2)-
A. Exponential Pundions 0° have differant ordonr F
qrowth fei different values Ra.
Bre sie | Ravic Efficiency Classes
Nare
Clas /_
i rovtand [As input Sige Prows
* we qt tanger sunning
loan Logerith mic Phen wr g of Logsith
hat he
algoyithm alevr nat
ex gil ats
repel rexthot the
” linea
ont the input Sige 1)
n log ny n logn Some. instamee Ae Sorting the
input 15 comidered| toxin merge Bont
dor the Uist of size n| ov Quick sort
nt Jauadvatic furten the algeaiel
has hoorade " f
lonps then ghir
Supe § efficeiones
eccunt. .
Conmencts Cay
fe
time Hen i€] Seared op
; . Performing Sea,
Search openast
i
Scarmey army 5
s Gy ek,
donne
nhy snat vin
Clermont.
2
| Cubic | usten algorithm hay Performing. matrix
three rere expos
them thie type Oe
Mbicioney Peas
ser
type & elficieney ecush.
n} oviak Itelonanalyerithns j I
| [fee Useless
hehis ty
wucltipdication
7 xf y
Fd ef A Cteenabing ap cubs
Kgrevcth ten tes | & 1 elementl.
Gemerafirg al
Welborn ey, Permadbbtyont
totic cary ob, On Malhe matical Anolpis of Non vecursive Ayonthons
Example: Consider the problem of finding the value
of the Lange erent ina dot of N numbers
For simplicity , we avume Hat the lot a4 implemeddy)
| as an array. The following + pseudocode ofa
standard algorithm bow solving the problem.
ALGORITHM MaxElement (Alo..n-1I)
1 Determines the value of the langet clement
1) in a given arog
i Tnput: An axrag Afo..n-1) of veal numbers
I) cutpti The valet, of the Langpst cliomendt in A
moxval maxvel
maxvel — ACI
return moxval "y
|
eld ws, Jomo C0) the number of Eimer thik
cemporiorn , 2 navel wd ama try to fH a
jE or a Pumction of siden.
formed omprxirg
mab one Comperinen On each
vohich » repeoted. fe
5 vastahl i within
the bounds 1 ond noi, incloive ;
ye Therefore wre yt the Pollouing eum, for CO!
ne! 4S Ht tm be path
' cen = & | %
The an cov Cie to comput becuan
je is nothing other than | supeated noi bime-
| eee hur, -t
eee Sy =n-1 EOC)
jet
General Plan fox Arobyging the Time EPficiency
of Nonvrecursive Algorithms
1. Decide ona parameter (or parameters) indicating
am inputs rac.
2 Tdentify the algorithms baric operation.
(Aa vuk, it is Located in th, inney mort Ap.)
3 Atck whether the number & trots the basée
operation in executed aleperds only om the vige
on input. Tt it alvo depends on rome
Additional preperty, the wrnt- care, average
“art, and if mecevarg bet — care efheicienct ,
fave Le he investigated repevrately,
4 Set up a sum CARTE the nitmben By
mes the algorithm» barie Peration is
executed.
5. Using Mardand formulas and Pucker
ALM Manipulation , either finda clred
fom formule. fon the crn Oo at the
Vergy Mott , exlablish it, orcten J groucth.
We we wrpecially tu barte ruck of seimm
manipulation:
u
uw
= 07 = CZ, Rule)
“ Be ta
= (ajthi= Sars
jah ih fa
®
%i —>(Rukesd Ard. ure we two summation fotdnedas:
iy
(= a-L+l chore ley
&
are vom Lcserand
offer integer Hm)
nr mn
ESi=Sisisee.. ts 2H). Lnteaent)
1=0 it
>)
Exa> Conmder the element Uniguench problem :
Check urhother att the chmentt in a
Yiven anray,
S 1 elementi are distinct. Tha probhm can
he rolveol by the Pella rg Atraght forward
algorithm.
ALGORITHM Unique Emer (ACo.-n-/])
MDeterrnines whether oll the element ina given
Wartoy au dittind .
UW Input: An array ALO-.n=17
I] Output: Retiams “Lyric? iPaU the elements in
WA ove aitinedt and false” 9 therisire
for i<0 to n-2 do
for Jit! to n=I do
if ALiI= ALjT veturn bbe
vette true
not mel
7
Coowrse OM) = FB Baty Geei]
20 jel [=o
‘
Af n- nee
= Soi Sony E}
i=o j=o j=0
= (n-1) zy = Cn-2)(n-)
i=0 L
= 71) Onn) ~ Enen=s) cn-) [Ot (0-9)
7 cnn [ ea
x= @M-0Ne
ny
n EO )
=
a
Zz
Enangh3 Given two nan matrices Aard 8, find
be Hine efhicieney g te chpinetion- bard off i
fa computing Heir product C=AB. By definition,
C Ban nxn matrix whose vkmertr are conputd
as the scaler (dot) produch & the pout & matric A
ard the Lolumns G matrix B:
A
bel LE Jb]
Col.
where Clif] =ACL 0] BLoj7 4... ACL KY Lk I4
+ FALinA-1] BEn-1j]
for wwry poss ] indices 061, ena,
ALSORETHM PlatrixMuleiplicabion( At, t,o.
Hea bipeerd Blo. .n-i 0. n-17)
Yee matric
Mblopinition bared, algorithm eee ert
UM Input Tun nyn matrices A ancl B
WM ovtput: Matrig c= 4%
fox 1£0 to n-1 do
* §£0 to n-ado
Cli 1 obo
Count = count + |
neLnp)
aun count
Mathematical Analyrir, of Recumave, Agorithe
cep echt The the Lachrial function
Examek i: C
FECn)=n! Por os arbitrary rieonnegative ee
nN. Since
“(n-1)in e(n-]in brn}
nla
ard O1=1 by definition, we can compute
FU) =FOn-i)en with the fallowing reewrsive aber
® ALGORITHM F(n)
// Computer n! recursively
I] Taput: A nonnegeetive inkeger 7
I output: The value of nt
if n=o return |
dhe return F(n-1) ¥n
The bare operation of the algonithm is
raultiplication , whose neumber db executions
wt denodi. M(n). Since Phe fumetion Fln)
13 computed according Lo the Lor mule
Fin)=F(n-1).n for ndo,
the number of nultip Lications min)
nedd te compute it mut satisfy the
equality
Mo)=Mln-0 + I fr nso
vo comnputi — to multiply
F(n=) Fon) by n
Tndad ,Min-1) rucltiplications arc spent to
compute F(n-)), and one more mecltiplication
wD recded Ae multiply the rernult bg n.
i? nzo veturil
This tet, urs two things. Fixst, sine
Phe calls stop when n=0, the rrmaltert value
of n for which this algorithm js executed
and fence Mon) detined 0. Secend , by
inspecting the pseudocode 's’ oxiting Line,
we can ye. thot when n=0, the ith
perfor no muttiplitadtens,
© [ws
>
Thenefrw, the jnitial Condition coe ase aflen is
Mo)=0
th: call —P Eno multiplications
stop ashen Eee ees
This we suceeded in setting up the veer renee
welation and initial condition for the algovithon'y |
number of multiplication Mn):
Mind= Min-1)t1 for 130,
Mo) =2- |
we use method of backward substitidions
Mn) = Pin-i) 41 no |
=[M owt] +t = min-s) +2 8
=C™M (n+a) 41] 42 = M(n-3)43 |
= MQ&m-i)ti =...= Minznltnen
Meny= "ks
General flan for Ana
+ Decide on a parameter (or parameters)
indicating an inpubo Sige.
2. Tdentify the algorithms baxc operation.
. Chock chethen the number of tima He baric
operation jr execute can vary on olifforert
input of the rome we, if it cam, the worst
Lark, Nero, - care, and best-care efficiencier
mut be invetigated repanately,
@ ee
4. Set uP a vecrence vyelation, with an
app vop ria. initial condition, for te. necmbes,
R times the basic operation is executed.
5 Solve the vecuyrence ov, at eat, arcertrin
the onder F prowth & it rolution.
Example 2: The Tower of Hanoi puagle.
Tn this puggle,we have n clisks of diferent
Sige thet cam slide orto any of Hoa pegs.
Tnitially all the doks anon th pirt pep
in orders of rige, the Large on the bobtem
ond the smelled on tp The goal in te move
all the dit ty the shirc peg using The
écord one ar on aunibiong iP ;
We can move only one dk ot a time jard
it 1% forbidden Be place a, dosger dork om
fp ap a wall ove
The problem har an Tecuspaive
vodution, uhich is UWwitrated jn the
eolorsirg Ayre.
® pa
Fiqh Recussive solution tor the ToureaR Hames passe
To move n>| clks fiom peal & £45
(with Peg tas auxiliary), we piwt pete:
rec wanively n-1 disks from pol #2 Pat
Cusith pig 3 a ausidiasyy), then qmove the
dik directly ftom peg! te pegs,
and feral p move Pecurxively n-I Asks
fiom pegs Lo pegs (ing peg] an auriliary).
Th n=1,u% simply move Ae sarge
dik irectly from te source peg ta the
dentination peg. 7 iol choice
The number i doky n a the obviow <
for the input » Hae inductor, ond 20 >
ciciy ou ok we Be cee es
oe Charly, , He number Fe moves
ncn) depends on n only and we gt He floc
socurvence equation fot ii
fn) = macnn) +1 MC) dow n>
men)= 2 MCet) +1 bansl,
moi) =!
We solve thi recurrence bg the method h
loackward ubshtitions;
Pn) =2MCn-1) +1 substitute Mn-)= LM (n-2)+)
= 202 m(n-)y+i] +1
= LM Cn-v ter) substitute Mn-n)= 2 Mme
=2°[2 eens i] ee +)
= 27 M(n-z) +e tet)
generally after i rabstibblons,
pan) = 2! M(n=i) + ay Le pet chine
Since he instal condition inpocified fa 1-1,
ctich ib achioved fizn-i, ce get
_ yn
Mina e M(n ~C-)4 P_,
= 2 my +
Feet a My
@) | le xample ee
AlL@oRITHM Binkec(n)
/ Input ‘A posifive decimal, integer n
[outputs The svmben & binary dighs in na
II binang poprocrdtetion
if n=l netanl
else roturm Binkec (Let) +1
The recurrence selatin :-
The number 5 additions made in computing.
BinRec(L W2J) AC Lanse), pls one mone
addition ix made by the alganithm de increase
the return value wy |. This Meads te the
TW ENE |
ACn)= ACL*/21) +1 foe nd1
Since He veewrrive calls end eaten n is ogualtl
and Shere are ro addihivs made then, the
initial condition i
Ali)=0
Let ws amnume nark
ACL) =AC2S') +) Ax Koo
A (2°) =0.
Now backwarcl xubsttition:
AC) = AGH substtete AG) = Ae S41
=U) +i] 41
= ACE) 49 substitute AG YAK pI
@ =e ACL Ey a ee
ac Aa ')ti
LA ae
Thee we end up, AlsX)= AC) +k =k
oh, after pohvenning fo the onigi nal variahh,
m= 2K and Aonce k= Jog, n
A(n) = Lenn € OC Lag»)
E, camp tei~Conpubirg the, nts Fibonaceé Number
Fibonacee numbers,
O11, 8,3, S BIsersy,. -
that ean be opined ay sinple pleltirance
F(n) = F(n-)+FOn-e) dor n>,
Flo)=0, FC =)
ALGORITHM F(n)
Computes the oi" rp snaces number Peeurvively
Wey wing it3 definition
Mo Tnpt:A non negative, jnitogr n
M oudtut: The nth Fibonacet riumber
i? NK! retomn
be return F(n-1)+F(n-s)
@® Araby
Aln)= A(n~!) +A(n-2) 41 for n>)
A(o)=0., ACi)=o
The securnenc, A(n)~ A(n~)-A(n-2)= |
4A simile £9 F(n)-F(n-1)- F(n-t) =0
WR Lar sewsits,
(Aen) + 1] -[ACa-1)+1] [AG] =0
by substitebing. BOn)= ACn) +1
Br) - B(a-!)— Bln-2) =0
Blo)=! 4 BCI)=)
ACa)= BOA)-1
