amiC Poqamminq

Onamic Po91ammi
DyT Pyoanic poq amoiq
S an alqoritbm desiq mehod that Can be
used wbeo the Solutioo o a Poblem Cao be

Vew e d as tbe eSull of Sequeoce 04

deciSio0S
Priocip)e of optimality
The Piociple of optimaliAy States hat

an Optimal Sequence of decisio oS as th e

Properdy hat ohaBevey tbe initiat slate and

decisionS ae the emaininq deciSiooS muSt

deciSio0 Seguence uoith

ConS-litute ao
optimal
eqadinq to the Sate veSulting 1 m t h e

PinS decision
A PairS Shortest patbS5

Let Gi=(v, ¬) be a diected qrapb d i h

verdiceS let Cost be a cost adjacenC
mali fo1 G Such hal Cost of Ci,)-o
1 i s 0 Then coSt ti,j) is tbe leoqih (ov)
Cos of edqe Ci,i) ;f Ci,J) e eC) aod

CoSt ati,j). if+ and i , )qe l).

The al paiS Shovtest Palb Pvoblem iS
o detemine a
mali A Such that
ALi,TDS the leogth o Shovtest palh
omito
 Let uS eiamine a Shotest ido Pa-lb in
G j The Path Oqina-teS al vertci
and qoes tbrouqh Some ioteroidiate v e ices
and lesoinate s at Nevtt1 we Ca aSSumbe
hat bis palh conains no Cyctes foif
bee iS
Cucie beo thiS
a
Can be deleted
ithout ioCeaSinq the path leoqth
f K S ao ioleioidiate Vertei oo

his shovtest path, theo the Sub paths

om iAo k and from k lo i must be

shoitest pathS fomito k and K 4o

espectively 0theroiSe the iAoPath
S no4 oioimum lenqth
of
VSinq G,T) o vepresent the lenqth
a shotest paBb rom i lo oin
eat hha
ao
Abrougb o e r te of inde qea n

K, we oblain Ai,T)=oinnin ki,k)+

keo
A(K,T)S, cost GD}
Clealy AG,S: Cost i,T) we can
obl ai eccuren e of
K-1
Find Abe a paS ofShortest path
path
Problem fo lhe qiveo 9a ph

3
A 2 3
2 3
O G 15 O 6
15
2 SO 2
2

3 3
3 37 O 3 3 1o

2 3
2
6
5 O 2

337o
(,3.oio C,), i,k)-+ (h,)
23 2 3
C,3) in2,8+151
(,3) i o 2,23

2,3)-
C3,) mio j),
3 2
i, K) + (k,3)
C3,2) min 3+9
3,2) min 1}
C3,2) 1
(1,3) min i,j), ti,k)+ (K,j)9
3 3
C1,3= mo15,y42
Ci,3)miof15,6)
,3) 6

C3,)mio Ci,j), Ci,k)} (K,j)

3
32
3, m i n 3,0+8
C3,) mio t388}

C3,)3
C1,2) mio { ti,j), G,x)+ (K,3)
23 3 2
C1,2) mio
u, 15 t
C,2) mio u,15
C,) 4

2,mio Ci,j),Ci, K)+(K,i)

23 3
C2,) oin,2+3
C2,1) oio 8,5}

C2,1) 5
Algovit hon A palhs (ost, A,n)
cos [:o,1:o) is -lbe CoSl adjaCenu
ofa aph oit
Iovevdices : A , ) i S the CoSl of a
shotcs{ path fom vertex

i o Netea CoSA Li, jj=o.o,Soi <1n

fo i: =| Ho o do

fo a l Ho o do
Ci,j7 Cos4 [i,i3; // Copy Cos into A

o K =| lo n do

fo i=I Ao o do

o l Ho o do

ACi,3: min (A Ti,i) :a ti,k3+a[k,jJ);

 O Knapsack Problem

>1f we a iveonobjecA and oe

ae KoapSack o baq io which the

Objec ias a wetqhl 'w S Ao be

Placed 0n he Koapsack ha S a Capacil
'M' heo the pofi Ahal cao be earne d
iS Pixi The o bjeclive is o oblaio
fiNioq of KoapSack witb ma Pro-fi
J
aroe d
Maximi3ed Piai Subjecl to Constvain

and is to-lal
idi M, whe«e 3n o

Dumbe of objecls i =o (o)1

Hov
The eedy methopd does nol wo1k
tbis P1oblem To Sotve tbis poblem
method
uSinq d
wil
dyoamiC PrOqO
pefovm the
Dinq
fo Il 0oin Ste ps

S+ep: The oolalionS uSed ae let :)

be be value of oplimal Solution
hen s' 1S a Pair of (P,w) wheve
P : 4 L ) and w:4j.loilially SCo,o)
we
Compute S -fom S' These
Cao

Com
Pu-lalionS of S' ase basiCall
Abe Segueace of dectsions made for
oblaininq Ahe oplima Soluliun
Step:2 We Can qeocate the Sequeoce of
decisionS o Ode Ao oblaio the OPtimal
Solulion of Solving the knapsack proble
el xo be he op-li mum Seque oce heo
two ioSlances x'1 aod
bere ae

x
Xo- o-2
sep:3 The ovmul a bal ase uSed uohile
1S let
Solvinq the koapsack Problem
Filyi be tbe value 64Op-imal

Soludion FilymazF;- L9, F -

oitiall opute s -
o,oy S, P,w)
CP-P w-) S
S S'us
Puqinqa ule
S't

ConlaiS Pi,w) aod (PK,k)

hese too pairS Such thal f P J Pk) ancd

theo PT,wT wi be elimioated
w)
Calied a s Dominase
Th:S Pu ule is aISo
Nule pu qinq ule .Basicalythe
1o shovl
Shovl
domina+ed JupleS qels Purqed
and
emoNe the pai with 1ess profi

more weiqhi.
Solvethe KoapSack inslances -6,023
Shown below.
wi are
Pi,
Pi
2
2

s o,0)
S o+1,042)
-,

s sus
lo,o), 0,2)9
S t3),(3,5)%
sS'us,
:(o,o, (1,3),t2,3), t2,5)
S' t5,4), (6,6), (7,1), l8,1)

ssusS

to,o),1,2), (2,3), (3,5), (5,4),(6,),

C1,1),8,9
Note hat Pai (3,5) S Pusqed om Ss
becaus eASSume ba P;w) 3,5)
and Pk wk) -
(5,)
Hee (PPR
lw >w):s lue
 Heoce oill eliminate the Pai Ps,T
3,5) from S
AS m:0 tue will find the lupie deoofinq
weiqhi6 (6,6) s3
,1 lo
C6,6) is oblaloed by addioq i30
S
S
So, 131|

Remainio q Capacidy iS (6- P3), (6-w3)

(6-5,6-4)
1,2)

Check CI,>) io previouS Staes

Check
s".
C1,2) is oblained by addioq i, to
So , 1

Renaininq Capacit is C1-P, 2 -w,)

C1-2-2)
Co,o)
So ( , ,X2,73) : C1,0,1)

SP13t-Px,+ Pa ,tPaM3
1xI+2 Xo+5
6

wiai<m
wi twX,t w a 2X+3 Xo -9x|

66
The avenioq SalesperSoo problem TSP)

le G = (v,e) be a diected q a ph wilb

ed qe
ed qe CoSt C The Naviable CC i i s
defioed Such hal Ci0 o alliand j

Cio
e lv\= o aod asSum n> a lour o
GiS a diectrd Simple Cycie that
includes evevuetCx in v The coSof

a lou is be Sum ofCoSt of edqes

o0 he tour

The TSP P oblem iS o Sind a Hour

of mioimnum CoSt

A ou be a Simple palb hat 9HariS

aod eodS al vevtel ¬ver lou ConSisis

edqe C1k) for Some kE v- {iy and

o ao

a path from ete K lo vevtet+he

each
Pa-lh fom Nestex 1 qoes hrouqh
Netei iov- { , k eiactly once

leoq-lb o Shostest
Le qi,s)be be
J
Path Sardinq al vesdex i qoinq-lhrouq
all vevtices in S and leminateS al vetex

The funcdion q (i, v-S 13) s lbe leoqth

a optima Sales PeSo loo
rom be Princtple o optimal-
C1,vi4)= min / Ck9Ck,v-i,k})}
2 kco
enesa we oblaio a (:,S) : m i o C i 9 - U
SeS
 15

ConSidev the directed q1aph given below the

edqe enqths aeqiven by malrid Find the
mio CoS ofhe Aour

23
1O o 5 20

25 o l0b
36 13 o 12
48 9 0

qCi, i2,3,4)- mio

ftqt2,{3,43), t 9 ,
2,9), Ciu+9 («,i2,33)3
g C2,3,4)=mio i+gl2,i93),t9 C4,3y)
gC3,2,4)=rmin fG+9 ,4),c,+3 (4,i230}
q(u,2,3)) mioiS44g (2,{33), °u3+9 (3,{23)
-

C3,43)= miofau+ fu,0

m i o (12 9 )

mio l20)

(u,33)- min y3+ {3,43

mio9+6
min153
q,f3,y})=mio + 9 (3,u ),¢+3 (4, f29)
mioq+20, 1o+ 15 j
mio 29,25

2,14)= mio (C+i9,0)

m i o io+S)

mio i8)
4 , i 3 ) = mio C eu+i2,6y)
io 3+5)

miC13)

C3,2,9}) = roio + g (,i43),4 9 C4,23)j

min
13 +18, 12 +13j
min 31,25

gC,i3 min +i3,b

mio 4 6
mio 15

g (3,i23)- min
+12,01
i n 13 +5

min 1sy
 16

g t, i2,33): oin f ,,tg (>,13}), +9 (,113)}

mio8 45,9+18
mio23,21}
= min23
9 C2,3,19) = mio t 9 ( , {3,), c, + 3,11

Cyt 9 (,í2,33)
m i o 1o+25,15 425,20+2*J

mio35,45,13

Bmio35

a)Find he optimal CoS o the touY

13

g , 3 ) mio isq (2,i3,1), S+93.)

qo,i)- rmin, ,+g(3,iv),,yt9,f39j

oio9420,9 +15 j
mio 29, 17

oin1
g C3,iu) min f 9,43
129
0

g (u,i33) mio Cu3+ i3,43j

9+6
15

93,
12,) oio i cSt9 (2,{4) S+94,i23)
( 2 , i ) - mio e,utiy,43}

mlo2+8

(4,123) =mio yt{2,9

8+5
13
9 C3,12,4) = mio13+1o, 12 +13 y

mio 23,)5
23

(4, 2,3})=« 1(2,31),stg6,175)

t
C,29) =mio iCat 3,43

15
C3,3): mio 3 +,43
13 +5b

g(4,,3)) mio fs415,9+ 1

mio3,)1
23

(,,3,4) min 1o 417, 154 23, 20+23)

mio21,28,13

Bina Seach Tree

Optimal

Suppose we ave Searcb a wovd om

di cliona and o eve equie d wod

we ae looKinq up i he dicliooary

heo becomeS time CooSumoing

POCeSS.

To Perfosm lbis look up move e44icieocy

we Can bu:14be Binavy Seaach of
he Comm0N d$ as keqword
wO«d

elemenS aqain we make his bina

Seach Lvee effictent.By anaoqinq9
eque9iluSed wovds nea o the

oo node l e s S sequeocuwosdS
aua from he ool Such a
 ee makes Our lask mose
binau Seavcb
Binay
Simpify
Sipi and e-fficienlTh:s pe o binar
iS Called OBST
Seach ee

be Sel o ideolifie
l e lea ,a,
, a,a
, -.. ao a

<a,<* an
Such thal a
toith which

P C)be he
pobability
1e
Search fov a be the
we Cao
element 'x'
of Seacbioq
Probabilit
Such hat a i<{a1+1

lo otber wovds PC iS the probability

SuCcesSful Search aod 9C) S be
o
unSucessful Search.
Po bability of
e will apply t h e f o l o w i o q foimul a

fov Com putinq each Sequence.

C +mi Ck+CKJ

, J-I +PS) +9 (T)

loitialy w G), vi=0, Cii= o

iveo a,,a,a3,44 (do, if, iot, while) P'SG
PoP PP 909,9,93
&s ae3,3,1,1)G (2,3,,1) ave Succe SSful
unSucte SSfu e S pectivelu ConSlruC
an

OBST

Sol we know -hal

Ci +mio / ci, k-1t CK,J

w ,J-1 +P CJ)+9LT)

Joilially w i i : 9i) i i zo, Ci;

we have to ConSlruC Ahe lables -fo

ValueS ,c, 3

wool |w 3 qu

2 2u3 y

3 34

loidially -lbere is oo oodes bJuo i i

o wbich iS alsoi J ov) J-i= 0

Wo09 o) =
2cD 3
 9(o) = 1

3 3 9(3= I

Now 1 node beloeeo

eoiaod i
So
So we bave -lo ConSlucl wo, , 3 3 4
o o0t PCI) +9) 2+3+3 S
2 o+PC2) +91) =3+3+)
3 tPC3)+9C3) = 1+14 =
3

34 3tPLu) +9u) - 1+1+1 3

Now hese is 2.
nodes bluo i j j
So, we bave o Consruct wo2,
ua24
02 w PC2)+9t>) S+3+| =12
=

W13 u tPC3) +9(3) =1+1+1 9

ty tPL)t 9u) 3+1+1 =5
No tbeve is 3 nodeS blu
So, oe ba«e to CooStruct
j
u3,
t03 +P(3)+9(3) =1D tlt1= 1
02

w+P4)+9tu) 9t141- 11
Nou here i S o d e s blwiii
S0, we have o CooSlvuC
oy o t Plu) 9 t u ) - 1441+1 16
Nou
Now we bave lo
Computt cand Values fov

Coo O, Cu:0, c,,+0, C330, Cqu

Sioce 0

o0 0, Y0, 9 , ' 3 3 qyo

Now , oe have 4o Com pute C&R values fov

j-11

Si-(o,O,01,2)C2,3),3,4)}
have o ConSAruct Co
we
C12,23,3y
C +mio C:,k-1+ CkT
Ok

Co otmiO 0,1-1+

woto+0
Co +o +O - 8

Ck=2)
C12 2 +mio 11,1- +,1
OLK2

14C+
1to +0
 o i o jC, , 3-1 3
O4K3

3
3
C34 tmin
34 y-+ u4

3+o+0
3
Fo
j-i2
Si-i -2to,),i,3), Ca,49
Nod we have lo Constvuct Co2,
Cia,Gu
C tmin C,x-+Ckjy
k
Co 2 in Co,1-1 +C12if (k=i)
OXk2

Co2 Wo2 +min Coo+ C2 y =)o41 =1

0<k22

o
O4k2

KMio Value 1

o212 t7
 20

13 tmio c t c 2 :2)
k3
043 3
k3
tmio
<k3

K miO valut - 3

Ci3 q 43 )12

=)0+3 3
2 oio
2k
k
3+o -23
C 3 t Cuu

C 3

woo0
Coo 0 Cu0 C22=0 33-0
22 0 C30
-7 33 y3
1E8 C12- 7 C233 34
O 2 2

212 3
Coa19 C13 12 y-%
o2 Y132 3

P C o 3 2 5 C u 19

Woy l 6

Cou -32
Yoy 2
wwmb mi
Now o u wi be the O0 ith Cos t oy32
lee
Tbe
h e 1e f1 l+ec of ; iS K I and qhl
S K,

Oo

4 int

híle
,3-
o,