Scribd
Upload
25 views

MOM Post Minor

Uploaded by

ag125aa
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
25 views

MOM Post Minor

Uploaded by

ag125aa
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/ 39
1219123) a ~ t “we can apply Gractre en Date... Page cul ORM TZOH OV) classic miro/max convex /Covicave = se ge etaom — es Pex) = 10K + pows* + CoS( 2AM +E *2) tema St —> Non-conve® a - ot clescent, as rt 15 ayy RY mained _ Function zB / YT] Wf ae = fool ep MUN ne z t Gradient TeseaF MAY BIVE FETS Voy : ™. Rossror) ity Por trapping tr Loco ORAMUM. convex Functen #°s —> Uuntmodal Functen. aynammic heunsHe wl alse not user mii CASe . Ft wii also get fappecl 7p 2ocol opnmals. Metaheursiica —> wri) work here tah MR cag HE SEE Solutions - _—|—— — es . evoluticnary competing : i kehasiees Bears Approach = Genetic Algorthy + | Proctpat? i 2 Survival CF Hoe Fetteat 4 Ppanvrw Heect | prenctjoie cl 7Nherttance. General framewort! single soluton x Popwotion of solution + Sol) > Chremer™s. begin iEntrastxec) po a. Evaluate c pct?) 3° generation 4. while not sents Prec G17 poo!) Selector C pilt)) mate peer = Evoludron. cele) ) Stze pool = Stee evoie ce te)) APF AT SOT OY GenerotHon selection Cpe), prce)) t 2 Peers volume: rath > Vv - 2 SCeiume) | Kedyan c cler a | Reol | Sh __ @ltmension) | | Bina coded GA bodes ft Sa SelM 4 chromesom [C100 = eloeco [oarescio} Ee tere oe or may not sane Ftmess cf solution 1 may be same as opectve Fitness = f* 0°64. Glo) 4 -ceost = 23. V=s380 Ptiness functon Randomly gem 7 Soludiens - +++$44+44 4 44 Evaltuscue 3 Frtnerr. | [a3] [a] +29 cons fed nect 1 Vtolotrer) [eT] 4 T= Tournament selection Bihan Teurnam ent selector J a TAO Players =F [ireae ed TF) es ea ae Pe joc Eat ee ee ee BA te tay rfzl =< pa] [4 t _lglas | Genetic Algontam : ’ begtn ]. Rortalixe toe population P | Randamly gered. sy 2 Evaluate Pp” 3: Whine 4+ Using KKr Mon nD a is leg [23 Log Cit QRy a _ De A Sith a Z B=P 3 Ryze ——- ~ eet ———— a = ) # Genetic Algorthm: a > Fnrhaltsaieow 2 Evaluction a] | 3 select Crating Pook) y Wr | Te PAsert good) gndlrvraluods Fa the _| wenrerotion . eo | ~— + + Tournament selection: 1 —-| | TE x, BS CONN? =a 1 Sty os Siar ed pea es fas Friness of x, 63 be 1 T = sixe FF tournament # partteipands ="K! AL \ Ki Gorerns +he pressure Selection \\ i Rt Pe ere Leads Stagnation 10 eocad “| KE pessurel > eltwersr Fy 7S Ath COVEGEN&e BS Slow. 8 ‘ Fy, | Rowete Mee! solution: Gr) _— sol FrtMess Bo Co BE Bo 26 Ae ge eu i alale|> RRR or | er= Fr eae t=! Procedure: ©) -h = ranctCe,t) S00 for each gnolrtdusdt | pees Sf wx, Tnetevttluod J: | Se =O yen e, r= 0's eh Sos ¥ ts clone USINS Of4 T | PCa) = 100K ~*a) + C-#) } pa © ESS) is eee wa apralat [eel ee a Chromosome lation ak ap ___* cress Over* 3 Ma5n operctor 197 GA ~ used we ge enerate chrid Populod green porcar Population? = “single _pornt crossover operator : | 1 sefect randomly RIO endrtduals fr popwledioya 2: Randomly Choose a crossorer por both ?*orvecuUads s- Exchange whe gerenc moacderrat befor anc! after crossover poret 1 both Hy solinions. i » | RCGA: { an * parcnti) P ! [4s 23] 3alic] o< | ae 5 a | paca)» [eo faale[as/21 | b | 4 Lame | : creseorer pret iio PLR A — _obriel L: Sea [i ere er Fo es Pie hd & $2. | Explorotion: explore ail the Sol convergence 1S Slow ; opttraay net rene Texplerrasio®: Shen we know the i fe. Zoi] exPIMS IO He Victntpy of x Convergence fS Faer, Trappecl tn 2ocal mimarma exploration: we search new postiHoms 7 We search space. we explore the uanexpliocd region - Avora Locat otimur Pathic osctiiore , fT ttme to rach oepitma, EE =e “| convergence ts sland, . Explottation : we KNOD the best so)” Hen ae artll search an the vietnrty of tHe cUcady | Known — BalwtIOW speecl up, comverpena frappecl wn Locod Beene for efferent cigemthy - proper Tcorplot nadie) oF expucradion & = xplottediey msl we Need to -expaiore the SOlUHONS AS vell a we ness! te chare best. GaeecoT 7 jor HOV? Milearpossove: oro le for exp Blecenas 2) SPS majorty res poner late For] xjaloredor, leeeent ; t+ zssues' - Pp, = (22) \ c= (4) ' Pp, = C44) Cp = 64,2) a { chricl popetoto worl) [fe 90 the bo undor, of parcrss, TE RS reshicH © the explore tO u @. f Lack of Orversmy + chricdh Popalaken ail # an the ‘region “hounded toy the parca Solution . eT PT Ea ee Le a Ge) re at oe = as =) : es Cae) +7 K€ Eee] Seek age Acta nS SC. ied - 7 01D) Sm tegen GMAT acre 7 T consrainst hanolltng rs cltPPreath and rae constraint vtoiation may arse Prequé I crossover Operators: Meant for exploroHon 14 single porar Cross over a 2 #nolrvrcluods b-. Crossover porat Cc swap the (Gevere maderad Drawbacks: @ rack OF Sttyarsthey (2) Consraset vrolohens a Arithmetic crossover: Process: @ select 00 enalt AUS Cparerts) @ perfor? arntamettc combination of those parents i Gy =)kIe tan) Poo) P= O° Comes pee ren Oe) 7 | “ae CoD i : P, = @3) i | Pasa Se) aw=oey | Cpe Oe 28) 2 6 4 acer = (ore, 1:24 (2°63 j Fe = O'S) (23> FO DOs) | = 04,422 Chi Sat Cons, 2-s) =e: Ors SI+ CES Crs, 4) =O a) trae =e fe | t Te " pikes > RRS Ras) z = = = : 7 + Lack of rrersry = _ _ fe : _procluce ve Pspangs Sted ap + toe locorod 2tneary berves, —_ rs 5 7 if T T: the porenis | op : ss ee — 3 P —— —_] at << = - 1 ~ — “ 7 =F : 3 ; 7 8: | Simulated pray crossover CSBX) te SIM UlodeS Stee $3) Pr Potet binary Crossover AS) Po = RCGA ~ CGA es! e, = S| Cae mtta = BBtaC L Sy Average value © parents and child wtll be same. ot esa - lemmas The average oF clecoded values Ff binany Stmges before and API crossover are same Ce capeeee sory srogie porrat Crossover eT oa—— lelrleleol ti] ee afa® alad'a® alae? se Teele o|) £18 t ay >t Bt aim 1 . i ‘ ' \ ler? | fofrfe}ije\! ] | eet 2eftL most bTt 18 the highest Stent ercant (Sea re @ binary eNncocl?rE » tet clecoded value oF tet most K ralUues » a | 7s Ar ana @-K) 7s By) I [Ye ee Peitwene(Bpeeayaretss) | —@O sarees \ tot Ya 2. (@ptBs)2 tetra) Ge B Baton = 2 Want @ are same. snack of Orersty _ ? a cee: wes +e 2 produce efFfspenes wat cr tocord Lneary _ bere a 2 ‘ten the pores ——— ?; “ simula = Crossover CS8X) Stmalaks SiR 23 Py > [tfepa ofe ao potet+ binay Fetal Crossover Ce el ey ! RCGA ele: [TelTols o[2] Se 43, = Te TeP eR] i ee COPA Sy ee | =a ez Be i Yitde 2 F8S¢45 ~ ey | t a x ‘ > pe Average value? parcrts ond chiid wri) 6& same. a + _BCSA ~Tenmai_Th © average oF olecodled! values F binary ~ Stings before ane after crossover are same singre port Crossover proc®: 7 > [etre elt [il afas al et xePt most bTE 18 he hrebest Sten? eaar bre 77 brhany eNcoclfrE - femelecoced Svaluc of rtebt most K ralues eg A; and @K) tsB) x= By) Qk +A) \% > Ba att45 = By ak+ aa = B,2*k 44) =H WFR eS (Bitba ak tires) s sear a 2 a i+Ye< Ba) 2% +@itA2) a ad @ ancl ® are same. 2+ Spread Factor Cp) p= |o-c | = spread of of esprng Shustop Pao -Pi Spread oP Porcnt => condrHen : B>IL q crossover 1s Sapaney a Reinge occupied ky. chtid popwoton ts greceer thon the parcot S68 popwiosn ran ze | => e Fy, ; + fee , expanding, AF = —a Se ia, ia aly B 4 | To TT o property ( © Average Property - ,© spread property —__| © select two soledions —T@ a = vs CCP) P +G-B) Pa) ae ca = os (Citp) Po +OP)P) yt Co Py Pal Ppreperty: 5 else Ne user Aefrined value I z = Go te Up were else. | aE arya ” —~ ye oe (ere) pite-xol] (a= " ¥Y, =e [Cw +p leo-7] ) — ZX D> parerr = LY 3 chrict -“ + | Pseuclococle: a [Znput: Parent Popurotion a | Poi Crossover probaltt ty p= (Hp, t1 7 | Ne | User deFrrecl value xe- | | [Re te 1 to Ts fe ranel Cort) < Po for G1 tod > generate Us! & compute Ry! 2 generate <4" & compute 1g Say = 0S (Hay +)- BPG tu)\ x1 =Cxit, Xi -- Xi) [os Sy = #5 Oat ay) 47" | tap >K5 | Rate Car tea) aS Ye = CY, Gig et) = EF Up Mepet Ne = CYay Yoo -- Ye) Yog © ay Rower bounce ors = fF Yay > %y MAE jo ettmenstons _S > Yog ay MAX ae = 2 WAS 2 = a eS ee | aye be ssices Upperibound Yi og | = ey a Xo) | ay — = —— b———_—Surmendions. = Jo atmensions: Po example example i zu a X= (-2°393, -7 79) Woo Hat Xa = (79742, 1ISY¥ ) Pe e'”9 Upsrancd (e,1) Qc = 1 rand on) = °’6 U) = 0236" = 0-dsy Pu + Gr = orn (= 2393 t Pa) — 2r9[H8 MAF? 34y by ee Ah" = oaF Biawe t | Mutcettoy Operator: Explorrastion Polynomral Metation Operator L_, pepwotion ; chticl poputation => Select 1 solution | > Mutation probaieris hy for re ton re rand Con) = Pay Por Ggertod gererote Dy IHG AS ONC xp — xp, OP Bounce) checic ud a - Multloezective OR mpr@dtion rvs (6]iefo02a -G) #agran, jagsrcot opinmtzasionn _—_ a rere re | Cea | | [meta 4 heunstic: GA. - aI tie a at cere erenssanpreen cninsiisctiabilaiiii —4 ———|censnfoma Por Classical eprmizaton “| Be Funckon : Nosure OPP. ComnnuaUSs ] comelttion on domain of fun = 7 ef olzscrcte %7 thot care we Cannot apply classtco! opt KxokeN) « Ga alent leather akoouk Hee clomain os tae 7 function te gnadejpencient ef nNorkuve of #7) = + | stngte objectre OPATINROHM : Problem f manpmart _ #C>) - Sh: Brn me) ) E mone than one cages FUNHENS - 5 ot * Probiem: ee Lomaxsmrn FOO hE — List ©-@ ‘J | C oe Fol Py a Hy Lay 9 SIMs a { Loe’ take optrmal value fm, __ "Feed Gsm 5 Teco _olgeotve #Y— Br-obsectie To desrgn a car obopeckive object res : spe i | ™m79 “cost” a2 | Max “com fort” [el ete a re , Fe [ez os Pe aa opyecte = fe — Sa ee Sen eircng | j a t aa i I searchspag AEST? Bparce 4 e i Two searc a ee + dlecton Spee: Randomly set? nati! 2e sel Cex. GAD ? Goal Se Find x, OR 4 [xe peccle Map Sel” Rens clecise7 spas 4 . obyechre Sac 80 wore know yr tSiay pic Peren@ cOTte~ Ho Ine TT Knew pecal to con}? are £t0e LIES. ee ohh DO pe Qaieat Sirer eegect as Reaembenrs (Maprre__* ; Se He Colne =a vare = clove “fo poleos 5 Sofe=HIORs already enerSh | | | Pormadtx ation : t obyect ve fr mPn we 7 CANNEL He | Se}ucHiory ___pareto beim Sd/uHore fi | (0a operat ar TSO OOS pare aT i pelze23 | Aes [Mut obyechre Opn miRohon) Difference bro Mop & Sop O Mast ple Copectnes eta pena) [oErAIch ip) @ tadtale Solcrons e eae he TWO SEACH Spaces Tin proving. an objecté, NS aeee net sue tot aba Lim prevemery4 of anit ORISA ve Space = jay comvergence : Locode tre solehons Clove to « ewe to optim Spaz. o> Drrevstty : Goludtom Should be as ApcetHe ao poserble ever pareto Fret, we Showtd fy te cover tye ortre parc No artifical Preup” te requrcd « eerie el en la ee m_olapectrve, feinchons. a gum or Lear Combtnonhor) Om Ss | ard 0K 1, War 2 Te ee E FONE Ce nicon (eat ep em Pe rt | = gevel t#0-) tl Pee erMaly +> a IY a rs beHer than i a £ ae f, (xz “a x : (aur a 1 = | tyCe;) ¢s_betHer than, Acta, i. filxiy ts beter than Fi(xa) bat 2 can't decde |fa(ta) ts better tein far) a ae concept & clomrnance : pemrnance Relctten “< dominane reloton e eperntor \\\ lee Kes ky CX domtnates xa) Pee 7 Cas es Py Se kaestesinin AiG. = Pole) i 5 L a) < Se) | i a ‘ $n domtnades Xz f i < fad eae Tice Gey. yx SS aed evn ton sale) fa and. there extst tc cee a Fe Ou), = For atleast One ebyecte # - wri) be shicty greatet Batten god wore i x Shichy dominance | ys Re xD ha CH. Qerdiedks | a A oer: te = fy xp) < fi (a2) 2+ Ferete optrmak froet or pareto frear Pareto onHmal Set/ éoludton i \ 7 8 oe IOEACAY) a> a id Sed Z eh bey ~ 2 : 7‘ i * 7 Ren earn) a men $2, 8,2 1 Sande? Adomrnate each pp,er theytare nor dominated Solufions Hy ClammatesS 2 FO) < Fe Xa) ? th noF seuts real thy Frere KK % R% VE pon- Fes) < Fee) t Sara ee | “There €xtst rth cbjectic Fev wg, R35 editions Aces het hold pus. - 1S TTS lagga eee © | Parcts ORimaltty * DomFrnance Relatrow i =o UP XK) < Xo = £000 © Fa Cx) vee Folominases better than “ Lr(%e) < Fol) = ule Fgh < Fate.) - wv | = KK PEGS Fats) ~ * | Parct: Optimal ser [pa= Seni 7 Ra x" # EF Feasthle seanh | Space? f pop: alt conctponding Sofuttown Ceonesponding To at) = | | | | ] | | Frople Region aa Pend peveto opitmal POC Bohs $153 ae | - a ie dlomtnading 2 oe TL domrbahn : ey T Bo | 1 dommotny 4 ; | 4 1B not clominakad fy ). Pe eh ce observostons : Wl: set oP non-c dom nestect Solution VA SF 8nay Relodron CdomInaten ) r eo C2,<) Reflextre or Not 2 not rfleone ~ | Vora : “= L,<%, Trteorfotse => Fore SYPIMEHAC @HeERS XS Xo Aad xy “CerajveER F Apttsginmaernc 5 Transtitve: . & S75 eo eG ete een) + -aee FOV AU Hee Obyechve bal Li es ir +—= = utoptan vector Non erisknceal CR 7-6 pcan emer eco) a . AAVADIAgVe : i jae Coeh Jo @= tol® Wa i t ; oo OS ese ae ig i Afferent fange, Aecteroy willbe lotang CP. bkena th order to tee Haun an same ran ure real hnormaltsoHen for tod wre necal Necltr porst, polesl potnt, went porrt - j

You might also like