Scribd
Upload
47 views23 pages

Daa Unit 4 PDF

Uploaded by

Kavya
Copyright
© © 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
47 views23 pages

Daa Unit 4 PDF

Uploaded by

Kavya
Copyright
© © 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
You are on page 1/ 23
BRANCH AND Bounp THE METHOD: - “The deen branch and bound yehews 0 All siatespare. tree) search metiods in cont al childven fie E-made are generated befor | ey Ke «Tode. can a. the E-mde- pe" have alacady seen coo raph search eeonique am ame BES and mae i m eahich The er ployotton f° new pode cannot begin unh?l] The pode current! bang expired fs ty Spore . Both Utes jo branch and 'boung lak a Th Branch ard Bound -faminglogy » a BFS tke Stole Space seath wil be mately" Letet In Tha out) georh as ‘he Itet Ive modes © 9 Fim in fret out: tor queae)- A "D-seaach like ‘Bole spare Search col be called “wen” clost Th viet out) georh a the tet tive. nedes 18a. lost etn. “AMT Got Not Cor stack? astm, The cose bo cetacking . boundeng funtions axe used io help awid ‘the ae 5 acl “rat di Nal tontarns an ansuwa rode- 5 Beane. ee Teaver E; Prog Sra rT ; : eps, cat G ¥ thone ane ncibies anel 4 ma if to othr city f given i is ed the pee such Bot cack thy visited exactly Once ond then netwuring- te Atdting eft, Completes, the tour d aed Jy dypl transiting Salesperson fuotlern ts represen |* weighted paph — Conbitler an dntinnce for TSP f& given by G an, | oc 20 30 10 re Ll iso 6 ¢ 2 G- |]3 5 0» 2 +# 99 ¢ WB 8 3 of le + 4 16 —y Hee n> 5 nodes Hence we can drour ao state Space tree with 5 nodes as shown & the above figure - CPT-0) 90,9.92, 99, 22,929 Pe Pe Oe dao @SOO ¢ 4 Oo 608006 ze je jz js ie © Gee is le je |e |s fe Is. je le le le lf |* OD OOOSD DH @ OO DY DLO OQDY OOOO @ =z ‘Wife € he Swe sd Cw Y Wey = WA f Re © Q © : Sy 5 E we € eSaeat uy ) and return to soot - ex & jo node ¥ fi 8. Be: Sees ad Sy twanch Orc Wound Hokey eat tach "ad computed * ae had, iy ing the nade wlth epbhuun cott Hence - C(x) = cost of toua(a) where, %—* Leaf node e'CL) + app hoxtmetbion cost celory ‘the path Rony Site alas | Row Repwersers * Yo wdanined sehuhg of TSP using i phwmach we will reduce Ahe cosl of in u, by salng fotiourtay feturnla, * Red-Reus (x) = [ ymin {yl tej- 16 2 " O 9 [2 © irenerte Repwerion.: |} [eat est | | —"|-¥ In the aloue trample ; aye e@btathe the graph Aare, h the oulgnat é oa iS+ het means all torre ie | a feng atleast 25 J EL ; eT rake the chotee of \ q | J | edge fay with efitireum cont } e f i teu - 5 gs see a — woe of HOt, attr Ee LLUy, Step-_ + Pea & bine dpaee ee ding sok HOIEIT= nf Obtain Ae j St nag and Je column. erties O ag fe optim | - + cost of connespondg node x with path 'n) Siep-3:! aac]. test + nredutced cost + HEI gerunade © bet nace Sei akaaumn ak a Ele pure See idlinen « Raped atep 1p 4+ PA Completa wlth @ftimsurn cost the pe ah Kem spe tag 0 ec! - note Bs ate «ng He Tool sete b colerslate the or ai ~o 20 jo «10 ii} ir © 1 + ® gr ee tet 9 6 Bo 4 oD fy? fovtion of Stele, Space i fans and 2% solu Consul path 12° Hake set H{2J[1]- © o 0 2 co wo | Fangte o © I 2 o | ignore oO 0 a OO 2 or fgnobe 17 [2 | Fen ote NH mw 6 iz co | L bE v L L imore Fanere Hoe Hntre Fonone (grote Cost of node 2 25 + 0 + ie} ih t + neluced td value of Cost = GE HD Jez] a55- mm oO: And » Modee st FOU = Se 0+ 3 column « ra] and OP 0A wo Oo Oo 2 ° 3.0 0 2 3 2 oO o © p - 4 +i t cot = CEI | Consile, path V+ | oe om 0 oF oO Db o H © o 03 0 0 * mw IF @o 0 oo 8 @ be ott ade & = ast oto =a 2o 5 5 Considat path "1S | eo 0 0 2 jn oo Wt Cl aac ; 9 3° @ eg me a a 2 9 ‘0= 2 is of stad space PE wo ile, bee nade opti Gt Boas ab ee a6 BRR? BROS 1 Cot of rele G = ard M[4,2.] = e7t3 = 28> Gonaa path Ved foe rate #8 P ilad ge Fl fo %- tek 4 column, rd estamn bo «St fet HEHICI] = H(3IC1) = 90 A 1 0 0 12 o © « O 2 Oo GZ Oo OD Oo @ CO OQ of Wie ie; BOE - Ut pede t= OF 413+ Ml4](3] R412 “ Conuilor path dpa. dev nade: as ft oO 0 ap S IR 0 Il 2 oo am) 0 8 @ G8 oOo oo oh Oo Oo @p 02 Cost of node § = 2s+ II = 36 36 28 Fig: Pectubn of Now as cost of role am G-nade: Hence gentitte chitdsen Te! Conmfelia pot 42,3 40 node F: bet aM row to. to o- get © column, fet MELE], weLi27,H{43) @ 9. 2% cotunn ancl stole Space tee . @ ft mbhibuwn, nede 6 Lecomet | node 6- Nade 4 and 10 ane children nodes of rude ©. i Pros, 0, gmt column 4 ieee oe eo a 2 a pam el 1B 1 «© 0 @ 0 t It tot of mode 9 = 28 ; 4 13 + e Optimum ctl edacectead — 2ST = §2. Cordldes path |,2,2,5 fer ned “ i, 12,5 ety fet 1 now, tour, Fug to 0 - fh he tN eotumn pan éotumnn and 5% Column to OG-F = HCl) HfAesJe oo we set path A 1 ag 1,4,2,5,3- Je complete chun te | Hence the State Space we EP, + Sette §pacce. pee - Fig A dence dhe ophmun tott of the Four fe 28. a : ack ler. Using Braxb g Bowe Of _krapeacs foes ae roto + ee rr gd “tits tues te. use bounding. fare Subtrea- i : i px ae athens, “foro ies in & different ” s : cy can be ee cls, bis ea {a bpd thesg “the Search t— dé bar branching qunction whi BV GE has a boundin dunetino, whic so fenstbiliy be 0 a rea to pre eAictently eee wehave te coleulete ine ¥ latest bound, ean Tobe doarmt take. ey ttiond froblemy— | Ac we kmw oft knapsack. (s a bag in which wwe shall TE wily Some wetyht obfects whose Weight awe leasthio f -eoual te meget Weigor | = ney Ruta k Pa Pu= (to, 10, ite) (1,609, uh, Wy) < (a4, 9) % meh Cenapanck use ight) pei atts = Vole ystonject =1r-2=18 | OPP bound = Wht ‘a don yu =4 =lb4io4 ne i on =13 ae nt =3L gid a 24-6=3 Se { ©" (parte bounds 10+ 104 gh et i meet Wert Bie = Bk. } Comestion man a Yate mo, we get 1K UB=-37 (Bo as. UBs-3e LR=-38 wef =0 a LB= (3) 3) MBe-2e LB=-2 (B= -3.. i el Fy ees . for oder, _ UB= AARIB UB= RAR= lota=a9g = 17 = lotloq\n . (6= to = lO-o+ ene ANE Corte main B. lonoHatSwe “a ene 3 236 =a. calculae ditleren a blw (BS UB in Noda & Nok 2 for node 8= —3.+(-3¢)< ¢ node 2 2 3. Ghj)eto Ew nede®) i eae : d=0 OR= Warp. ANOt hE don nodes) pie bi, moe —UBs It? = ABS Or DHT Bat Aa\e = al =—3k LOe joriat af =-3e Dt\ference tt rT to Mode y , d= 32.48% = 6 For mde $, d= -p tee fy for node t, dir node, DB= letis tie - UBe lox tot = 3p = 48n ’ WE 4+ Ay Be ADH § vy Fils wae “2 =-377 ! Pierce | dev Mode 6x ~ar+ax og {ov node Fu-Bf438 =0, jor ode © %y=I tos neded, %y=0 OB= t+ fo220 [Be 10410 =20 UB = 104 fot B= -BE (Be 1 Hoty xls =-38 Difjevens, fore poder , = 364 =o ox Oude oy | = -lofts0<0 Ak -nede@, we got moaritrnuns Upperbound , [3 28—2yu —3 4-28 ATL hy My, on | ye Malmo py our pees, 2 my PA ALPE A My Pa Fy Py ! = ECs) 4110) + OCD 41) = (OF IO ATE x, I — of { knapsack prbler ate branch § Bound '- Ala L Bound Cobol. prnjtt, total wt, b) A total parit denolés cument poi A total wt denotes cement total wetght “HK denotes tinder A, Rirmoved object- | eid de “represent profit ot abject i. ! ‘ Ie the “welght capacity q knapsack pr <— total tit tov (i ket ty a) do we Hae wl on thon pe pk +A ae cpr ai Gorm) uted et) ( han pes . y Bound (total proqity totally kyrn) a profit denolés current pryit i] jota) we denotes current total wueight Ik denolés fader of nernoved object |JwGd denote weight oF weaved object felt “apne papi: oF object 1 J is rhuerghl: Copactty ot Koaprack { Pr <— tolal-peofit wk < total-wk; for (TERA to odo if (we-+ wo) carn} then { pe < pt-pt]: whe—wt tuff]; 5 4 Fehuun pt, 4 b (Consicloubians§ —————] ap hnaicanee—ie aa | => ane Can Pose several questions concerning the | performance choveckeristics of Branch and Bound | olgettin's that Find Leack cost answer nodes: | 1. Cree the use do better Ateatting value for upper olkoayy| clecvease the number nodes genevalta ? =. 4k is possible to decreme the number by expanding dome nodes estth TL supper? of o better B always nesulbine® | | e in) the number] d nodes ope 3: Does the use decrease in (ov at deast nok an fneweas! dq ‘nodes generaltte ? 4. Does the use of Gn the genevadion d move nodes dhat be generale ? => AU the Following dheovemea Gmume dhoat the | branch and bound algerie ga to Bnd the minimum | | dominance nelokiens ever nesutt would athersise | Coat solution node. | Theorem Eas tek ty be a eholk space bree. the fais of. | RRR i | nodes qd t generated by Figo, LIFO and LO branch and bound algorithms Cannot be clecreased by dhe eupansion | q ony pode w with €G5 > upper where upper is dhe | Current upper bound on dhe cost q a mistmum- coat | Splukion node In the treet \ ‘Peco? :- The dheorem follows Ram dhe absewation that! cthe Value | upper Cannot be clecreaned by €xpandling ys x as CCW Upper). Hence , Such an expansion Cannol ablect dhe operation ¢ the algorithm on dhe nemelinder | A Abe hee” | Theovem 2 let Ur and Us, U,, be two imbal upper hounds on dhe cosk qo minimum cost Solulon I nede in the alah space tee , then FIFO, LIFO and ic | branch and bound algeith ms begining with Up coil | Geneva no more Node then dhey would 7f they Atenled “ tyith Uz as the inital upper bound. Proo:- dP they have sented with uv, they might generale nodes. Theavem ‘- the number a nodes genevaled during a Figo or CIFO branch and bound seanch for oe cost daolubton node ney fnevease when a Shenger clominance elation fs tecl. Prk Consider the efali space ree the only Adtation nedes ane deal? nodes: their cost % covitten eulgels | the ‘node. for the areal atn Nodes dhe Numbeyn Gulstce each node fe ia ‘ce’ value. the tuo ownuvt | SONI He AB BRS. (Gi =P pal Sen oaks Clearly D2 78 slvonger ¢han DP, and a ec general uaing D, nadher than Pisx | fa alco =xample :- By s - @ . od Abobodidd ( D ODO LKOOHH 6 @ HH Theorern i- 4P a better a Lunction te tued ina le. byand and bound algoridhen, the number q nodes generalitd mney tnerease- Proof :- Consider the chalk space bee . All teof redes ane Salulion nodes: The value outside each teak is Hs Coat. From these values %t Pyttows that clr) =claj=3 | and c(S)=4- outaide each d nedes 41,2 and 3 ig q | | Pair q numbers. cle & 18 0 belter function | a ij chen & - However G 7s wed, node 2 can become the E-node before node 3-85 (2) = F(a) | Ee alt

You might also like