Prof. Dr. Syod lkrrm Tirnizi
Transcription
Prof. Dr. Syod lkrrm Tirnizi
MESHLESS METHOD OF LINES FOR NUMERICAL SOLUTIONS OF NONLINEAR TIME DEPENDENT PARTIAL DIFFERENTIAL EQUATiONS By Nrginr Bibi Supenbd by Irr. Sinjul Hrq Prof. Dr. Syod lkrrm A dism.ion submiEn iD p.nid ill,lllndt of lh. Tirnizi Equimdt of th. D.sre. of Dclot of PhilMphy (PhD) in EnsiMring S.i.n@s M'Y 201I Chulam Ishaq Khan InstiNr€ ofEngin€cring Sciences and Technolos/, Topi. Swabi, Pakist4 /& (+ + 2 * +, $\l t-. w (v/ Inih. iLm€ otAllaft,th. Mo6t ..r uL $€ Mort &n.fic.||t Dedicated to my Parents A, My Sisterc dt*<rrf rL.6j,ii:rre!.JrruDLllr+ GhuLm bh.q Khrn In ltut of Engh..rlng ScLnccc and T.chnology Certiflcate of A roval -Meshle€s lilathod of Lings for l{rlnedcal Solutions of ilonlingar Time Dependent partial Difrerential Eouations', phD ft6-b b, N.si!. 8ibi q is of a very sood qur,ry. she l:TllT3 dete*'€d ry]t.th 'l$(esrurry 'l'j :q,r h.r PhD TlEs's on " M{y.201t, h rlr Feully ot Enqlr.rin!_ \,.4es.urK Inst,rute u,e srotrgly Ecomend d. awrd of phD Dese lo the ctudid;k. VJq"^-.-t A{.* Prof, D!. hf,@d **- ttdd r@ \\' ,.'^ \-za) ,'n)A L r/,t1r4i6,, PreLE.. sy.nt anrnntzt ul )/l Dsdco.suEtuisr Appov.d by tcRflor tAcadctucs) ;---t,utlt- Ac*ioeledgema6 thBis ca@t be the sinslchrnd.d etron of d individ@I. Th. sinc@ h.lp m of i'i@n* dnd c@!en1io! of all thoe in clos @nt!41 wifi the in ils valw. I wish to exrend Dy h€art-fclr Srdirude to .ll tho* who helped 1o wne a l!ffhq n ud foenost, in all hmilily I thrl ny Alhn for massinS sill-pov€r dd @Mg. in m. lo tulfiU this daulirg t8t in rhe fae of ndy ch3lldg6. I ow. ad illinitum ro by Grrcios lad fd mting m. @! this fd in pww@ of my Fi!$ My sup€Fisoi, Dr. Si6j-ulHaq. who Srotly hclFd me b th..nli@ poccss oftbdis complelion ed rude me b€liwe in my pot nlial wilh his positive ed en@daging My gntil|lle is rc l.s lo ny @supdvi$r, Pofes. Di Sycd lt@ Tidizi. hs b..n . @.sbrn bsror, guide and suppon to me rhrcughour ttis disnario.. I indcd, ind.bl€d to him fo. loding ! pdicd .d ro my dicli€s ed problem ad thc fNiltul disusioc on $c ropic always cad my last. u, danls ae due lo Prof€ssor Johr Bulch€r, Unire6ily of Arcklmd (New Z.{lad) for exlcndiDs inlaluble euidse in the slFF ofFqsidins dieu$ions. H. is a peMn of grel tnowledge and inr.ller. I m also gdlrful ro Pmf€ssr Srd Son, Mdshal UnivcBity for previdilg @ with .xtr€mely uscftn nar.rid Elded o ny Effih wort. I sould d$ like to thanl all my t lchc.s to! lh€ intidl lhey e!v. Specidl lo ne dulng rh€ cow eql, I m .lso $rni(ful to Mr. Jehegn Ba8h!' (R@ror GIKI) ed Prol Dr. Fdl Ahnad Khalid (SI PGR4lor, academics, CIKI) for th€i. ad$iristniiv. md noral suppon .luing Dy sludis .1 GlKl. No hour of rhEtftlns @uldjusdry d. oL played by my loving paanls (S. M. Has ud R Jd) End my swer sis&u (R ai and B!bli). Th.i. p6ys ad coopcEtion r€mdin€d a coNlet snid€ ed a b.eon oflidt o. lhis ardmu jom€y Md rcplenished ny spirils, Without their luppon ud modle-boosting, I would havc n.vr b.o succc$ful. 'nr€ glid in rheh eyes spaks of th.b mlold joy on lhh Mey thlrks io my dd oes cul+S.tu, Madiha Malili" Ohoai. saee4 shas!ff! Tahir, Robi@ Sulta. dd Inm Hasd for.ielys supponing ed en@unginc nc in I d thdltul ro Deparrrn nr ot Highr Educalion, Govem.nl of Khyb€r dd HEc Paldslan for limcial dsislancc PathiunklNa for grer ol study leare duing ny PhD prcgrmnq Th. codenb oI this dkstution arc my oiginal *olk lhis acknosledsemem is given, pln rc Nny orh.r dis .raF *heE sFcific uion h6 not bed subniled in wholc or in Uriwuirt. Cdrai. aters of rhis dissctulion nave ben pubhhcn Plblhh.d /Subtr r&d Joumal P.p.6l IrI Si6jDl Munmd U3nb, M6hl!* nahod ol x'h ia oJ seroti&z Ktu siwh,l*) .q@iq , Ha, N.3lq Btbi, tiB lo. wiat S. L A, Timizi and ^gdif,l MrrhdMis e Compldio. vol. 2t1 \201O).9p.24t*ulJ. l2l N{inr Brbr, s. L A. Timi, sd sndrl 'l.tq, Mdh'd aI tk6 ualbhed ||ik Ronizt 36l; Fu.ti4 (RBF,lt Nwti.ol Sahtim of KMhM,pe eq@iM Marh.Miq Vol. 2 (201 ), pp. ffii13. Applied I l3l N.th Blbl, Sinjul 8,4 and S,l,A Timi4 "Nwriat Soh*r oJ &!tutlar b! Mesh|s Melul ol IiNt Applied M{hcmdicll Modelling lal Nreiu Bibi Sinjul gq ( Eutl Su wiah bn ittcd). Timit, ''tktlod al LiM Jd tt* sohtim tudittd .tt,.t ,itxh |MEE) .q@r!, ai'la tdiot tail jtuiN , luiel Ensinedns M.0rtrics (Submidd) I51 ord S.l.A. NqfM a|bf, S.LA. Timizi ud SirErl Hq M6^1es nahad of liret lor salrinE nunk u s.ttodinea.q@kn , )oudl of conpnhod Mdne@tus (subhirEd). Table of Contents NetuxrnsF6l.oiMJ*Eiruuwo iudEtua lsorw|on of Equ[wm iioDtrEo Ee![ wm IMMI €qu nor urrM N|. tuid t a.2 htidiqoItro:o oaNa$ w' 5.. t hn.rtb olkt| t.a.t Shtt. 'rtibd '01hry nG noru,iEra s.NRoor,ic$ EauAr Ij'st o! Figuns tis. 2.2: stwri Fitw 2..: tbdr 2.s: rh. w Ponas ot @.$tusrtutY 4di.r Foltt btnq w. Fopaqtunot e.Eoti,.d *irg.M @ lsoB k'.rd r sMtukv 48bttu tit6 'd!tbn) chod. totahn ol th. $4!onor *ttttttt c.lrbr htt'd.otditu E4 [email protected] tu^tu.Mld FrM. z 7: riu. s: EtFowtr5 2 ^.{tN rqwe 1 4tu rtu tute.r s&d ot,h. 4q b1)@l. 6'a N. t21ll'2d1, tu borEt. e ds.lvott t lcEc D) 2. t,Fr €b4tu tje,.2.i2: [email protected] '.oH ol ".p w t'@ ted 4 rn by ' 3h4 @0 a 4\ . lot t4^,h r ae d nar torA a't h. ol tu a!El,ludtfN^tvotEsol'tuFpma" ld i.t alw.pna.4tM4Ftatn*? ncp N =0.@r , tot l dtfqeftrdE olsr'p. p.th44 k' FElt. 2.13: €s.hat2s s.'kr q dN 4.P A =OaDl, t,cillcQrtac q Ntu E tbut. 2.i4: aq.twttet td.d l22tiott 3tu dd' M'icot cwi.Ntld ' r Fisu.z.ta Et$nd!4 xokd b/ th. tuF tu@ 2.tt: 1tdid ttus rcp@m dt t 12 27) n.p Lt =O,gJl , lN dttd.rtrct@ ol nd.23-i stipt Ntffiet , r' -00or,ldd!l.td$tu8Ntn En@k 2 4 Fi\!.. 3.i: hN w^n Nm.ta .Jot *6eh 3 1 'hoq Fi4w t t: ft ffi\@ dtu ol tuh{o.@tbakokt6 stwiat nntut t./rtw ord Fiqde 1.ts: Eisaw@ s.ot dbr.t6. tkP N btupk t z 'hq Nn4i td ttpa 1.: ertuy @ tdd6.l hodittd x8.h6o $d,oa P N@eoo eh d iltut:Fnkofid tn6 rho* rnatut 30tu tor dd .\haMc 46 totltbr) adn r33 Fbd.3.s: etv 68 'h4 [email protected] tt@ 3.6: sotr;qffi tuol r4r tuhn 4tu lt ttcttu'ho, ^Med \tu6 d te'rc t.3. 4N wE6 Fi4u. 3.r: tntldb^ . ol tn tuBld IilrotFquEs 4cr. t 3 tnb'daa4 rkeeldd'J6 L.nFt. t t hrlt. t e k@dM ot NvR lot aonpL t 6 hout.3.ra sF4nrnlot FDlt. t tt 5.d2d E onph t dqn6t!6 t lot .tonpt. t 1. rw. t t2. spdM lo' Eoapt.1 2. hc&tB tudqavMld z ^@pt.t hsw.t.t Fe.rvot*r lat hdnot.3.3. helt.3.1st spedtrn lt 4c6pk 3 3. Ftelt 1.1:.nuu*6shot Pama.r lot Pnhkn talt 1.2: wsv @. p.fB vith Mdnld. ot, Fhu.. r: sEtu tututu.hR dt ol ol t k en tubta.1 ol3^q Nmd4, tw totnoryeM Ftola 1.5: hdutdtut rtolr5 or dtlld.nh. Ftelt ..7: oodubtd pofr$ d dtn44. h. kwtt ltt F5 br.tt ld d4- .@|.'':shaqWfu.eld|g.m' Fi.u4 s 2t ti@k 3dto4 w wi6 d64.oi onplltud.s hrua 5 t 11. 4 tod.t 'ffitu FttR 4..: sqtuo ot todbtu mnlu dtr.t't!. heet1.5 t*.a{lor 1 'tt tu @rd trE t46.at nun"nrot spdrdtrckol@nE ee(? wd\ hithsk,ottutt$u.o.e2qd Malb)6 k) tMo fttd2 tllil; stobt2.i@tot6l&5ng[ tedt w ac2q l4 Ma@ e k)tMa Fbu. 51n:eobitity Ftcn st: ttt todhn4tN aru| s ' 6 hn.odbr ol th4. h4w'ehdtlld,rtoarrtud. tuory w!6dth dlflRnt oqpth'h' t9u. 6r : hd. dt 6 r.ro.rQ D w rgu.6 ) ,ehdtdlr art.lott.q htw.6. httddw 4tp etit6 Fi't|5s'o^d|njra]n@A'l ,isd.63: Buad nd2 ol dae, FiN.6tsod,ot ol sotutbq 4.32 q - te li:tt of Tabbs r.a. r t: crybqafwtu rdr dh le2l ld 4k ?.1 r!a' 2.2: wtu* aI. qd Nlq.'anpb cd.atbn tur,brt al ^eittrdtn*tu Nnod' ai lezl lq cwruq aI @r Iou. 2 a cry'6d4wr nn^o^wtth ftbt zt lesl dd* tu'.rrtpk 7.1 2.2 2 1 r5 s4not.a. ol.haz.$. d t-5 rru.26 rie tu. al.N\|ts.t. fatL t t R.,ntIu Ktutun.qdb^ a.ryiw nh l l t2l touc t.2 spdpl d. [email protected]. ald E attk t I t . ,t rda. t.t: R.rtE l@ M,4i.d tu tsn.Edia tou. t. Wd @4w4ed!e Eqt r2t=s rrtn ).5: tutu xav Kontrtu @idrr t-s fabt ol.q|.ryE. dt lu Ex@k r,r t - 5 rabb J t hn4qrJq turtud al@ tut'ttu ld Etqapk t.1 fabk t.3: tihtuA ld hbtut'd al thn. tu M lu Ea,qta r.t fqut t.o tn\aiq, h, hE4!i@ ottb4,olnn tu, tunpt. t 6 robL t.t. k\on@t@n{ ,ffilt't tntk'ottt.lde fabt 3.6: s4ti.t td. rctk..2: kw,iN: u,t no, M Ju lhtt. ,.|ibo w|. tod. Lr) kwdds fq id.retb oltw vow 'o1ib4 rabL4.1: D.nto@d olu bd'tat t*. tou. 1.5: 7abL 1.6; eFlwn 4u a. t6a@ r.d.1.7: M.tqwat to 4d atAtot bdr lE5) af q sl4t tbo|.ld-2) M to t4bt'5.r. kvnqLai.40 tu to.,,,tk hl,k , d d,f.4N vl4, [email protected] rubt, t,t, kw,iub ad 4N, fttl. t.r sp,,t td. ot.u||,su.Ia r.A. t.1) rou. t x tndd'Jvtu sal@r4\.itueM tMdBtq roLt t7: P61d otd ie6'i8 t-tu 'tu. 'drq|M' arttud..hhre w d hbt.t .(ryaqoltqt ,ohdo robl.6r: Rd. al.d|.a.E. ir rlat 61 atcq|.z.rc. h tiw ftbtt 6 t 'N. Ju 3hd, tq 'iryt, cdvnd q@nb, h, bo,otia$ Rde iltitu ot '=l ldrqk,N=to. roht 6.7: sinhQJ rar. 6.3: audi,t 'atd 8iah.t tuta^t'otxq ra* 60 3,^ oItubtk !.r,@ rou.6.10; 8od ad. al a= 'otdtu t.]t Abstract Much ot uselul wo.l beln! don€ todsy aor num€d61 $lurion of garlial difieEhlirl equations inwlws nonlanear ec u.laons .risins in dilteEnr usd num. ol ii.lds 6l si€n@ and .nlane.dn, r'6t widerv t.diniquer at€ 6nit€ diffden..3, fintte element', sp.dral d€rho& 3nd @llftation m.tho&. Howd€. these n€lhod! fa.e bme limitnions like @n*rudion ol re8ular gad for irf.gu ,. anorher cla$ ol and con pler methods Beom*net knNn :, s ow conve Gen e Ere sla bilitY .nd low accu rdv m6h free methods k expe.ted to b€ supeior lhan convenrion.l mesh ba*d methods in prdidlng mor. aGuEt. and ltable num.rkal solulEn rnY.onn..r!. mesh. M.lhles d€lhods using hdi:l basis fundioni a.. mo'€ flerible with hish 6nw.!.nc. 6re. Tn€e nerhod! pBid. v.ry.cd.are nune ol elurio^ with row withour Jhe .4.aich pRlented in thh dlssen.lb. ls brs.d on m6hl4 m.thod ol linee using Edialbasn fundons for numencalblulloB of nonlinear tlme dependent pafti:ldifieremial equations {PoErl namelx Generalten (lEfrolo Slvash nskv (6rG) equttlon KaMhad tvpe equarbh,, €qual Wldth (Ew)€quation, Moditied Equalwidth {MEw)equation and No'linea' giwn PDE in ro s.hrcdined (N|J)€quatio Fidthesparialde Erlv6 are dkcreiize ro'wniru tiur ord.r odinary difid€ntial .qutrions (ooEs), sharh is lh.i slwd bv 'lt$i"l pro9'ni's ol rhe Jourd, oder Runs.-lGtra {Rx-4)sheme Eishv.l!€ stabilitv and.onre4.nce €'ior method are ako di*u$ed. Ac.uracv ot the method i! mea5ued in t€.fis ol ir and fa svsrem ol norms and conrerualve oropefti€s ot mast mofientum and enetgv The pEsent method is smples ava'l.bl€ dh ertsri4 num..al re.hh'ques inplen.nt.tion a^d il lh' lteGru e rh. n€$od shs rup'no roEcY €fiiciencY of m.5hle$ n.thod €d:e or Chapter 7 IntToductlon 1.1 Lit€rstur€ SurveY of numc.ical rerhorb for th. solution of nonlircar Paftial diffecntial tlo tm fmn bolh equadons hd ctjoy.d u intere period of &rivilv ovd the lsl $@crical dd pncd.al poinr of vi.v lnPDvemnls itr nuDcrical t'chniqucs, rogeoer wi$ rhe mpid advanss in compul.r lcchnologv hav. neant oul mnv of Panial '"The nudy difloE.tial equatio.s .tising Aon .igin4rinS and $ientific applications' which *ee pEviously inkacublc, c& no* b€ rcldrcly solvcd Ge Milch.ll and Gifinns [1]) \onlned panial d'ff.Rndal equat'oru dc{nbrng panicul& mhaty vavcs die in vuio* plsm physics, guanrw Physics, dd most ess, rhe cx6ct slution of th* $lution ol liclds such nonlre d $avc Plrmmna dnd in mnlid optrcs hydrodvnmics (s Dcbnalt [22]) ln so one ncds ro find thc oplcat fibd @municanoi cqu.ions h .ol availabl., such eouations nunencaUv Numerical meuods sEh N nnne diffeEnce method (FDM) fi.ite elercm mrhod GEM). finitc volure rethod (FvM) and bound{v €le@f,t ft$od (BEM) havc ber usd lo solvc a widc nnge of panid dilfcFrtial equ*ions Thce rthods di*Eriz Oe doMn ' or m.$, gnd or a {r ol G€ne6lly, $e fornaion of a PornE tqurnng t liicd conn4rion dnong fien squm ot t.clan8lld gnd is equiEd in FDM Nhich is lery dd FvM ac fleiible in d€aling vith @npLx gcomuy bur fmtion of ! *cll suiled ftsh in 3-D sltucluE of $e dam involvod ud coDpuar progming is difficull (D.ntowicz .l .1. l2l tid Rachowicz a chalLnging vhen problem dondn is nor regulu FEM o/. t3l). ftee @ sitlatioN wh@ E cEucdo o enlargcneDl ol m'sh ov'r tine is ned€d, like pbblems of cnck Ptupat.tion ot fngrentadon Also ii boundav fEc problems, arising in fluid mhanics dnd finMce mod€ling, lne donaiD b€i'8 Pan or a.P!4 b. DEdicled in.dvo@. In continuous mhuics one h410 ddl wilh fic snudon whcE ftsh may beat down bc.aue of d.fom ion. ln all th* dss it is soludon c@or v.ry rougn lo deal wirh a strtltuEd mrh. Apan fron difficulry of .noti{ gensaion, prcblen is achieve@nt of accutrcy, which is linked with oder of approxinaton schemes used. apprcxi@rions In order to aroid polyiomial snating prcblems only u. acb$ dE msh bul nol ils used to appoxihatc d. funclion oDly low ord{ e@c Fd tuE .pForimtion of derivativcs, hi€b order *hcms neesery which involve addition l computalion l @st. Whil. using low ordcr panial dqilatircs. e mh she@s ecurscy c.n be inpovcd by lclining lhe @sh but again at lhe expense of incr€ased conpuutional cosr I4l. Merhod of lines (MOL) is eothcr well es6btsn d .u@ticd r*bniquc so[ine pdial diffcrcnt.l equatons IPDE5). The us.d by mth.nadcians for solvins bourdary merho<l for sas orieinally developed and vdE pmbL6 in physics (z,f@lhn l5l, caer dd Hinds 16l aid schiess [7]) yhich w6 furrher d.velop€d by PEsla.r al. lE, 9, l0l fot pmblems dising in elatrcmgnetism. Tne rethod oi lincs is a special form ol finit difreEnce m.thod which is noE efficienl thm rgular FDM with Especl to aeurcy dd @nput timal c6t. Tbis is a emi{r*cie minod wbiclr is convenic.t aid quit @liable. In 0!is ncfiod by dicrtjzinS th. spadal d€dvativ.s only dd ldving tim vciable co.tinuous, th€ original PDE is convcrted inro a system of odin,ry diflerntial eqlatioos (ODES), which is dEn properties of and i egrat d io tim. The Brability and conlergcnce tn @dod used in MoL .4 4sily b€ e$rblishcd tire disEriation. Preglzlmire .ffon ce b€ conpe.d Tbe dilTicultics dd emci.ncy of diflftnt aily by slving ODES I I I ]. b.i.8 spfc is ato Edu€€d by nakjng use of st n&rd ODE soltc6. Thc accuncy derivatiyes by sepdating rhe rpprcxinatjons tor spadal faced in Gsh-b.6ed nelhods noivated $e res@he^ to d altemliv. ro tmditional grid-based nu@ical D.lhods. So a Dew ficld of n4hlN netbodr dergd ard thc fid n€shl€$ mcthod 'sn@rh Panicle lmk fo. Hydrodynmicj v.s pBncd ir fo. lh. simuladon ll2j dd Luct ll3j of atrophysics poblcms. The mcahlcas mcthorl! for n!rcrical 1977 by Gingold ed MonaShan .slutioo ol PDEe hale becotu very oppealing and .chieved a enarkable prcgles ovc! the lat No decadcs, TIE min s@l of thcs. mthods vas !o elimitut€ lne ditriculties rhlt conpLr ge.neuies wheE lomtion of con@cled nesh.s or c1'@nt6 $ a malor Drebl.n vhil. usiig radltimal dcsh bas.d m.lhods lit finile diff'Fnce mthod ely on (FvM)' vdious GDM), nnire.le@ rethod (FEM) .nd linile tolum tulhod nBhlss tuthods n ve been develoFd *hich cd bc dssmed into dE greups; (l) d6h f@ Paniclc mthods in qhich a finii. nunb.r oI discele panicld tr u$d ro describe tlE slat of a systen and to @o.d tn. novcNnl oi tne svdemr qdk fom of PDES; (3) nethods basd on collMlion technique dd *ork o! sbng iom of PDES Tnc ke, tqtuB of mcah fE. ft$ods is ro poaidc a $abL nu@dcal elutio fot PDE9 or irl.gral equadotu *ilh all titrd of p6sible boudary cond,rions usinS a sct of sndEd nodcs or eanictes wiino dY mh. A *idc cl6s of neshless @$ods in prcgrcss bdav is bs*d on colldation (2) we.k oeshlcss @thods rhai work on neshless method using adial bdsis functions, methods using ladial basis functio.s solurion oi panial diff.Ential al$ knoM 6 (RBF hav. bsn me$od Mcsh tEe equations involving multidiDensio.al complex doroins alPonm.do involrcs pan poinrs so RaFs rcrhodr suffer no @nplicarion whil€ in $e el4ion b'l{en lhe drta dd neiibilitv wi* dislanc' wqting {ith highfi dimDsional @ RBFS appdiimtion e@s bacr to thc v6t of R L fttdv inrclpolatron of $au.rcd d.t involved in geogEphical surftces. ud pDbl.hs. Thc Il4l for $e s usal suc@$fullv fot nunerical RBFS metho<b haw advdtag€ of exponential conv.ry€rce of nodal l@ation. As RBFs Ksa liLElft RBns for sraliralio.al and ms.eric donalies. I. earlv 90t E.l K@a [15. 16] used solving elliptic, p@bolic dd hypetuolic PDES Golb€r8 ed Chen i30l cxPtinenDllv prcvcd thar RBF l.chrqne in strong fom is vcry useful for elvinS Fnhl diff'ential d equtids. Ealy sort ot Xris I|6l od Zemu*ar 'r'zl Po*"and Chet llTl shopcd 6!r $e Fd()!m. of RBF bled rethodr is mrch b€llct tlts FDM Ll6son and Fohberg tlSl md. t conporisn b.tw@n RBFS nc odsr FDM $d well s intcStal Pseudosoeclrll mthods md lhev lound thal high odr 4mcv is mhicvcd by RBFS ne$od. A dircd quantiBrive conpeison b€t*een RBF derhods aid nnne dement retnorh was oafied ou by Jichun.,4l ll9l, Poving suFrior acu'acv of RBF ne|hods. ln F6.ke's Eview papq l20l duldquadric Bdial bdis tunction 'Ppox'mton of Mo for inl.Aolation of rclhods nub€r b6t dong a ldgc re$od w.s dt d s th€ din€nsional scattered dlila baed on then stability, &.mcv, efticiencv, ease of inplcmnkrion a.d mDory rcquite'n nt, MiccheUi {2ll Prcvid€d a oathedaicd fond.rion to tua@te nonsinguldnt of thc RBF coll@rion oatix. Th. eri$eme. uniq@ne$ and coove.g.re of RBFS wts Ptolcd bv Micch€lli l2ll, Po*ell t231. Madych dd Nllsn t241, and schaback 1251, for mullivdide sereF! dlt inrctPolalion. l27l lnd Wentnand [28] .slablished rhce f6crs ror $e $lurior of PDEI. Two importa asPccts of RBFS nerhod had ben obstved in Lhes while w! l) t261, Fnnlc and scnaback ruty rcshlcss altodthn in qhich coleation Poims can b€ elcct d fEly withdt ey cdmctrvity, h.mc @Dplid.d ftsh fomatioD is Th. m€thod is 2) ft a is indcpcnde of O(lr'), wheE dincnsion. spaial I! in the sme lnar th. conrcrgcrce is of sFtill ditunion n is densiry of coll@orion oiher words ord.r of convergence dircdon *ith low nunbq of the pobl.h and hcrce or @ll@adon Poi !. tu points and I is spatial incrces *ith incMs in accutocv crn D. pr.eryed In panicul{ Golb€rg and Chen 1291. direNiond Pobsn €quario! with d v 60 tudomlt dislribded nod6 ro gd $e sm 4cufuy !s sith FEM redDd usinS 01 000) lined solved $c Resedchcs like Fombclg Dd Dn$ou ard Tanalc 1341, sd t3 U, Hon and wn [32l, Chen and Hon (33] chei ch€n d, al. I35l aho conuibut€d a lot KlNt n an usinB RBFS to solve PDE9 H.mile lTe collddtion mcthod Laler on Hon !r 4i. funhct wid€n the RBFS n.rnod fd soldng difr.Enl tyP.s of ddinary od panial difi.Endtl cqutions imludine Scncal inidal value Problcns I3?l ionlind Bugdt .qBlion wiih shNk ww IIEL boud..v lie pDbl.hs likc thc Amencd F6shaler [36] nodined lhe thod to a id!8uld boudsnes sucn as shallow warer equation for tidc and cuftnt sinuhtion l4ll. A clss of Kort€w€dg.'dc_Vties equalion option pncint 139,401 and prcblens with ionliner s.hrodinse. cquiion [45, 461, couple Sinc cotdon equaton Md l4?l ed rcSuldiz€d long *!ve iqudion. t*o dinensional CouPled Burgs cqualion MGrlincnsionrl R6rion-Diftusion Boss.lalo. Svslen (se L48l) hs be'n studed using cotldltion apprMh bas<t o ndigl bsis auncrios Ovq atd Wnghl l49l usd t42, 43 a.d 441, tire &BF @rhod for $€ fmr and to $lve shlllw w{Lr €qdiiotu Pict l50l sohrd mv@tivc in [5 sd esy i. Fomberg ], 52]. of lifts witi ddi.l bah luNrio6 CouplinS rethod ncrhod nore acc@r sn ed uing RBFS appmximioB. Funher PDE on splPE d€hih csddins RBF rcthodolosy cd b. on sPlEE. (s [?8, ?91) ner.s tnc impl.n€nration as no nesh is equiGd in thc problcm domajo, Hi81l quatuy ODE elveB ca. b. u&d to obtain solution fot the systcm of ODES obaircd d a rcsulr of RBF appbximtion of spalial d€nvrdves. Thc liEratuE for nshlds tuthod l2 of lires l4hdquc i. lcry limit d for Method of LtH depen.lent nonlit.d PDE5. usiDg RBFS Befor de$ribinS fic rEtlrod l.2J tim s 16l prEs.nt thc folowing d.fioirions; Definlllotr (RDrs) A Eql lalucd dishnc. d:!8'rS,(d=1,2,3), runction ES' b.tNai ond sooc nrcd point whose value depends only oo he ,j €t8', ./ =1,...,x (knorn as ccDrre), so rha,rG,,,) =r(1,-,,1)=r(,). *hc,! . is Scnehly tI€ EElidm disoncc- fi. r vnal blc in then application, ehich is dcfincd positivc dcfinitcms of RBFS pl.ys 1.2.2 D.ndlion A continuous function /:ry J$,h srictly pGitive deflnile of order u if for ev.ry ser ofdistnrl poi s r,,r:,....r"€ $i,i.rrlll, ror -,.1 a|l,i,,4,...,rx 9'. l,o satisfying 24Pt''t=o for all polynooi.ls p€ also c.lled d r: . A @idnionauy pGitive definite function ot order r=0 posilive detuiL furlioD. ln ord.r to apply msl cs mefiod of liEs iahniqe E mrde! PDE of rh. fom h $*r1,t=0. *o, 'a o.l) *1r". ,=,(r,r) sd L is spttirl d.riv.tive opedor' kr us slPPos' lh't ir.r,,.,r,€ oc9r'.(d=1.2,3)be t giv.. st ideas proposed in $e solutjon r(r,r) cai .xP6sd "(")=:,..i,(,v{ll,,,lj. wh.E q is $ne adial bst untDown ceflicients lo domjn FolloNins nethod I8l foi 6lime dependent PDE Kee's be of ce.ten in prcbleh te apptoximale as ( tumrion, t/'r t.2) @ anr6 dd t,(r),(i=r, ../v) b. del.mired. similuly lle appoimte soluion fot ft sPatial dcrivltivc opcnlor cm b€ Mitren ar tui,)=$ r,r,)lL,lll' . Thc abov€ apploxi@t ons (1.2) dd ( l3) ,{r=". r =[,, t).,,(,).....,, al. = .(u ) can b. wtid€! in mtdx fom a (,)]' I1.4) (1.5) whcrc A hs cntries of $e fm /, =,/(l', -,,11). i,i ='...,,{ and , is (t.l) ll)1. lnrisynnerric matrix wnh cnnics of ,,, =r.y(ll'-',1l).,, i.i= the fom t,....N. usint (1.4) dd (l-5) wc set L(u)=(t')o. Equ.tion (!.6) m be ( mido 6 l,(u)=Du, After die€lization in smi dncrefted 1.6) (1.?) spd vith sditl sysrcm eiven by baih functions, PDE (l l) is rosiomed in to a 0.8) The sysrem of ODES (l.S) ce lhen b€ solved wi$ mv have ued classical folnh o!d.r Rung€-Kutta mcthod IlE sumcient condniotu for t/(r) to ODE solv€t gladte m!{ingolditv l' lhis thesis we A wed or rhc marix Schc'btte ri6c by [email protected] t53l btcr on Miehelli tzl I 'rttntlcd idea so lhat a ldSe class of functions could b. @nsidcFd' w€ fisl siatc som elevmt dennitiois b.foe Soine to schoenb'B 8d Micch'ui given for Oe fid ftfiridm r3 A fumlon (i) (ComPleEly Monobnc Furcdon) r. k sid lo bc @mplerely mnolorc '/€c10,6), (,ii) ( l)'sL"(')>0,ror.>0ed D.fiDiriu r.4 (Positiv. symeuic dd r'A r r cigenvahcs of Tbeorem con$ant ,4 wid' oi 0 for every Posilivc rschcnbe'g l5jlr [0,@) , lncn ror ouis a,t (.) dy st =l(lr,-rrll) lf @r''-d.t' a comPkklv , or distirct poinb foloving r...., '}. RBFs e $c nxn {r/ i = 1. . mono@ne bu' nol r} lne ix' matix n pcirivc definn (od theeroe no4ingula) tt (t)= ((.1;). c"lo *),t'(r)> 0 for '/ is conplctely morctore bur not cons|Ant eoDs {r,. i = The @uix ,4 is Fshiv. dqfinite il n is a ,iimtuiond 6lumn varor rto. or if aI tlE Definite Mattix) A tn tutrix ,{ N l.l I =0,1,2,.. Theor€m 1.2 (Miccherri I2U) v/ if: mtir ^ oi (0 @) wi'i cir.i6 zr &eqEndy uscd,n li|.ra{w: thcn for =d dv {li, - 4l) r>o.and scr or n is distincl noFsinsure Tru.lJ ryFs /(r), f>0 Idbli.ly doo& RBf. M"ltiquldric (MQ) tnv.* mlltiquldric (lMQ) PL..rlrc .Do$ tAFt Trin pl.E splinc.s (PS) ft d 6ily bc pm!.d that inv.Gc Ellli{urddc oMQ). inE$c quldic (lQ) .d GaNiln (cA) ddid t6i! tfttiG snitfy tlE suffci.nt conditi@ of thc Thc.Em (l.l), whilc multiquadric (MQ) sltitfy thc aufiici4! coiditi@ oltlF nE$!n (1.2). dd lE@ for ilFe typct of RBFS cqultid (1.3) is uiqucly slvrble for 4y 3d of dktiltt TIF slullcy of RBFS n thod! d.pcn& or thc choii. of oplMl val@ of thc rlEF p€r.iDct.l inlolvcd in idiritcly 3oooih RBF3 lik!, Multiquldtic (MQ) Oau$is (GA) and li6c oolti$ldric (IMQ ndid boi! fuictioN By oPlind vduc. q. 6@ rhc rloc th.t poducca th. Mt @usic rllult! ln Flctice, th! vduc of ih. shlpe pmeter deiircd . nust be adjusted with the nunb.t of enrc^ so rhat he inr.rpoladon n.rrix A is wcll condniomd .rcugh to b. inv€ned in linic pr.cision silhnetic. Ii has do.sny of poinrs as ben shown d th th. accu@y of nuldquadric int .polalion is rclated to well as th. valuc of rhe !hap. Pu|neter, The@foc, aEUrcy oi the nunerical apprcxi@tion c.n be inct ascd by cithe. incrcNing nunber of couocation points (i.e. by d4reasiry mcrh siz.) or by incrcasin8 tbe value of the shaF Td*arr [54] fouid thal lhc rcol m.an squae of enor dccF.es up lo corlain Oe' inceass npidly when opplied $e p@oct r. techniquc FontE€ stably conputing €r fic .+*. of o6s linil a.d [56] atd Hickerell and Hon 156l validadon lo obllin an oPtidl !.1u. of th€ sbap€ Golb.q er ai. I5?l 4lablish.d a 41. Contour-P.d( aPpoach which is caPable of RBF lppoximation for develop€d RBF QR algori$m ro lll c>O olcrcom ill-conditioninB of ln order ro inpove condirioniry ol RBFs colledi@ > pmter. Fomberg ahd PiE 1501. RBF8 interpolation on n.triG, th. de folloving altemalvcs Conpedy suppottcd ridial basis functions (csRBF have b.cn inmduc.d by Buhmm [60j) in q{tq to obbin a sPaE ini.rFbtion Datrix dd th. F6sibilitt ol6Pid cvaluarion of fuetions. Unlike audlo6 0l.dl&d goballt supponcd fixed dimal d I5E, 591, .d RBFS, CsRaFs 1581, ducared polynomial, nE G rc srticdy p.sitiv. dcfinne oD 9' onlt fd a bed on a aftitrlry dcSne of snelhMs CSRBF co.strucEd by Wendlad 1591, pcnirc dcnnnc having Cr, 16ll. wu t62l offcrcd dorhcr tcchniqu. to consrtuct the sde csRBFs. buI pnliding a high.r d.t@ polyiomial lor a giv.n l.ycl of snoolhfts and > PEconditionins techdqu€ (B€|6oi er al. 1631, Ling and Kansa &l b6ed on appoximte cadinal bash functions (ACBF have effectively > i. which is bai used inErpolation and in PDE stting. overlapping and non.ovcrlappiit don.in deonp6idon methods (DDM) coupl€d wnh meshlc$ RBFS nethod on Datcbing and non-nalching grid have becn dcleloped. TIF ovqhppins DDM less spltial Bohion qith RBFS [65] p.rrormed bener wirh as compaftd wnh FDM .nd FEM. Liig and Kaisa t66l vqified dut combi.inS ACBF pEcondhidi.s .nd domiin &@mpGilion ma*es lst6 s .mpdd ft) di@t i.vdsio Lealiz€d RBF netho& in vhich ndy stull ovctl|pping RBF sysldns m usd to oblain a fle$ble solltio for lal8. sllc prcblcms wnh m is@s ol conditioning. Liu sd his collsgus 16?, 68, 6q 70, 7l iid 721 lllst inl@duced l@l RBFS appmrch in invdsion ollbe > Ndll systcn io elv. fEe vibradon of radial poinl intcPol*ion mclhod 2_D solids and de incompEssibb nuid flow 3inllation, lnd difieMtial qudBluE oollocation nethod for Defidiilon elving $o dincsional inMpesibl. Nrvierstoles 1.5 eq@lioDs rcspect'vely Vetor Norn r is . @l Posnivc numbo Stving n.6uG of lhe tiz' velor and is de.oied by lJl . x musl $rhfy the followins 4bn' (i) lLd>o if r*red l{l=0 ir r-0. Th. nom ol a veclor ci) lql= kll4 rd ! Ed (iii) lt+rl<l{+lrl. o, @nphx civ€n the terorr=(r,,!,,--r,), $h. D.fitrition 1.6 The l-mm ofttF rdlor of the s.dr .. r of 0E rcll kNM vdbr mm ryps re h thc aum ofthe moduli oflhe @mpo*nb rhe of l't, =t',1.t', *.....*,. =Il',1 sm of $e squd€s of dle moduli of the @mporchts of n =lr'+l4l'+ '' The 2-nom Sives + r-'l'/ r Mtof the is muinun of lhe noduli of rhe Dennition t.7The2-oom (orEudidsn nom)ofthe vcclor r is lhe squE i.e, ) h' L; l =L lsgrh' of lhe volor. DefDitior 1.8 InUniiy (a) mm of th. wctoi l4=s,, x kt /(r)€c[a,r] D.Itbition r.9 k,rl . o / G). t, k,rl ti t, =lJ[/c)] t' d, ). In oder work done on usd the following approxiMtion for I, ercr nom of tlte ir> and {- cxetind numncal dunots rcsPectvely, lsis sbbilily of RBF trrthod ro undeEland h6 her nrdiom I I Sbbility An continuous l r l3 dl space of rrrc rcar varucd runcdo' dcnned by have c=lJ"l,,*, (rllc lin@ s for m [73, 14, and ?51. A $ablc a very hlle dm i.qtuion $he@ Equrs thar d.Fndenr PDES o.ly ha rll th. cie.nv.lues wnn m4ositivc Eal pan. A sFctrun having Eal pdn of all eiS.nvalws in $c l.ft half ple is known as. skbL thc spetrum of tte sFdrum 1751. diecri&d A ole of PDE 6unb for sl.biliry of Nthod of lim is tlEt "the eigenvahes of tim slep At, lie io fi€ st biliiy egion of ir sotu ces rh! d.tals of fie stability de norc in $h disenation ve @ dealine wirh nonlincd the spatial disEtized openlor, ealed by thc time{iscerizarion opc6ror, alftouSh rcchnicol ed rcsrriclive I?61". Sinc. PDES, so after spatial dhcFliation we havc a systeE of mnlined solved by cldsicsl Runge-Kuttr rquies that all eigenvalucs scnem i.€ for s d n thod of odd fou (RX4). fte ODES, which is then snhllry of $e rcthod ol thc Jacobian mat ix sadsry stability condirion of RK4 sufficiently stull li@ st ! & , {c have (-2.78 < a& < 0, fd oll a). sm lim RBF nethods cannot bc @Mt sd un*tuinry pdrciple Th. condirion numb.r ot llE @ll@arion @rix n dcfin d by r(,4)=l.l[,a'1. [251. ecll cmdidomd.I dE due to tlE 1.4 Solitons ard Solitsry wevts Solitary waves @ wav. p6ct $ or pller lhat popaSlt. in nonlin@ disFBive nEdia- ir fluid @chsics, pldna physica. solid statc A wide y&iety of phemnena eising lircrics and g.Ghcmist y de descnb.d by elilary pbysics, oflical fibe6, ch.mical wayes. Sohon is a sp€cial typc of solit!ry wav. that dain ns shape whilc it travels al a consht sFed. Solirons ar produced duc ro cancell.tion of nonline& snd dhpersive effecE in thc nediun "Dezin !.d lohnsoi l7?l de$ribed the followins $r€e ProP€nies . Wavc of pemoent ioml . t calied wilhin a &Cion; . Inrenct wit[ olhcr slilons and.ftree exepr for a phe aftcr collhion witnor chmge in shaF, shift". A elilon wilh ldSEr anplitude novcs m@ apidly 6 conpaFd to a soliton wnh ploc.s of prbP.8ldon lwo such wavcs witn diffeEnl anplitude getl clscr. @llidc ed $cn ep@rc into two $litlry wav6 dint ining then odginal shrp.s and rclcitics. smalle! @plitude. Duing lre 1.5 Overvi€w of th€ diss€dxrion o. th. d.rclopncnt and .volution of I6Nss refiod ol lies usine Edial b6is furctions for t lving nonlind qave equ.aons like dE Tbis dissnarion is pivorcd genelalized [email protected] .quation, KawalEa .quation, equd wadlh (Ew) equation md nodified precne nftri.ll .qu.l qidth (MEW) cq!.tion lor obtaining high ecmc] and solltions of these prcblens. ntc dkscnador is dividcd into 5 chapren. In Chaprer 2, the nuhencal sldion of thc generalizd Kuonotesiv6hi.sky equation is pesented by using neshlcss method of lires (MOL). we havc usod globolly supported ininitely RBFS for rhis plaos. Mde in l.ms of global elativc.rcr tc' RBFS nedod of lines. Stobilily of rh. in spae dd titu h also calculaFd The compuison with lhe exhring techniqle is povc lh. limple applicdbility and superio.iry or poposd shene h also discused. Conv€ryerce showing th. .fficiency of the schene. In ChaF.r 3, RBFS-MOL tahnique i5 fomulalcd kwanm eqution. Nutuncll Esults in l.ms of tft rhc nu4ric.l elulion of mr noffi I2,L re conpded virh liEnn!!. TllE c!|gvcd quetitics ot n ss, drty $d nom.nmm ,l! .ls Edli.d. Thc inlrdioo of wo rid tlulc .olibB is ltco discu$.d. h Chapcr 4. s pEet [email protected] io|lfiio of .qurt eidth (Ew) cqu.tior Srbihy of dE rEtlad i. inEti8.t d by cdcrbrnt !E .itctv.locs of dE Jaobi$ rhc lslts mE in the of $c ilsultin8 sy3l!n. lilc telDiqt ndid bcsis fuFtions for dE mMicd iolurioi of rlE Dodificd Gqrd widur (MEW) .{u.!io!. Soulit} daly3is is rlso csLlblhh.d dd protld by gnphicol llPGsttrio of cig.nedu6. TlE mr .oths Thc Chapl.l 5 cont!i63 rit$h I,, Le using cdculrtld rld coN.fltrivc poFnie, ol tn! s, mmnlum nd cncr5, e lle cquttd by msh lcts mthod or liF. Moaiod of singl. solitotr, ircr.irioo of t*o sliioG &d Mrl(elllid initid condinon fll bould 9r|! of elirom e di!.u3!.d Nuincricd Gult! e cooPlr.d with cri![ry l! ChaDcr 6, s elvc d iEe S.hrcdi.gcr Chapter 2 Meshless method of lines for the numerical solution of generatized Kuramoto-Sivashinsl<y equotion 2.1 Introducaion The sen Blired Kumot,J'Sivshinsky (GKS) €quariot is a mnlined cvolution equaton =+,=+/r -+v -+, -=0, vh€E d (2.l) I r<r., >0 /. v ed, @ calcosla.B. This is tlE nonlined .quario thar Eainr lhq .sential clcren|3 of involved in vave evolutio.. Th€ tems instobiliry .nd enrgy production, * sy no.lirc& ,* k lnc doninmr noilined for sr.bility od tem, ener8, disipadoi, * ud # is $e dispesion t€m. This equation was dciived by Kuranolo lo study dissiParjve structurc ol re@tion-diffusion 1801. fth cquaion was d.rived ind.Fndenrly by sivahinsky whilc sodying fi@e front pmpagalion in Djld codbusion process For !=0. Eq. (2.1) is ! 18 U. norlircE.vohdon equation called lhe Sivahisky (KS) €quatior *hich desfiibca a vuiery of physi€al Kuramolo_ Phenomena disi.g in savs oi rhc incrf&e betwn two visus n!id. l82l and unsbbL &iff wav6 in plaM .rd @don ditruiot s,stetu t83)Chaoric h.hrvior h.vitrg F.vcling wavc s rcl ions of p.maftnt shnp. is obscry€d when Eq. i2.l) is int Stlt d orct a finil. donan with p.riodic boundary lons wavcs on $ln filN. long condiriohs 1841. Thrce clsses oi fdleling qavc solutions temed ds Ilgular shocks, osillatory shocks and solitary wales for KS cquarion wet studied nufrericallv hv H@per and Crinshaw t851. The travcling wav. slutions wcr. aho studicd bv Yms v=0. nis u*d |o dcsnbe tbe Patt n fomtion on unsbble flmc ftortl od Lhitr htdrodymic nltr 189. mt Nutuncal l4hriques like tsal Ds@ninuous Galerldtr nethod l9lI Chebvshc! t861. For r=4=l and sp4ral coll@aion rcthod 1921, Inplicn_ExPlicir Boltznann method (LBM) t94l md msh fc. BDF n€rhod [93], rhe lattce collocatioo mthod I95l have beei used to sody Eq. (2.I ). TadmoE 196l poved the wcll'p*d.ess or Ks equat'on ln this cbapLr, w€ d€velop nesht ss ncrhod of In4 with ditfcrc tvFs of infinitely smeth dditl bsG funclid3 fot dE nun ncal solulion of 0E Sendlized Kudooro-sivashinsty equation. R6t of thc chapt r is organiz.d 6 follo*s: l. slion 2.2, we briefly.iplain akontnn. In Secion 2 3, sLbilirv in tems ofeigenvdlues of nne iitegation schene h discused. In s*ion 24, wc p@sent sone nunencal exanpres Elaled lo diffc& q?4 oi CKS eguation. Scclion 2 5 @tBins discussion @gddrn8 sel6tion ot shaF pdaftler od tale of conv.lSencc ln Salion 2 6. wc sulmane $e 2.2 D€icrtplion of Method coosid€r YY+ tlt cKS Eq. (2.1) u"!:+ u i\+, "=i+ n wifi rh. following initid il =0, o < rs r.,:0 and boundary condnions (22) <r<4 !(r,0)=/r(r). a zk.,)=4t), !(t,,)=4t) Q3) To appl) m.shless mthod of lines, *c first use ladial basis functions to approximte poblen domin ta, bl n divided into nodes,,.i = l. space derivativ.s. The of$ese poinc {, i= 2.3,...,lV-l dc incriot loinll- Thc appreirut rurction of !(r) is Poinls 2 . N Out vhile r' and rN aE ln. bolod2tv dqot d bv /'G) la u I t) = >)i !'lr' )14 = Y lr)t (2.41 wha.t'r @ untnoMtn d.Fnd.nt $8titi6.nd P $m ndirl Y {4=Ls, (r), yr(r), -, w. hr$ u$d {,"(r)l hosis turctd t d . tlE followhg RBFS y,(r)=!(,-r,) +.:, {r0) r, {r)= cr r(-. (r - r,)'). (0,{ ) IINQ) whc|! j =L !d . Z ..., iv r,l !'(t,)={, b.ing 6h.tc dEn ir mtrir-v.ctor !.ntEar. notltio \2 5) whccu ={4(,). v'(', t)t' ed r=tr, (,), rh(,)......1,(r)l' - v,(,,) '!,1\l ,r,\',) - ,/,(\) v,(,,) v,G,) ) Y'G.) Y'k, !,(,),..., !, ) ' r,,(,") r/,('") ... e"G,) F om Bq. (2.4) rnd E9. (2.5) i folows dEt r'(.r). vr(r),r-'o = w (r)0, (2.6) *h@ w(r)=Y'(r)r:'. Applyint Eq. ( 2.6) to Eq. {2. | ,. ad @llocadnS .t 4h node /(l..(r.'u I r, (W,,,tr ,)!) + r=1.2.1...,1v, l+.. ir{$.,,.tr,,!,=0, l:jr+,.rW.tr.)u) + 12.1) l7 w,(J,) - Iryr,(',)w:,(r,)...Pr, lr,,tr,)=;*r(r,), (r, )l i = r.2...., ^, v.,(r,) = Iy,,, (r, ) r' 1,, ( r' ).,,x' r., (rr )l t(' t), i=t,2,...,N. w ,,\r,)= In onl.|toMite lne.bow sy$eni! i;v u=k,,,...,,f . l'rn of tic colum @tos, w, =[w,(r)]"., w. = lwb (.,)]*.w* =[w,* (r)]*,, w- =[r^(r)]*. dM, q. (17) @ -+U.(w.U) | he Miro a folb*6: /W"U)+(W_U)+'(W_U)=0. (2.8) R.yiri.r &. e.8) a 9 - " ,u ,. t2.st fl (u )= -(u { (}Y,u) r r(w,,U ) + y(w_U ) + ?(w,-u)) ad th. sFDol ( * ) delo(6 cory.nr bt cdEpoent hdtiplird@ of two vcc!06. Ti. U(,o)-[,'(a] !"(a)..- !'(+r] 'd(aU , rtd nF bou!d.!y c@ditiE d.sibcd i! Eq. (2.3) r, (r)= r, ('),!,l')= rr (r) {2.r0, rtl (2.rr) ofod.r four ro solvc Eqns. (2.9I(2.1l) 8i!cn asl i,) r, (u'). r, = l' (u" + +,(, r(u^++r(, = \2 t2) .] a{u'+Arr,) ,J.r, ct!. ol dssic.l Rungc-Kulla m.lhod (b.inl one srcP $hcn.) no irui{biury n6d to b. f@d ptovidcd th.t lh. tiFe s|.p & is sumci.ndv shall ccc Coltltz t tol j, pas. I 12) shich h s in N casc Thk is .PPlicabtc lo srren of cqu.tions !s wcU s singlc cqu.don. Fd mG d.uik E.&s ce dso visil LrVcqE It h Ehe*ld 0ul in tl|c ([02], pats l5G!60). 23 Srab ity Following thc rcfiod of tin. rppmrch airy ndi.l b6h tuncrio.t Eq {2 conven d !o. sysrn of ODES (2.9). W. wric 4=ir{u).0.,<r. t .borc sysGE 1). ha be. a (2 l3) urot = u, wh.E U. is inillal veror. For srabihy @lysis wG [i4 g:'l stabiliry bv rim scD a, -J, +1W,, 0). Thc of lh. J&obie r.lrir {',2-a&} b.lonB! to $c Eeon or abelur. our cs ld cbsical rounh ddd Runs._ Jrcobid is gis. oftt milinwllrn [>,{r)w, (t,r,*{r(r)4 {i t))..,. 1,. lr(44(i,t). Ir s4rioi by +vw,,,1rrw,,.. whce J, i! th. rr.obid J. = d "."lcd or tirc inictnrio s.h.nlc, whi.h i. Kuru rct||od is c2-?8, J= will pov. lhrt .ll .iSdvdGs (2.4) bchlvio. ot cig.nvtlues for (2.14t U*(rry,U) d.dEd ror d'Asona] a cr.mnb fo no'tdj€dal.lelmts difidlnt nu@dcrl .smPl6 2,4 Numericrl Applietion In thh seclio. we apply lhe abolc mention.d ehere for nun€iical solution of OcncaliEd Xumoto-sivdhiBky cqudion Thc tesults @ comp€Ed *nh the exacl soludor s*t $N nention d in I92, 951 Tt'c Slobal @lativc c@ (cR.E) h calcultt'd bv rh. fmula d.fin€d in [9t zl!4 tr'.'J-!* t,r.,,1 OiE=r , j=1,2....,/v , ii t'l =i LV'-\'' whee !a ed rq, m aPprcxim.i4 t Fot p = a = ^.d (2 l5) solution and ex&l v = 0, Eg. (2.1) becomcs slutio. Bpetivelv. th. KS equarion [96] au" Au a'u . a'u Jr )r Jrr , {2.16) dr' (,.,r', . *\Ft', *'' T|E hdldlry @ttditios e (- (, siEn -,.))-' trii lr (r - r.))l (2 l?) bY f \Et',""''(- (4-,,-,.))-qtrii (t(, Dr-,.))l (2 rE) ,(,.,)-,.i+rF1,,","(-(,-,,',.))-,tr,i(*(,- D1'.))l (2.r9) -frffl|l*r'r-(r- D,-,.))-,nri(,(,,'-,.))l (2.20) , (,.,) =,. ,t".,r=, For nurctiol cotr pulation numricd eutt\ N v. h.ve scd MQ. GA and IMQ Edial bdis functions Thc obtdftd wirh tE pe.amE6 D=5r-+J"!.\--l2 over 'h' dotuin ta bl = t-3o' 301 ulins 1v=t2l Thc tlobrl tlat!rc @r (cRE) is calculal.d ar , = l,2,1.4 wirh dre srp &=o.ml dd tlE Flults e lhdn in Table 2 r' Thc oflimal valuc of $c sh'pe p6rame. c for MQ b found to b. iaid. ,E inteflal (0.7, 3) B shown ubh fic rcsulB givcn in I92l using r$icc Bohmam wlhod with lV=600 .nd Ar=0.0001 e al$ shopn for conpdison. Wc ctn se tal our mdhl$s N$od of lin6 (usi.g MQ RBD F.fom ve[- Ir'e eguld .hck wav€ prcpagalion is .lcarly visible at ditrcEfi dm levels in Fi& 2.2. In ord.r 10 wdly sl.biliry of rb. m.thod for the abov. mcntioned p.obl.m, nsr in Fig. 2.1. In this we fix rhe nunb€r of .odcs { obsefred lhat all eigcnvalues . within the inleryal (0,?, 3), ll is lie in th€ lcft hrlf plane a long as c lies in a rcighbodood md changc the value ol of optinal value of |hc shrF psl"@r.r. So ln. rethdl Etuitu rtable in dis shown in Fig. 2.8. 81, h l2l andc=2.5, lh. sond ce E kccp . fixed &d v..y /v. In this all eiCenvalues hav. n.gative (e ce, as for /v = 41. Ed pan rs shown i. Fig.2.9, So.gain ncthod is st2ble. We know thal the accuncy of RBFS sohdon depends on valuc of the shlF pa@tcr c dd c ud number of nodes lV /V, but at the .d of al$, Accu@y cm bc improved by iicrclsins conditioniB, This f&r is shou in Trble 2.2. Er.mpl€ 2.4, Consid.r $e cKs cquation [%] a! au a'{ 1-+-.0. J'a J'a =-r-r:-+r ' 3r < r. r:0 !('.0) =D+e r5[ldh(](r-r,))+'dh, (r(I-r.))-r.ih'GG-r"))] o 2l) \2 22) ,(,.4=D+,-ri[hhtrk-Dl r.))+bn,(r(, D' r.)) dh,lrk D, r"))] c.2r) ,(r.r)=D+e r5[6h(r(D-Dr-r,))+hh,(^(b-Dr-r.D-hhi(r(b-&,,r] \2.24.) Thc cracl solurion for this eimple h given by r.))] (2.25) Thc GRE is compuGd ar l= l,2, I and4 wirh tin sFpA r=o.ml d p&amt€n valEs e sl*led a: D = 6, I = +, r. = -10, o€r $c conpur.liotul dotujn !(I.r)=D+e rs[b$(r(r-r-,")+bnh,tG-Dl-r"))-@h'(i(r,Dl h, bl = ['30, 30] whh N:121. TIE @ulB ac shoM in Table 2.3. Fron rhis rable it is cld tltat dE Esulr ouaincd by ou @shl.$ Nthod of lin s ($ine GA, MQ dnd IMQ RBF3) e bed.r 4 compoEd ro d€ Esultt given in [92] with /V = &N ttd Z | = 0.01)01 Th. optinEl value of th. shaF pam@cr lu MQ li6 insi& th. in|ryd (0.5, 2) at sbown in Fig, 2,3. ln this c6e th€ solit'ty wave prcpagadon of GKS €quation at difteent rine levels is shown in Fig.2.4. For this poblem wh.n ,{ is fix.d .nd c cieervalus stisly stability conditi@ of RX4 difreE val@s of tlE shatc pdmt bclongs etlFd. Ir r is shown. Fq to (0.5, 2), dEn all Fig. 2.10. shble sp€.ttuD ror ! 6red c, and diffcdt valEs stable nurcrical rcludon. eiplcn RunSc.Klua ne$od Frfom wcll giving lhe fte of /v. scrLd eigehvalues for this ca8. de shown in Fig. 2. I L Esmpl€ 2.4.3 In this exmplc w. consider the XS.4uation tE9l, #.,*q.#-#=n a<x<b. t>o Q-N) This examplc rep@enls thc siEpLst nonliaer partiar difredntill equarion showin8 cnaodc behrvior when sDatial dotrEn is fiDit., with the Caussian initirl condnion, \2.21) and lne boundlry conditoE, !(d.,)=0.!(r.')=0. t2.28) The conpuntional do@in is [a b] = {.30,301wilh and Fig. 2.6, wc cm ob€fr€ th€ AJ= =0.m1.In Fig.2.5 converSent nlnedcal Bults by our m.shless ne$ols 0.5 and A | of lines with compl.tc chaotic behavio. at | = 5 and r = 20 Esp€ctivdly. Stability sprctium for ditr@nt nunbd of no<les lv and shapc Fie.2.U ad Fi8,2,13. Frcm ltcF figurcs $e w parh.cr c is showo in tnat fcw cigetrvolues have s@ll pGitivc Eal pan showing a snall .xpoEntial grewrh in po6itivc h.lf plde bnt erhod wotu w.ll in $is tE c& alF. In this €ianple, wo consider the case with 4 = 0 and regative siSn before I . ln lhis case CKS equltion is converred in to (dv-Burg.r equadon [97, 98], t-,*-,*-,*=0. asrs6,,>o Q.29J !t).n\=)o-----j!- (2.r0) lr+.\e(r) , )l' and the boundary condino6 Ir | .re 1,, {" -,!))]' (2.31) (r, (, - (2.32) + Ii Tn..4t solution i5 ..e eiven ' ))]' ! 6 (2.13) l0v 25v For nu@icd c.lculnios dr=0.1 .nd & = 0,01 fte cRE calculltcd 2.4. Fom thc tlble - w Th. ta&. 0E comput{iorl domin \ElB of th. bl = la. al wilh p!F!eie6 re t=0.0009 dd v=0.000@. ar , = 50. 150. 250 bv mcshl€s *e cd se tlDt Ir turhod of lincs is shoM in Table rhe resulB obtained by ou nethod (usine MQ RBF to 0!. Fsul$ siv.n in [9sl *ith !=E07. Gnphic.l Epr*ntation of Esults is shM in Fig. 2.7. Fq sbbiliry ealysis se F.f@ two diflmr qpgitMls. F6t w. chdge the Bl@ of conpoEd the shrp€ pdam.t r. insi& fie fixed. Wc oh.ryc that $ long i!l.d.l s (0.01. 0.6) by lccping the numb.r of nod€s the vllue of. cnaitu iosidc this *lecled interv.l conraining optimal value, au eigenvlhcs have negarive ical pan, when we iocEasc c oltside lhh intcNd some eigenvalucs novcs to ngnr half planc showing very s@ll e4onenrial grcwlll val@ of . s shom i. Fig. 2,14, In s@nd is kcpl fir.d. A spenub sirh sheme is showr in Fi!.2.15. .xperiDc w. ch6ee x dd tlE dl .igcnvalus imidc stdility Mge of RK4 CUpE4 2.5 Dbcussion Ihe .. uu&y ln lhis oI RBFS elutions hcrvily depends on thc choie of th. shape wo* $e idea b€nind lhc peamcr Pm@cr is b6is functions is in*niblc $l4tion of lhc optimal v.lue of th€ shaF inBal for c in which |h. narix A of tadial aftl tlEn to sler a value fom thal int.Pal giving fi. mst @mtc Esult3 ln our nuBic.l exledrents wc $eh lhh vllE bv Plolling 0E Eladmship bcrwcn |n' shaF panmte. wiih sEp o.0l atd lhe oaioun+ru' Fren Fig 2.l 4nd Fig 2 3 w' ee $ar $cs vdB e *le.t d iion the int dals O 3). (0, l) ud (04' 22) Esp.dively in 0!c cN of MQ for ExMPlcs 2 l_2.3 |f w. furth€r inc€& lhc valuc of ' $c dovn because bEats RBFS soludon $e Mtioned in&ryals th.n fie above fon cocflicie mafiix A becomes highly ill{onditioncd, fte rade^ d€ Monnend'tl lo to fiBt find the rcad Madych 1241. wendland 1991. chens Il0Ol ard cgdditrg pobleos of conditioning Tbc point vie 3tcp ! Lherein lor morc dek'h and accutacv rac of convergen@ in spoe and timc de cdcltar.d using lhc following '*.q1iJ&r,^l)-" whcrc RteRms *ffifr# EprdenE.mct slution ed U{ cPGscrts !h. (2.32) Nmicd elurions *ith sParial sia,r,. we calculate spatial Btc of convsgene by ktrPing nme srP A'=0001 fixed and varyi4 the nunb€r of collocadon poinrs Table 2,5 we cm sik, In se thal the conlcrgcnc. rat ('^r = 30 60 120 240) Frem lhc incrcases witt the snallet spalial steP tire rat of convcrg.ncc wc tcep the nunbet of coU@atio' dne steP si4 & = 0.001, 0 OOO5, 0 00025, 0 0001. It can be nolcd order to calcubte points fixed fion Tablc ud vary 2,6 thal tbe tinc BE of co.rcrteic. &cca*s with @dla dme $ep' T.U.t r: ConF.is old rErt &*in[y2]fd{nPh2l otE tnl T.U.2l: Cdniuo nudtd of A vi.hdiftrs v.l6 CIE Co.didddiha I l.23ao(lor 2.ms?tld . 2 1.1655!l('1 3 r.6arxtor 7.lp35xtd 3.il935tlo'l T.bL rJ: Cor|Fris of N ,v ,rt 6r l2l nErh.dt wnh [92] fG of c!iNford.ntlc2l GtA 65t26rtoj Cddrion i!n$.t 9.l.lgtrld f.303&l0J 2.t24rlt 4m26xlor 1lo4xl0n .mplc l 2 te2l 25 rrbk 2.4: conpds ot N n*rhods with 1951. ld 4Nnpl. 2v) hbL2J: spdial l.Dl. Tire | -5, 2.6: dt * iv= 24o,.It of of mvsge${E5 dvd$re lr3l5 2 4 ! a..F*'r ItE FtscXA SheL r.rr E&|{L 2.l rin N F!fiL *rwp&fitdd otlt{tt|M'Sl rd&!.d IiF !'|tdiu.rd !i fllslol r I2l rhiaty oqdio.{Solid blditr n pdiElt). lird I 3|F!||fu'l.l*. rta ttE 43. B[taL 2,2 {ith N= l2l. rE Fltgio olldrlr.d /@nbstiycliEtt lqdio (soldriE't!.En.dDhrid.n.idFlirl'r,Fielsl{ir)2-A saliEy Flsc 2l: I'. Cladc 6dM F4a l& TIE odih Sollnjd ofri. r\S Ec. G.27, . r clsic ltol'rid Eq. o,2?) I lqdid w h th. . 5 b, iEdtL. isiod ofdF r = 20 by rls.qM crsir or IiE. rnh di lEr'ld dhoa of iniri.l c.a lic. iiirirl rtn a7: S.ldo c@ ad sludon for ydv Bfls .q{dd rd i' .[.vlu numdci lsold liE nrr.sr $ludon). e € B I B s € 1 Ite i.b 2l! Fi8dds dLd ol by [email protected], frF Fr@ Icd{ N . A rr 0,()01, ailt l2l (t!.d), ft. ditu BrqL z, vdEl ., 3! oEooo o oooooo@(Eaa@DrlrMn o -r.2 n 46 4.6 4.4 {.2 0 flg@2J: Egavrlua*d.d bytlm&pAr=0O0l, forBffipL 2. din Eorvrlu6ofN'r.ting'. 6red. BiDg d2 o6 g-"o o o o o ooooB 02 @ !" E E"' -oo ^-^no.Qj 46 R..l ulc ilN 2.r0: E!ffns sLd orttlr. bt rir F|.dg ep a, ftr = 0-q)1, Esqk 2e tu ditud r.ls F {,6 {,3 n -1.2 n llctt} {.6 46 lra {.2 d6t &!oEl6 FLd bt dE & a, = 0or, o hdlti.d :ta 3 l0 E I n {,3 {.6 {,4 4.2 0 0,2 F€rl{2) rtd 1ta EisordEr add b, tim .L! A, = 0i01,l8 dilft'at of r[+. Fl@c ? 6r E qL dc 2l- 35 I'E 2l .6 -1,4 n,2 n +! {),4 .0.4 {,2 0 F.dF) ftlEt 13 EaNlr.rLd otN !, de q |f Strlb Ar= 23. Oml tudift.d dq 0,2 E -l {:6 {.C a.Ota .o0l 4d 0 0G F.d{z1 ItnA+BtFv - sr.d olai.tc ht td a{ A t - o.ot, tudfrd v!6 c 6.Er.4L 2.a. a7 N=A -l {i-7a.54€{-rO1 hd{z) r r{ rtn ll!: Etovds sL.l by iiD r.p A, r 0All. ft.dlbd vdE 2.6 Corcluslon In this cMptcr, wc hsvc rpplicd mclhLss n thod of liM (using MQ, GA .nd IMQ Fdial bsi! furcnoN) lor thc nun ricd lolulioo of &n dizcd Kuuoto'Shivashinskv .qurion dd thc FNllts a$ v.ry cl* to th! .x&t $lution. Fom rlE nudEncal crFritunl!, *. cln c 0|!t ou rcthod b€lcd on MQ 6dirl bair lsltctions p.rrm w.ll s coqor.d to o0s trDthods giv.n i. tlE lit'nnlf.. ft ha to bc .npnlsi4d th.t shlF plnmt r fd .ll th. calculnids Ftfom.d in this Pr!6 *.r. roultd dp.rilMnlly. ln all nuo.ricd crmpLN . tt bl! lFcEun cln bc gn TlE PrePo*d @thod h ustul iD $lving tin|. &pc!&ni @nlin.r PDE3 AIS tha lrdial b€re tusd@ t v. l|. Flp.tty io tivc lMh higEr .clEy with low nmb.r of sPsid no<b. Tw n jor !dvr!trt!. o{ d rthod at! ttl. Bltlets proFny rtd us of ODE elwB of h4I gldity ad tlFir .oib to rpli@h {E solutioo of PDE!. Chapter 3 Method of tines combined with Radial Bo,sis Functions (RBFi) lor Numerical Solution of Kawahara Ape equotions 3.1 Inaraduction l. lhis Cha0er. we .heuss followin8 du ttu du du 9-,9.i.*-*=n fom (l.t) a<r<h. (3.2) 'zo (3.3) d<I<,,.>o in pl6m {103i, wirh suirae Ension 0041 &d capillaty sraYity *alel waves tl05l physical ph.no@D shallow ware. wayes th€ ^ 3.*9.*-*=0. A vdiety of Kaw.iua type egutioos of lik, na$clo loustic tares se desrib€d by l(awanda €quarioi (l.l) and nodificd Kawanda equatioi (3.2). Kdv_ Kawahda equation (3.3) is used to dcenb€ the one di@nsionll evolution of small bul tinite amphudc 1o.g wav.s in vdious Pobl.n! specilic fom or Benney-ut equadon [ 106, l0?l i. nuid dynuics Thn €quaion is a DifieEnt dalytic and nuhcricat rctho<h inchding $. Enn_function nethod Il08l, Adc'mid decompcition fttnod ll09l, siic@sie relhod [ll0], varialonal iEradon D.ihod. honoiopy Frturl i6 n.thod [lll], Cr.nl-Ni@len DifieFntjal quadrarE ncrhod lbonlntu lll2t. Pedioor coreld @thodr tll4l ard RBF @ll€atiot mlhod rswah@ rtT€ eq@tros. [ll5] [l131. Dual-Pctrov calc.kin h.ve ben ptoposed for solving lhc In lnis chaptcr, wc prc&n1 m.sh t c num.dcal nerhod fot solution of above equdons. Conenadve Prcp.fii.s of lhc Kawahda lquation Elaled to tus, monentun aid erergy @ ale iivcs!8aled numaically, ln se.tio. 2 thc nu@rical schede h rhc nuNrical examPlcs fot the jlslinc.ioi of te This cbaptcr is oqanied in th@ se.tions oxplained md Fclion 3 contaitu nelnod and ve @nclude in &clion 4, 3.2 Numcrical Method For inpl€nentrtion of nurcdcal @thod, wc consder wi$ the fi. modificd Krsahm Eq. (1.2) following initid dd bounddy condirions (3 u(a,t)= 8,(r. u(b.t\= tn pE.d@ !,(I)= Y r,,, (r)= v !,,,,, (3.5) &\n Following atlogcd in Satio. 2.2, wc rtPorimare 3podal d.avatirs 1(r),t-'u = 0, (r)i. 1., (r),{ (3.6) -'! = 0,,, (')o. (3-?) {r). v j,,.,1r)r''i . a,,,,, {,)!, using Eqns. (3.Gl.E) in Eq. (3.2) lnd collcrtnrg *+{!, )'(4,(r,)r) +(a*ia)!)-(Q*(r,)!) (3.8) ar = ach o Q,({)=Lq,(',) q,(t,)'-' 0*,(r,)l ,i=1,2,.-,d, Q,\',)=iQ)o,), i, 4) j=t, 2.,.,. N. r sPadal =1,2 rcde .N (3.9) 0-(r)=[g-(r)'o&(r,} ., q-(j,t. =t'2, 'N' Q-(.r,) = [q-(.\), 0l-(..,), - , gb a@@=#ap,)'. j=t'2 .. N, oBG)=rta'G). u=1,'.,' ,'. .,,. = x t. i Now w. us. th. followiDg o. i=1, 2.... (r' )1, t = r, ? . x, mlum wctoB not tim " ""J'. [o" ("' )],", a- =[a*(1)]*. o- =[o* ({)]* Eq. (3.9) k th.n conren d to ih! fom #+u:'(a,u+(a-u)-(a*u=0. Aborc calatio can be win n i! the fm 4=yo wh.E v (u) (3 ro) (l.tl) = -u' + (Q,u) -(Q-u) + (Qsu) aid (he synbol (*) d€ioEs coDpo@nt by cornporc nultiPucation of lwo ve'tm ftntiord cdlid i! sation 2 2 u('")=[!'(',) ,"(4) ... ,'k,-,) ,"(',]]'. F oE rhc bounduy coidhions q(r)=c,{,) Nov &snh.d in (3 5) wc (312) g.t -d,,(,)=c'(,). o.rr) w sc clasicai founh odT Rurgc'Kulla slm (3.13). N!rcdcal sheN for foloeing dt .bos a orher two c$adons siE (212) to solvc EqM (3ll)_ in se.tiot (31) ctn be obancd Pd*, 12 33 Sbbility An lysis ln rhis scrion, shbility of tn. first oder systEn of oDEs (3 l l). obtai.ed RBFS appmnma on r"-bhn Dr.tJ =-J, Q* = fd spari.t d*ivativs is ([o'(u)]=[iP]) c.4 dalv4d as a by calcul.tinS eiSenvalucs of the bv, -(O--a*), r,{ ', md Q* = rcaul or (3.r4) c.4', qh.E A antl C m 4!4 m[ics whoe cnri6 @ of $c fom a',2(,- rl) - a'*{1"-,,1) -4=-r''"=_-F ti Eq. (1. | 4). J, is th. .l&obian of nonliicd Em Ur ' lt 3l!tr rr J lk (r)' r, (r, t l'. .(?!l,1rl*ls.li.*i) ). (f U), which is siren bv . lor dt'8onal clcmnb. for no.diasonar crcmnts In the n xt srction wc will fhd $ability sFctrum for differcnr nurcrical exmPles ad di$us $abiliy of thc rethod, 3.4 Num$icslrpplication In this sdon rhc numdcal Esulls fot K.wah@, lEdified Kavthe and KdV Kawdhan €quadons e Pesented. Fo! Kavalea eqution (3 1) the lovesl tuee @nsp.d quritiA Elaled to nN. e8!y dd notuntum delincd ii [115] @ r,= ld'. t,=l+u'dt.t.-l l++ *r'++#rJaI consider rhe Kawshda equaion (3 1) a! ar a'! a'! ^ wnh th€ following inidtl md boundary cotditjoB !(!o)=15sd/,'{*(t-4D (3.15) !k,r)=0;!(r,?)=0 (3.16) The exrct solurior giv.n in ,(,.4 =-@s/" ehetr 1= For (i--!,,i)). (* (l nurencd computatio. we t l@ [a,r] using MQ ard GA tirc l= ndid thc neigbborhood of $c l?) l r. is cMi.d otrt up lo in Il12l is. 4 25. r? atd L basis functions as shown = [-20,30], nom e a = 2 and ,v calculalcd vith th. valoe of shaPe ,t : 51 rhe sinuladon r:0 5 15 dd 25. par.ncl.r found m be in in Fi8.3, I lnd sinildly fot GA $e value is fond to be neigbborhood of 0,3. we have seafh.d $e optirul vah. of the sbape Paflmercr bt plorting naximun edor 864 shaF pslnel.r wi$ st P 00l Th. lhr.. consrycd qudtiris e ats shoq in the TrbL l,l,Thc upliru&s and Fir po6nioi of $. sotiary vava e.lso cnlculdt€d. Thc ttulls wih preent m.lhod 6ing MQ @ betcr thu the polynomial basl dinemtial qudr.tuE (PDQ mthod Il | 2) and re v€rv closc to cosine expansion fovdd b.sd differcntial quad6ru@ (CDQ) nerhod [l 12] notion of lie slitary wav€ in conP!tuon qnh qst InFig32rh. $lution (3.17) ar ditfetnt $abifty spcculn lc this preblen is shown in Fig 3 l0 ud Fig,3 l I for diffcEnl values or sbape pdmcter and nunbet of nodcs Especdvdv The physical benavior of cigenvalu€s has nild exponential grc*th, which is cxFcted in nun.rical apPrcimtions b rhc @thod rcnains appli@ble in tl s ce. Th. Poinl *is tat. of convdgpn@ in so@ is calculat d by Lcping ti@ saP Ar=0.m1 nxed and vatyiig de nunber ot collarion Doinls (N = 20, 40. 80), from $c Table 32 {e can se€ $al the order o' co.lergence d.cE!$s wilh have usd MQ and lh. smller sp6tial slcP sirc' ln dll nunericol €xanPles we CA in order to calculac ord.t of conv€rgence coNider d€ modifi.d Klwaha.a .qutioi (3.2) -+r -+---=q wirh rh€ followin8 initid md boundary cordnions !G.o)=Dsr,1(rG) (3.r8) !k.4= D*ci:0k-rr)r!{r,4= Decrl(r(r-84) (3.19) 'Ihe boundry conditions e extiacGd ftotn thc cxact rlution (giv.n in {l16l), r('.')=Dsr:(tir-84) ',The calculatids @ (l.m) 1l tri.d out by ErinS 1., bl = l-30. 301 wirfi t = 6l.wc Ds Mq cA ard lMQ ndial b.sis widr sh.p. paElrscl found !o bc in fic migl|to.h@d of 3 ror MQ s slb{n in Fig.3-3. For CA Espedvcly. The a, dd L dd IMQ tnc vatuc of nms c.lcubt d {r t - . lies in vicinity of 0.5 and 0.0m1 0. 5. 15. 25 @ shom in Table 3.3. TIE ordcr of convdgcne in spd d.c@Fs *ith imMsing lva show. in Table 3.4. 'DE $litary wlvc profilc at diff@nl dN lcv€lr ii.ompdisi with the cxet solutaoo is In this cas shape ssLd eiS.nvalues with pcarcter md nunbd ot nodcs d sndl pGitirc Ml pan for diffeGnt shoei in considd lhe Kdv- Kawalaa cqurtion (3.3) wilh the followinS inidal Md bound.ry cddnions valnes Fig.3. | 2 rnd Fis.3.13, @sp€ctively. of ,(-.0)=-qs4/,' (+(. { )) (3,21) (22) ("i;(.-#,-t)) ,(,.r)=-ese,,'(t!(r--4,-)it ,(,,.)=-s*r The initid conditiod dd bdrndary cddnions (3 23) N ext&t d fiom dE ex&t $lution (8iv.n inlll?l), )) ' 160 t2Jrr( :!:,-, "',."=tls*r(-!L r6a ".JJ rl2al The simulaion n perfoned by uking Ia, bl = dd 10. 2001 A, = 1. The dis.r€t ot l- @ calculaled using MQ. CA and IMQ 6dial basis furcrions up ro tihe, = 5. F on tbe rsulti shopn in Tablc 3,5, wc clD se dut the both MQ and CA !E showing very good .grenenl vith Lhe ex&t m.an sq!re €ror nom Lr maximun sludon, The spao.l Etc of @Dv.rgcnc. d.cEes by increGing olldltion crcr nom wilh is shown in Table 3.6. Th€ otdet of conlcrgcDc. poiou for a tiied time srp A t = 0.001. Thc fo sd !t dificcnt mc lcv.h in @npanson Mth th. extl elution (3.24) is shom in Fis.3.6 (s.@ s in [18]). Eigenvalu* ofthc rfobie mat ix with snall pGitive E l pdt hav€ ben shown movcmefl oi lhe soha.y wavc in Fi&3.14 and Fig.3. | 5 EsFctivcly but it d@s not ha& lne R(4 mrhod utupplicablc. For intenction of two pcitive solitary qavcs of equ.tion (3.1). we corsitu th. following initial cohdltron ,1-.0)=ia,*/,, (jF(,-,,r) we solle fie problem by usint MQ la, rl ro tirc = [-50.100] wirh r= cqladon x = 2ol and GA Bdial bsis tunctiom. The sFdal intctvd is scl4r.d. rhc nutuncd erpe'irent is 50 wirh dne sGp a r = o@1. Thc vrlucs of Oc e choen s: r, = o, 4 =N. o = 4l Jr(,f., csicd our up pom1e6 usd in dr abov. 4 =ro/o an 4=E 6 $li!.ry wav* prepapt roe.lde righr s the tie prcga6, The p@ss of im.rutior is shoM in Fig. 1,7. DunnS thb prec$ dE lueer wave cakh.s uP lhc The teo snallet orc a.d tlEn th. bolh warcs scpec frcn .!ch othe. m.intaining their @gimr sb4e. Frcm Table 3.?, w. can se tlut th. inv&iants of moridn €min.lmon consened .s ram incEases. The vriation in th. fiEe cons.rycd qutnlides is found to be in $e lL 140.509259< MQ <40.4838916.1 140.509259< t,3 40.412842, l s 45.84J648 r< /. <45.8509184.I. GA i4r.8jor4 145.8361413Ir L-J.2 J7:ro< r, < -32.15a9r fl2l?2189st.<-J2 I Fo. intc@fon of ,',.0,= f@ I. 14082201 solihty wN.s we cotuida .qution (3 1) vith the following i,c '*r'l::ru-r \= rl. we solve tlE p'oblcD by usins MQ !d ar=o ?5 TlE.alcularion h cdi.d our val@s oi rh. pal"@t E ued aA fsnctions orins [4, t]=[-30, r20] wid tmc r =50 with drc slep & = 0.001. The rh. abov. equation @ choen 6: up lo in =.xi, 1=0, 4 =zq o = tf ,!tu, e, =rq ". e, =u1" .a e, =01". Thc rhe solibry wa6 FopaSaa owards ddr, The pta.s ofi etution is shown in Fig3.8. Dring this proc.s tlE ralld qavc mv.t fast r dd .archcs up the sn ld n vavd and rh.n rh. th@ wav* scpatate vavs aler coubion motion Edin fon @h o$er. Tte shapc of tlE is mlnlaimd, F oD Tabl. 3.8 alnost consesld w. can s |h 6 tim inc@e' Th. vdialion $@ solitary de inv&imts of the in |he consrycd qumdri€s is found ro bc in lh. rdgc: <514?]E98. I [51.4726313/, <51.r04t2. MQ j5r.roo4ars /. s5r r5838. CA l. 151.100491<t, I. /. <-l]069?3J <-Jl.3036l l5r.a726jl3r s5r.568591, I [-jr.6]8063 In this exanple we [-3]63806sI. qill I show thc phenomena ot wale Senention for equalion considd$e iollowing iniial condidon ,(,,0)=/r.r'(#(,-a)). (l.l). we Conptatlonal domin [-40, l30j with ,r = 0.625 is coftideEd. The sh@ is Nn upto tim, = 18. W. tal@ /=10. IlE 3i!sL 0olit ry w.vc aplits b 10 thE€ rclit$y waves. Thc cqed.d q@!ti.!, h.idt &d pdition of thc els e cd@lat d .l veious tirc lcrcls d slbwn ir T.blc 3.9, With th. D$srge of tinE dE l@ditrg wrE owinS ro n5 f!.rd v.l@ity gcts f& t@ ttc othd two er!6 a 3bown in Fig3.9. veiaid in ftE coNcrvld{llD iries n fdDd to b. in 0r follosiry tu&: <r,<96.14165[ I 196.1a365r I 3 r: 3 329.65039, GA < l2c.6503ec, 196.14165rs/,<%.lij65r. MQ j 329.65O39e -< 1329.65039 l. 'r r, s-70r.426ElEl r, <-?06.E E96j 1,875.4160413 [.875.416123< l. chopet 3 Trbt 3. l: R6ulr ld Krudle €qunion in @mDtisn whn I t2f250 0.15362 0.163t3 1272502 25 PDQTII?] 2,5(t CDQII D] 12] t IrbkJ2:Sp.ri.lr .of @rrc4EeartuEspl.3.1 t=25 fi'd. MA Tdlc !3t RBult3 for Modified [email protected] L- _l 1j;Ta l5 5 0,311 o,m3 TlbL J'+ Spdi.l rE oI mqgc|* r for ExmPL 3 2 | : 5 LL 6.7te3 T.bL 15: R6!lt! fo. r\dv los.hn cqurdd .l E5 3266 59m0 59ti.8 t.n8 T.tL IG SFd.l re of @u8de { T.tL lq in|*rid 3?r trvMsts Ib. of r*o GtnPL Sl I=5 tolihr fd EEfpk I t ahopEt 1 T.nL 3r! Irwieb nr inldid or |nE rljhr f6 Fwnph ]5 -31.6]306 51,472631 -ll.l90t5 r.8rt962 51,63t932 5t.tcx53 51.r5333 r.bh !J: lnveieB fq rnE@io of 6E oliloB tor EN|Pk r.6 MA 1.5 t2.5 li:11. m tl* 1l &d ffid.F F.rec rrSurbl.l. FllrE 31. Tr.slung wrs toldid .howlni .rEl tolutid of &r*rlm lqurioi (!olk llmr {d tlhowiig Nrt&i toludon) 5,1 d|qPatt' nr! lt! rtn:ltar So|t.y &E tE 16.ondd dd rtid(.tl lh. d.F ot !ao' F,.dr c b B!!p& 3r, !.dn d v$br..$ao qd.rri. &.t lr or!.id en t tldl'l mElcd .l'tin), Itc!J! &ff |!..bFFn|F. I{a- !4 SorL, do{ rE dd.a r.dv bl44lt3 3, xrwls.qdbr (dn uc 4d ot|bn !d t'iFvia .uEial $ldid0. liaEr &t! bEed.n orrs .liM .qu!b,r, BUDDb 34. ror K.*rhra ctpoE ItF3t rrffdr oad6.lx.! fa BLrd. 3J S ItE:t': Gdid olffi ft. Er4& J,6 59 -9- I ls oo." tlr.llc $68 lo t., tu Erqb 11, s8: ftNarrr SaLd iaEnls b Br4L t. t, a4g1 i .;x ItN:irtr s!6n I F ! h. BrMd. 3.2. * 1":fru"'--" IlFaDr SrLd.ia.vd6 6. E ,aL 32 61 s t e .. fd.- " r|g!F iL. d il':.' EI3.lvdar ft. EiEd. . 3-3 I € ryEi ar3: SDdE ft. &.!rL 13 62 35 Cotrclusion In Oir chrpi.r, w. harc u$d @rh l€ss m.tbod of lims fd [email protected] Noluton or d.sibile oomn of singlc sliiity wav$ dd pl'.no@ra of wave Scn.rtion Xaqahda typc .quali@. Th. nu@dcal r.lults wav., inLnctio! of two .nd $re. soliury be! disused. Th. &cu..y of thc aolurion d.Flds upo. Ue cnoic of th. rh.p. pmnlcr, shich h.s b@ s.lccr.d o(primntlly. TIE .unericd ada $iry MQ &d GA td Krwrl& eqlnid N b.!.r tho th. Crut Nicols dif@ttlitl qu&|tuE dsqirhD 0!21. rl|e inv.rirnb of mtid F@i!.d cd*nld duiot dF PMlss or cdrPur.*io fd a[ c'$. hlvc Chapter 4 Numerical Solution of Equal width Equation by Meshless Method of Lines 4.1 Inhduction Equal width (Ew) cquanon was suggested by Motriso. et al. tll as a nodd to deetibe v.v6- This .quadon b D alt madve fom of w ircar disPe6'v' waves ro qell kno{n egririz.d long wrvc (RIW) cquatro. and Konoseg_ de vries nonliM dispe6ive (KdV) equalion. Thee equations hav. slit or packe$. Th6e wltcs popagatc in mt { shble wave tv wave slutions in th. fom of wave Pul*s itr disp.sivc nedit Thee wavcs na|nlan even atur inletution baause of balaDcing of nonlined ud geneEl EW equarion rls tav. eharv s.v. elutaons blt ot l's fom disF^ivc.ftets. ryp€. Thc EW equaion dErived fot prcpagation of a! d! J lar, -+!--r=tsheE ! is lhe dDlnud. of th. wav. t Oily a fcw dalyiicd Echniqua numnc.t fludy otEw equation got )t=u,^ is a posidve ot Lhe ' lotg saves in positive i-di@ion (4.1) padmrr 'nd !ro avanabb for EW equdrion aucrdotr of has 6' r..mlE^ merhods nlve been opplied to solve thh equalon CardDer LRT as rt1_ Th'relbe $c DiRcEnl nutrncal ettl disu$ed elitary wavs of Ew eq@ion Il2ol. Archila $.d spcclral mtbod to elve this e{uadon [l2ll Zrli uFd $e nethod of leasl squ@s linne .kncnl *ith sPace{ne lin'd finit quanic BiPliie tl2ll Ranos elemenB ll22l. Rasl.n ued coucation mlhod using a tr€d exDlicn finite diffcEne @thod for EW cqudjon [124] Saka ll25! @nstucd fini& elenent solution for th. EW equaion. Sat! cl al {1261 lsed qlanic B splinc Grl.*rn nediod (QBGM). diff.Endal quaddt@ Ethod (DQM) 6d mshlcss Etho<l S.}r I l27l Dag and also solved EW equaion by linn€ ln this chapter we dev€loP a mesh widrh cquatio.. The chapter is fomllrtion of nesh fe ftc ehnent melhod sheme for $e numencll slution ot equal orghidd s follows: I! setion 4.2, w. Sive @tnod. S4noi 4.3 conuins $abilnv molvsis and S.ction contains iuherical exmPles of lhe Pmposd trrthod l! ection 4 5 we the 4 4 sumnez. dE 4.2 D€s.rtptton of rh€ method Consid.t lllc Ew eqlation (4.2) wi$ the following innial md houndary condnions (4.3) u@,n = \!), u(b,t) =,,o To apply resnbs nelnod of lircs, vc n6r use 6dial bash functlons lo lppoximatc spa@ denvadv4. wc will use muldquadric (MQ), eausian (cA) ed iivdsc multiquadnc 0MQ) 6dial {,i=1,2,,,.,, xr =r. basis functions. In oder ro implen. be the collocation points in thc inleNal [a, In meshl.ss rethod of li.es v. b] tne tuthod, let such lhat lr=d a.d use the followi.S apPoxinadons for spdtial !,(r)= Yl(r1.1 'r = M,(r)u. (4.5) u,,(r)=vl.(r).{ '! = M.,(r)u. (4.6) whe@ M,(r)= YIG)i L md M-(r)= Yl(I)/ r, $ we get the follorins disclized @1 f;.n1u.111"1-,(x-tnr*)=n I!dd.rbirit i! Eq (4.7) '-,t ".n 6n ctu v.cr| mlr[ldt L! M,=[nr(r]L, u-+.(4[-' x.(r,)-[r.,,(r,) rr,(r,)-x,,(',)1, r,G,).*r,(,,), J=!.2... f x -(r,).lx,,.(r,) L :-(,,).-. |,,.,(r,)1, u,,r,,l= I&E $t,t,,r. 8q, (4J) i=t,,,...,". b.c!@., gL*'t"."1-,("-#)-0. (4s) 1r-,r-1ff--p.rr.u1, (4.e #=-* c(tD F (1.10) | =(r-6ir.f (u.(M,u)) @ |!c ilitid rd or hddly caddo. d.$dD.d b Eqr u(6)=[ro6],r'6),...,!"(',.r!'(,.f ,,,()-,,('),,,(')-",(') (43) $it (44) (4.u) (1-t2l To solve Eqns. (4.10 4,12), we use dalshallounl orde. Rung€-Kulta nethod: U"'=U'+ r,=o(u').K,=c[u'.+r ,J r, =c[u'.fx, ).r. = c(u'-a, r,) 43 St bilitJ In this stio we syskm (4 l0r. RewntinB E9. pEsnl sBbility 4=-rliU,tM u)ll wh.e.=(/-,M_) (4.l]) i,a@t ix with Nnstant c@fficienB. The Jacobi4 natii ior this nonlired synem k given by a(-.(u'(M .u))) (4.t4) AU allu'(M un) --:.:-----j---j-- $d its.t.ftrrs eolthe fm l(rr(t)M.(r.t)) +>(!(r)M.(i,r)), -11,141".1' " o; In ow ou@rical eMbples, 4.4 Nume.ical T€sts rd re wiu di!c!$ $ability of $e mfiod hv showing th. Results Nurcricd 6'nts of lfi! aborc shcm fq EW .quatior savc, int tudon of rwo slibry vav6 .nd undul$ bd., 6 discusr€d fo singlc soliury The Ew equation poss6ses thc folowing con6.nalvc laws for nass, mom.nrun and (4.1t r..l[4' .4,^)')d. =J' For singlc slilon slurio thc (4.16) ,(r.0) =3dscft ' {p['-i,]). (hc boundary condilions an (4 t7) a(0.r=!(30,')-0 The end elurion s Srtn in Il25l, is (4 l8) !(x.r)=&scir'(ptj ],-d'l) This equation epdF ts a singlc soliary The values of lhe Pdanele.s involled Tho spadal intedal t0.301 wilhn= cxperituturion- The @Ufuy of dE *!rc !c or dPlnud? ld velcitv d and $lcded asr d=01003 o15 @d P@ I. -l- ' '=land{=t0 aI=0o5 h slectcd $herc in lcm of P= for numencal aid L noms N d al tl26l Th. lolilary wavc PDPagates rhe spadal inledal t0, 301 duting th€ dme incdtl is [0 801. In ou nume'ical expenmnrs 4 4 dd | ' wc sld de oPtmrl val@ ol tlE shtpe Pusrclds in the neidborhood ol 0 compaEd {ith lhe Es!l$ givei by for MQ, GA antl IMQ. quan i.s for .tur roms 6 shosn in Fig d = O.l and d = by MQ saka 4l The emt nom 003 de listcd in Table 4 m b.ttq lhd n is cleu $al while consideriry l and and Tablc 4 th€ tcsuxs Siven bv Saka 't $Ee conFRed 2 Wc see that thc al (126l Fom !, dd I. , $e Esulb of MQ @ betr.r fin the tables lhai of cA Thc solitary and lMQ for d = 0.1 and CA is 3howing bclt€r performdncc for d = 0 03 motenent of wlve soludons fd drE RBFS m shown in Fig 4 2 Fon this figre $' norion lor a solir,ry watcs sith no choSc in shaF is obsefled Th' lhru invdidls of soliBry wave of anplitude 3d and width dlpeoding on p, (as my be .valulcd Mlldcallv siver ir Il26l) 64 t2d1 4au',E pp Far d =0.1, /, = 1,2, 1r =0.2E8,1r = 0,05?6, Fot d =A.$, I | = 0.36, I1 =0.92592. rr = 0.00156. Fon ln Tablca 4.1 md 4.2 we can scc d. ir thce inldimt! v!ry Sood N nurudcrl valks .nd fi@Etical valws oi agr.crcnt. In Fig. 4.3, qc havc Plolt d ln€ saled tnt Jeobiu mtrix fd diff.|lil mbd of nodes t@ping the lallc of shaF ptrs@ter fi-d. Fd MQ ! slobtc apcctruD b obbircd whil! ir cs oa GA and IMQ rcw ciscnttlB lic in pcidrc hrr pl.nc wry cloF !o am as shoM in Fi8. 4.4. h cigenvalu€s of rh. ou wond crp.rimdt E kep insid€ srabihy cej@ of RK'4 /V lixcd ad chs8c .. ln sh.m 6 furthd incras c fic aF.uun shows tlE vduc of . ce is ol MQ all eig.nvalucs li. lf n clos. !o opliMl valuc. . 6ild .xponcntial trcwtt in posni* half Pld. Sinile b.havid is ob$Feil for IMQ whilc fr GA strul! of shaF paFna.r tullts in stlblc nod.s &d inclts in vrluc of c cNes utr|oble which do.s mr affect $rbilily. valuc nod.s. TIE sLbility rpeEum is shoM in Fi8- 4.5. To obFrc rhe ini.rdtid of t*! slillry !(x,o)=!,+a, u, = 3d,.@ h' I e, \' waws wc coNidd lh. innial condidon (4.re) - x, - d,)), (.mJ ed bolrd4ry condnions N !(0,,)-!(80,')=0. 14.2t) Tbe valles of tlrc pEmelers .=1. Th€ p, =05. Esrl$ @ pr @ tllen s: =0.5,I =10, J: : 25. 4 =1.5 dd 4 =0.75. compured *ith Ar=0.l,Ar=0.1 at timc r=0,15,30, the pronb is |*o waves e moving rowards ngh wnh whitiA thrt &pcnd d drh dSrtrtudet dd fl ccnln slagc tlE ld86 waw caEhes op 0E snaller on. ar|d bofi lhe qavcs unite $d tlFr sepmt frnm dch othr shown in th. Fig. 4 6. !r is cleu from thc figurc thar ihe teir odginll shape. The prccess of energenc. .nd sepdaion lakes Place berween O. trnes r=10 ard r=20. Th. valua of th€ consmcd quaniri€s in nainaining comprism eidt qlanic B.sPline GaL*rn rcthod (QBCM) The rcsul1s 6E obrtned by using MQ TrE dalttcal valEs of rhc |h@ ndill invuidt! basis tunction e srH [ 126] t*ing @ given in Table 4 3. the valuc of tlE shaF bclos bv usitrg fomula givs in It26S t,=rz(d,+a)=zt, r, =2E.E(d,:+ll)=8L Frcm rh. T.blc 43, n is clc& rhat tlE MMLaF rhe very r, =5? 6(li +d:J=2t8.1. nu@nol valEs of tlE @NP.d qu.ntnies bv clos to dalyticd vahes in conpdison with QBGM 1126l. showing that prcsnt shcm. is fairlY conervcd. 10 study undutation the initial condition n @Nid.Ed as wr!. ,(r,o)=o.5t4l / \ I /.-. \\ |!nhl:---:c ll \.,./ &d bolndary conditions (4.22) e (4.2r) vhere ,o is the anpliturL of w.ter the slop€ betwn simulation is abov€ dE equilibrim l€vel at initial tine and d h th€ stitiondy wacr and .lccp.r @ied out lalint vatei !n Eq, (4.21). !0=0,1, Tnc 6=0.1666667,rn =0, At=l md Ar=0.2, oter $c in dal-20<J<50. IIE .onpot tjoo is cuutcd up to dme t=Em and rhe &rclop@ of utduld polil. is shown in fi$. 4.4-4.5. wc E odtiquadric RBF lMQ) with shlpe pamcler .=0.45. For wo diffctnt valEs ot d=5 dd d=2, rhc rcsults for thm iDvtridrs, pcak position .!d uplitude of dE l€ading undulalioD givcn in Table 4.4 and Tdble 4.5. Fom thc Tablc 4,4 md Tablc 4.5, ir is mximum value of amplitude is 0.18J030 for d =2, which slier e alnosr $e sme a sah tl25l, at r cler = 44 8 for d =5 and 0.186?4 ar in Ef.Ence [26]. l. @ thar the r = 45.8 Tables 4.6 and 4.? lhe and Dag and sala [27] are lisred md fmn nutuol conpuien w. sE that our mthod is capable of giving the ene ectmy *iih low spadal r.shtion and a ld8.r tim step, Our me$od Nunerical vdianon in the q@ttia r,, r,, r, e calculacd f|om T$16 4.4-4.5 using fic fomdd EsulE of studies by Saka et .1. [26] given in [126], I ,ld tiM =8..]lJ)- I ,ld tirc =0) Th.* nuftrical vdiatiors $eoEncal valu6 siven iD e shoen in Tabl. 4.E, and [26], ii*=i':=-'"'. !i0 * *:) !.: a, = = o f;i.t*=!':-.o*". aaoouo". e in very good agrene vi0r T.bb 4l: lnvlmB rd .m m for rlE ,ir8L olir.ry wlE QBGMll26lh=0.15 d= 0.1, Trhle h= 0.15,I t= 0.05 {2: Invdios old mr nms for iitrgl. solil,ry ; o0t43l d - o.oi h=0.t5,tt:0.05 wpc T.bL r|.3: hveo6 Id 'nhdior ot tvo slitt, v.€ QBGMIr26l 2'8.7028! ; 26,9998! 2ts.702tx 213.70130 213,69966 269991t Tbbh 44: D.FlopMt of u undula boi xflmt3 3,0(x)$4 iorlr16l o,t76r27 05630t5 !2.N T.hL 'L5: D.wlopnEr ord u.dllr b.c MQ.MOL 0.r8955 o.t9t27 MM t126l 0J3t62 T.bL 3OIr27l a.6: Drv.lopmr 0{,05) 3etr25l (lF4.0n of d ddulr hoa (d=5) ll.6 T.U.,Lt: D.vdo0|Mr of u udd& bd! (&2) 0.723!40 0:tbrD 6,00243 0,tztall T.bh 4E: Nmridl v.rl.!on ln conr.d [email protected] Itn'[email protected]!b{l, t5 chdot o2 0.05 Ilm a* soliey d€ lofils rith .rylibd. ol. Pr.bLo 4,1, rl I i s t d I I IEG,Lt Std.of Lcobilt tdir nt drE s or..da Pr.bld4l 7A I t "l I IlGa.+ Spd. ofr..otie dn tr &* y.!E of d!r. Fr@. Plobld a,l. n|Dar:rddCoottw.lic, sl$lFbd bM a0 c FirrE a6: Uiduldioi pbnhs,r dlllEm op@ 4 It-a?. urdb ,.ld|. d dtftrd riD LEL tae2. 45 Cotrdusbn Nwdcal t chriqE h.$d bEn od msh L.{5 ntthod of ibpi.|Ml.d fd th. !l|@i6l $lulion liB ui.8 Edid tois fidcric ha of Err .qonid. Tbc o€ibod pEenls dE posiiu ald v.l@ity of $ryb 3olitlty ewc. TrE tslis fo tlE itrl.ledo4 of r*o $lir.ty waEs &d slmdr d.rcloDdHt of urd rr b@ otrtrm sdid sludi! Stability !"lysis is p€Ifom.d by cdcuLtinS dE .itlNrlB of dt Jeobi& urir. .Dplisrde, Tlrce iNeidt' pFsenr work. of nodon Eflicidt .r. found to bc numerical cNblr fonll thc cM |. ult! at! obtlircd lad @np@d wolk U24 l2?, l28l avdau€ in thc lit ratu.. d.$dbed in the witn dE publish.d Chspter 5 Method of lines for the solution of modifred equal width (MEW equation using radial bssis f?u.nctions 5,1 Intmduction Thc nodified equal *id$ dhp€sion process has the Au equriion (MEW) Il l9l, dising fbn Lhe nonllnear nedia fom #.*(#)=' whcrc v is a (5.t) pcitivc pan@io ald a few $alytic elurions wilh available fd modificd egnldizd thc sub6c.ipb r ed v denot dificEntiarion. . cuicl.d et of houndary dd Only initial mndilions ft nodifred equal width (MEW) cqoation. Thc MEw equrior is elaled !o long w{vc (MRLW).quation (Abdullcv.t nodilied Kon weg de,Vries (MKDV) cqurior (Cordicr of MEW eqmrion wilh intercsr wi$ 0r vdious foi the @semhe^, E6e. and Kutluay er al. al. [30]. Finite.l.rcnt Nthods based the Il29l). Nunenc.l srudy bouodary and innial conditions has These hdrhods include finite ll28l) md ban the ropic of difieene ne$od given by on collcalio. a.d Galerkin dd l34l bave ben Y, DcEli ll35l usd iesl e$ n€lhod basd nethods usi.8 qladndc, cubic and quintic B4pliies nuftncd study of MEW equatoD. on cou@adon t@h qne to solve Dodificd .qual used for ull, 112, 133, width equalion. In this chrprer, qc solve nodifcd cqurl widrh (MEw) equation by @shl.ss mcthod of liDcs. nshlss @tnod of lin6 (MMOL) for MEW Setio. 5.3 is d.vocd to st biuiy ebsis. Scction 54 contaiN dE .u@rical ln S4tion 5.2 se cquaaon. describe thc exmples pmviding the validation of dE @thod md we €omlude in S6tion 5.4. cr@to 52 l'trcrlpdou of th. 5 dtod Cci.b XBw.q!dd 9*r, dt i-"91{4l=q dx d\u-) "s,s!, wirh tu folowinS ioitirl !(ro)=s(r), ud hdd&t cooditi@ a3r3r, !(., r=s,(r), dr, [email protected] (52) '>o (5.3) r=rr(t. dotr of tlE (5.4) r.ht$ ntu of li'E! s! divid. tlr 4xdd ifl.w.l ta, ,l [4]: t|}iIg { = a .tdt' = r. wc t!. F4a (14) a'd (25) rnd sprdd d.dv.dvcs by ! .in€lc nltir by vcct|r muliPliqliotr into co[ocldor poim! ,Cbc. r, (r)r, (5.5) l.(')r -'. = 8,, (r)., (5.6) (r)- Y l(r),{''! !,. (I)= Y = s, $/ht q(r)=Ylr,rr'!d s-G)=Yl(,)^". Nd couocrdoo of 8q. (t2) tt aclt rd. ++3',,'(s,(r,)o)-'[s.(',)41=0. u) 4 \ U!i68 tlE @lum rldt roLlid, rj , Siv.8 tlE folowiry fdm r.ra...,^,. 1s.7, s.=tr,(1)I-. s-=[s"(r)]*, s.(r,). Is,,(r,)s,,(,,)...s,,(,,)l J,{r,). ;;J,(,,), i = I,2,...,f, ali )s1,. (, ).. ri,,(r,)l )=;;'s,(r I i=1.2,..,x, s,, (r I = s,,, (, lJ,,(J IJL. :,1u . u ;. dt N 1s ,u 11- " f s., I o {l= dt) (r.3) =-rtu, wh€'c F(u) L) 9) =(.I-vs-)'' (3(u1u)*(s,u)). The initial @ndirios e U'-l!"rr,). !"(r,).....,"1r,-,). fon r'rr")l dedbcd boundary conditions r5.ur) in (5.4) 6 ,,(,)=a,(,),,. (,)=a,t) NN clNical fdnt $lv. tlE stsEm (5-9)- *!!1j1{{!1!1 u-' = u. r, r(u').x, = odhr Rugc-Kut|r (RK4) mthod is ued to ) = rIu'+]r, J r, = FIu'++,(, = r(u"+^, (,) ,),,(. dd the approxinar. soludon at fte pfen rethod can dy hlndl. point in th. the incnal nonlinor t r] [a, can m arr, pEent lineariation. TlF f&t h thal at r - 4 a can be touid frcm 'Ir'en rcsr is m.E elemc wisc nultiplicrdoi or vato . th€ b. roud ar qch ti@ in €qn. (5.2) withoul tilen iniial condi rron. 5.3 St{bility Analysis Fof sBbility malysh wc coNidd .qultion (5.9) ltd rc*dle s 35 4= -3p({u,r{s,u))). o<, sr. vlr.E P=(r-vs-)] b a squc mEix vith Thc subility of tlE mthod is of nBl (r.12) u(o) = u kmwD clcmnts. d$u4ed by cdculating ciS.nslud of tF ,sobirn n trix ddd sysGh (5.12) dcli|ld a, a(-rP (u"(s,! a{(!'.(s,u))) ))) (5.13) au AU (5.r4) TIF r, dcIEG ot./r .r. oloc tom [(r1,1r11's,1r,r1) -{ l(,(4)'s, Wc will aDlyz. +>(2!(t!(r)s.(i.r)). rddi.sorrdeEnB (i,t), .igcn!.h* rd mrdi.sorr.hEnrs of tlE ,rco0i6 mEix J iD n.xt s.ction. 5,4 Numedc.l Exrmdcs Th. MEw cmryt cqutid p6s.!g niw tlE following cons. hws rd nasi nom un and {* ll3ll) Bpeiv.ly ,=i*'=i[,'."(*)h'',=!''l^ (5.t5) Th. co|rwatio! of tlcs. iovariant shoes dE stibility of tl|c rul|i.ricd $lEnc. tar $qh lola.rt erY. Dodo Thc inirid co.dirion r(r.o)= f6 dE mrion of sitrr|. rolit ry q!rc i! dq As/r(t(t-r)) *ith dE boondrry dlditioB Th. cncr solirary wrE i! I l3ll (5.16) rJ0 solltid for a l{+-. cqurtio (5.2) (si6 in I 13l l) ,{r4-,1si(r(!-e,-r)) whcE t=*, l is thc mpliardc aid p k thc wkrity of th. sintlc elit ry s.v. h or rhe ou sultr wnn order to €onpde paliftres invotv.d ja s: ! dlicr = l, A = 0.25, P qort. w. Publislul =+,& = 30 dd clrc the vatM spadal donaiD La. bl = = 0.1. Th. sinulrtion is run for 0<,<20 *ith tine step & =0.05. The rcsulls obt in.d by n6hlc$ mthod of li.es using MQ, GA ud IMQ Ddial bals tunctons m @nP.cd wi$ th. tesllrs PEse.i in ll29l, ll30l. Il3ll md IO, 801 I win scp r l32l- In Table 5.l. tlE cdor noms, l3l I md 1321. t $M invlnan6 of mlion al t = 20 de shown Fom fie results by ncshlcss nethod of lines de betle. than [l30l. we navc aho conpdd our Flults qith finitc ehieir se$od [ 133] ror Table 5.1 we can t d se thar lnplillde 3 showi ii thrce differenr vallcs of $e rale with MQ at time ,=20 is Tablc 5 2. Single soliDry diff@nt amphudes in compeison with the cxacl soldion using shown in FiC, 5,2. Th. optitul voluc of shtF paraml.. h sel4t€d tom the inrenah (0.1, l),(2.5,8)ed(0.1, l)fotMQ,GAandIMQEsFctivelyasshownioFis 5.l The saled eigenvalu€s of lhc J&obian dd IMQ. Fi^t wc nodes mtdx lir . (sh.F p8!trct.r) dd (x =200,400 (5. 14) Epc.t e inv.stiSrted using MQ. CA .xperim and 800). Fom thq Fi8. 5,4{i), we c{ w for diffeEnt nnnber or thlr all eagcnvalucs wirhin sbbility Egion of RK4 $hcDa In lhc scond cas we keP /v fired value of. spetrun The is i.sid. the int€Rsls co .iniry oPlimll val@s. ln this obt ined ddrric , _'" ii as shown valu.s or lhe 0.785J98,/ chmee again a stablc Fig. 5.4(iD). itrvdidts givcn in I l30l: =:::- ::::=0 IE vslues of dese invsiants aod N lrso (e dd lic 16666?./r = =0.00520811. @din ddost consta dung all ii lery Sood aCr4ment conPurer siDllallots wilh tlre abovc analylical varues. Tne rate ot onvergence jn space is calculotcd using thc bc," (1,.-., - uq :^, l/ 1,.-. - ua, | iomtla deflned etrlier as: ) lo8, (At, /Ai,r ) whe@ aa r€pEsents exacl solution Ar, as sprtiat stet size. Ar w. =0.05 fi xed and varyins md Uq tPescnts lhc numerical sohrions Pitb cdculaE spatial taL of convcryencc by lcepinS line step sparial st p (d r = 0.8. 0 4. 0 2, 0 l) Frcm $e Table 5 3 w€ qn s.c thar thc mnvdge@ nls dect!&ss conctg.nce I 5.4.2 tu shown in witl the snDller spdtial stcP Fi8 53 e.rctloD of aro solrl.Y e.ve Tnc initiol conditio. for intetelion of two positive solit ry waves ,( siz. Ratt of d denned in .. domain lor tnis proucm is up E*i and Kurtuy I t30l wc tate 0<r<80 wnh l=0-2 ald tlE lo , =@ with tine st P qnh amplitude I stac of edergence at l = 35 thc vslues of th. &=oz At t =0 thc la4er {ave dd w!rc is ot !m pdition r= 15 is thcn sePar.t hom each oth€r and nove foMard with m io Fis.5.5. The daltrcal valus ot rhtc invsiur.s @ giv.n il E.Fi l, = =4?r2388e,I? = r, Tl'c splnal siduladon is run tor tnc sanc ahaF as b€Ioe collision, Th. Phcno@oa of int tution '(A,+A,) is: t( posirion ,=30 wilh anPlitude 0.5 As ne noves fasre! rnd it catch!6 the smaller wave Bolh wav.s ae in and (ne small€r p6ses rh. lsger wave l0l '51?) pu!ftl.R s v=1,&=0.5,r,=05,i,=15.1=3O /4!=| ar'd,4,=0.5 =0 I =>4 *c,r(rr lr-r,ll \.0) ln ordci ro comFrc ou Bults with , I j(4 Sraphic6llv shown and Kurluv Il sol * 4 )= 0.3333331 =i(4*4)=r.a16666?. FrcE the Tabtc 5.4 we position a.d amplitude s olr Gsuln ae v€ry cl6e lo dalytic values Tte or both wavs ar tirc r=60 (when rhe process inteEctio. is ca dtat conplete) 6c also shown in T.ble 5,5. 5.4J lrtffacdotr of tb@ soltlst wrv6 Th. itritial condition rd intemi@ ot l{o p6irjv. solit ry wavs give. in I l34l h /(r.o)=t,4,qr{*,1 r-r, ln o'der to compuc p=1, t=i-, ou EsuLs with D.cli ll35l, {e ta& r,=15, r:=30,4:45, r{=1, rongc fot this Pobl€n is r=0 (5.r8) l) valu6 or rh. !r=0.5 dd 4=0.25 r:0,35 ard rhe sinulation Ar=0,1. Al r=0 rhe ldgesl wave 0<J<140 wilh up to r=200 wirh time slep tne A r - 15 {/Xh fr. .nd to[d] rspliad. I .td 6. |0rlln mw i! rt Ftttb! , 30 vi6 qlitudc 05 .dlbl rlrt ! d Fddo t'a5 rft EC&d.05 T' tn'rtid " ft!' tr Er56 wtva .t dt6nd tio. Ln{. i lox! Ttu dlbdcd rdu.{ ofth!.lrYri.lt t\= tt r, = ivr[ in tr3al +L+rr>- ss?Et - !(,f+,ti +.lj)=as ltD =t(,i'+,{+ri)-Lan& r. co tc lh o rudb rl! trry clost b 'ldrlic v'\rGr ts ompd to 11351. It lodn d ddibd! d tvo ffi a ilD t=2m (dd From lt oqldlt TrtL tt. 5.5 F6 of l6rcdd) G rbar i!tlbL 5_?' !l(l T.tL 5J: hvuiuls dd .ru mru fn tinSL $Ii.ry saw st I = 20 t- Irrol trrll 0r?t MA IMA Ir3ll Il]rl T.ble 5.2r lnvdimts od Ir34l Ir34l MA 11341 .d nom ior sind. elibry d difree.t vdue ot mplitda A T.bL 5J: Sp.nd @ ofsv.B.N for singl. $lioty w!v. l. MA 5.t62t r71315 IMA 0<r<E0,/4=0.25!rr= l0 Lhre 5.4: lnwiors fq two $li.rry s.w jncdiont !.!n413 l.ttt2cx Il3ll T!Dl. 5J, MA IMA Posihon dd anplnld. of No w.vB .t t = 60 LtL s.6l I'|wi:n6 for ||ic $ltury sv.t hldqld tr36l Dra tr36l Trble 5J: Posnion {d mplitudd of $$. *iv.s .t t = 200 MA Il36l 54,25 tr361 IMQ I'51 93 tl.Gl'r:S||FF||6tfugu. FICG5:: siq|.DliE rE *tl| dltod rdiun . { r20 (S.l'd liD .a.lulibdq R.ta! tunql"rl dut r!, it ed Ctu Ers 5 I ItN l3: sr.rid @ oror!l@. c ."-l "" q ! I -{ i g IhE q0. sdinry {cfun h. sio* oliEy d r20i(.) MQ,O) oA (c) IMQ uE ."{ I I I t .{ tl-r$S*Erbri$rd. r&rwr|4(4XlOloAG)DaI 95 cloptet 5 1020siaoa) It@ 5.! LErcioo.a re.ritty ma 'ith dtu i.4liuda ntln It hcrutlotr rirh ot tlm olirty vrss dftEd MplifitLo. 5.5 Co[.bdotr t ta !.d fd uraL.l .oldi. d uBw otdd. Tb D.tod Crs U|}ry ..@r tlrl! 'itar IEirrdod of dE lollD.r !.n ilvolvld in & lGW .$rie llE D.d.o of siigb $lirlry wrv!, i!@lio of |w .rd tbF .liEy r.v.. |.d iltlldt of maid e itirdrd.EtitutlEbrurt3ffr.dilcqdnerhlt FDlit d{tat 1Bo, l3r, l3e l33, l3a d latl.ydH! i! lb liElc b rti. ctrF, o.tnod d lilr !ad.d dh nditl t dt fto.dG Chspter 6 Meshless method of lines for solving nonlineor Schriidiryer equotion 6.1 Intmduction The noolin d Sctu6din8.r (NLS) equaion ,! -a;,'v', vh€E i=Jr-, a a = ". -* < I<-,, @l con$lnt md als known d cubic S.hro<ling.r eqnal'on (6 >0. a(rr) h a t) conPler valu.d runction which govens weally nonlinw, strongly dispesiYe and almst nonochrcnaric wave 1136] Ixn nonlinw oanial difleential .quatioN tPDEs) aPFe in a wide ldietv oI applicauons (s l3El). In tanicul&, ir Provid?t a Mthemtiql mod'l ror pltnorcna i' fte tll7, of water *dv.s, nonlire( optics, hvdrcnagnetrc and plasma wates' prcpagation of hear pulss in solids md of Nilhtory wav6 in Piez@l4tric evoludon eniconilucros tl39l. Thc a$ltdc slution for initiat !(r'0) ranishs foi sufticie.tly lrtg. I I I ws silcn bv zath@! and of NtJ thar '4!anon Shlbar I l54i in l9? I using ioresc saltering D.lhod Fot oo@ Scn tal inidal conditions lheottical solutions or ehh rhis equtio. m nol knovn. TneefoF nanv studi6 in orde! to d'al 'umericar TheF conditions solurion of Nt-S hare bc€n m.le fo diffcEnt bound!ry-initial the rechniques include rinire elenenr nethod (FEM) nnnc diffcFne mrhod (FDM)' sp€.ral rcthods ed collaation rerhod sin8 nditr basis functiotu I 140_ 1531 In order to rind rhe oum€rical solurion of Eq. (61) Ne considct a bonnded inlcryil or rcal liN with aninci'l bound{v condnionlh ! condidons !(a'4=r{t't)=0 to hodcl 0E phvsical J0 6 r+l@. kl (6.2) u(at)=v(.'ct)+ir(at) qkrc (I,4 ed ,('') dF following coupled ", ="- arc rcalfunctions. By sub6tiruins Eq (6.2)inFa (6.1)'wecet s'6cn ot PDES +a("'+"')", (6.3) \= "_-q(v'+*,).. The bounddy condtnons e rk.r)=v(a.,)=0, p(r,')= v(r.r=0. (6.4) This ch{p&r is org.nizcd is thtce ections. ln sation 62 rhe explained and [4,'] sdon 6.3 @ntaim thc numncal €xmpl6 fo. nftncd eheft the rs justific{tion or th. Delhod and wc mnclude in eclion 6 4 6.2 Nudertcal Scbeoe o( nunocd rchc@ *e lint divi.lc sPatial domain la bl iiro Out ofrhce Fints {,i=2,3 , N_l arc i'knor points while J For implemmarior of nodes and \ 4.i=1,2,..,,N. .E thc bounddy Points The aPprcxim& turction of v(r) dd w(r)is deoor.d by @l dd indginarv pans v'(t) ed,'(,) r,r =>1 (,)r,, (4= v'(').r "^G) (6.5) it (x)->tovlG)=v, (,)l (6.6) whd ,l'r.nd t/s G udoown tiift &Pcdlnt qoaniti* a l v/lotu ndid b4is nEdd lrd Y'(r) is tlE s.e s &6i.d in PFvior cb4r6 w. harc u!.d MQ, cA ed IMQ .!di.l b6is tuDdc il @ nuEicd cx.DPLs. tar v"(a)=v, and / ({ )= a, rhcn in narix-v@lor notatio. (6.7) r=[r, (r),,r" ('),..., &t)]" '=r,,(r),,:(,),...,v,(,)f r =[r, (r), z,(r), -, r"(r)]', i=trt).'c(r),....*"t)t'. , ,/,1,,1v,(\) Y'(,') Y'G,) - e, \',) v,l v.l4 - v,Q') v,?N) Frm 86. (6.5), (6.6) ard (6.?) it foXoe! dDt "(')=Y' ' v',(a) - vtl'NJ (,)l-'r ' ( r ) = Y I ( r ) ,{ - ' | wh@ P {!)- Y'{r),{ '. (r)r, (6.8) =P(r)', (6.e) =P Apptri.s Eq. (6.D on Eq. (6.9) lo @upLd sy.rd! of (P,,(r,)v)+e(,i + PDE$ (6 3), dld @lla!in8 d v,:)', (6.10) t=,.r.r....,"1 -(P,,(r,)r)- {(ri +,1)r,, s,tw ',O=', P.(r) = ed 4()= 4 [4,c,)P,,(r,).-P&(,,)] P"l',\= *Pto,i, i =1,2....,N r,,c,) = [1i,, P t,,(',) Using = (r,) P ;i v,... v,1 - t(, ,r, th. follo*itrg v =lv P, @llm ( r, )... P,,, (r, i -t,2,. ,N lcctoB mt lid ]l for lbow 3y5Ln' lct . w=[',,,-,,]' [P" (', )],., . P- = [P. (,,)],., , r- = [r- (r )],", ' P- = [r* (,,)],,. . P. = tlE[ Eq. (6.10) s b. wird a follos: ro! J*=,t,,t,-0,"'t*',t lf,= -r"..* r- rtv " I (6.tD w 'rw . I 4IL= nrv.w,. 4!= c{v.w'. dt (6.r2) dl ls {v,lY )=(P,,r)+4(v'+w')v lc (v.l{ )= -(P,,w )- 4(v'+ w')w, &d injn.l ooditic tq dE cdplcd sysEo w('")=[,"(a), w"(l),..., r,"(r".), vp";=f"rar. "'rar...., I J @ "rG,)]', ,'r'"-,r.,"c,rf. Flom the bound{y condiliou &s.rib.d it Eq. (6.4) oi Iv, (,)= 0,vr (r)= 0 I I,, (,) = o. "- (, ) = Noe EqN. (6.10).(6.14) e (6 13) (6. solrtd by RK4 *lm. r4) w., w.+a,((,,+2lxn+x,rl+4,.) .," dr{x2L + 2(x): f:: r+ lj.) (v,w). H .E + I v'+1r, ,w (6.tt o[u ).,'= 'tr lv'*!r...w 2| )",,= 'fi =c(v" 2" nIu 2 6.3 Numerical T€$ Problems mndot.d schctu for nu@ncal eluiion of noolaNe In fiis sation oe apply lhe above schrodaneer noncntun (Nt-s) equation- Thc m olcllatcd by using losrl rwo coiefled qwdtid of m6s dd ttc f@ulls dcnncd in I l4?l as: t)* The eol,lic slurioi of nd in6 (6. Schrodingq .qudion {6. | ) is givcn by /{,,)=dli}'l.xp'[;pr-r(e'-a'Yl]sa1'a(t-p4 This equauon rcF6enls singlc pl'J'ljmErs is d slhon of mplilude a chosn fbm lilcnr\rc -2o < r < m. For a = s. q I enleloF rcliton =2. p = 4, and sPe.d / i6 'ftc v.lues of lhe is Siven by, I noving lowdds nght with mmtrnr speed 4, For our nunc.ical cal.ula ons w. usc r = 0.25, 0.3125 and ed 17) d = 1.2 Nd sPatial domain wbich EpEenls notion of a sinlle soliwy wave of anphudc us r6) rin. lMQ, md oflimal valuc of shap€ PdaDerd for ttrce RBFS is seteied from lh. int ryals {0.t - 2.5), (l - l0) and i0.5 ' 3.5) t$pecuv.ly srepAr = 0.0q25. we MQ, GA s showD i! Fi&6.1. In Tlblc 6.1 .nor mm rnd |9o coNmd qodtili.s.l r=l .nd fd a = 1,2 e dF$r ln T.bL 6.2. w. haw slD*n dr Esulli comFrcd by ftsNN Iie .p?@h &d *irh rhos. plsnr in rlE licnort 1142, 144, l5ll. wc can $. rhfl dE p|!s.!t ncthod F o.N vcry *.ll with low sprirl F$lulion ..d bc|rer rssutu e obtlirEd 3 @Dpad with dlid r|ctlods. TIE lDlfic vtlB ol i.vdiett e: r, =2 ad li =?.33!333 fora=l. Frcm T.blc 6.1 ttd TrbL 6.2 it is clear tht @rl|od or slB inEidt! @ Ery d@ to d.ltiic iolution .nd tl.rivc chd8. with rlspet ro initi.l ritu is wry slDll by dr Bhles nEllDd. ln Fis.6.3 ql lnd nutudcd of lwo inryin.ry c@poF i $d trrduLs of rrrclling loliioo Thc point wie rllc of c@!e€c@ in sp.e rd tirc ! sp.tial lal. of = A .m I is cdcqlct\?d by Ep6cn$.x*t iolution.rd Ur fll U! aFrscnB th. Fipccrivcly wilh sp.tal sap (l' 3hom. .Id nm st.p mwryeM E kep dm 0.5, O25) . Tltc $ing O. a"(1,- -u.l/1,- 4.(1,- -u,lil,- -u* l) wh.E e siu ,r slcp A, romlla: -u*,1) nunErical slltioN t .id Ar,. ln orLr !o c.lcolat =0.qt5 fircd td vory spec scp odcr of coiv.rtEE is shoM in T!bl. 6.3. By dlcrasinS ,, thlt RBFS b shown in Fig.6.2 by ploonS tlciproc.l otx(tumbqorcouoca{o poinB) !.n6 nqi!'M.b$l!t .mr. No* td tir nc of conrrgcrcc c.lcuLlion vG l.cp ,=0.25 {ixcd dd vlry tim stcp (& =0.@5,0.0025, o0l2i 0'0@625). Fmn Tlblc 6.4 w s tEr tlE Frc or is Educcd. Thc @rcrylncc in tim consltdrc. b.hlvid of &o!!g *ith smrlld tim scp. PmbLn 632 ln 6b plo6l.n ir|.r.dioo ,(,.4 = i",{r' of t$o slhdre is sNdicd by {-p,(+p, {,-,,)[*,'", (,-,,) lsiry dE initial @di[on (6lE) 105 In odct !o coDpdc our rcsults with lbose p€s.nt iD the literalut we ch@se lhe valucs oi thc pdancten 6r q=2.at=l,A=4.\=-10.d1=lpr= _4 and&=l0 The -20<r<20 vith n=0.2, ar = 0 005 tn Table 6.5 dd Table 6,6 the conercd quantitics ad their elarive chdges comput.d by MQ. GA. tMQ ed ate by thc mthods pacnt in [14], 150, dd l5l) aE 3hown nuncdcal rsulls m conputed over thc domain Two solitons of equal mplitude tnvelting in oPPo$b dnedon couid. then $Paratc fom crch olnd pNfring their qidnd sh.F. tlc whole p|@s of inte.4tion is shoM in Fig.6.4, whlch is iI ag@Enl{ith lh! bchaviorshow. by rn aulhoEin[14], 150. 15l, l. ed 1521 bodr rhd.rically and nuncncaly. rhis example, we show the binh of soli|on fot 89. (6.1) which vas itvestiSated by Delfour er al. t14?l using a sqwe wcll initial condinor C.rdner er al. [144] als studied lh. eme plEmnenoo using lh. Mdw.Uian initial condnion !(r.o)=,{qp( ;). (6.re) ln N nurcdol cal€ulations wc sld tlrc valu.s of the para@te6 6: /'=0.25.dr=0.005 md s=2 with comPutllional domb '45<r<45. Accoding lo J lhcn a solibn n genent ,{ = I < t, o$eNie soliton d4avs awav Fo! binh or a sllnding slilon decaying aw.y wilh ti@ is sho*n jn Fis G while forA dforavaluegt atd $ai - qhich is in 1.78 > tt- lgEEnt 1.7725. a shnding solrton of amphtode 2 wnn ficorr. Th. t =A'J:, t,=+A'lr,!z cA')G . 's )hoM in | Mlytic valks of No Invemb for '8 6 5, 66 initial Nunericrl valucs of l*o invair s e in v€ry good a8t4runt with ard)li€ vdua a shown in Table 6.? and Tabl. 6.8- lo econd sinutation wc afudy binh of mobil. ! (r.0) = $lito! usiry lh€ initial condition A.rp(-i + 2t). Thc spodal domin for thk cxF ircnt k -305r<@ litiry Thc situladon b run up to tin. | - A = 1.78 for q. s .nd dr =0-005. 6. A hobilc soliton of @p|nu.L 2 is dcv.loP.d for shown in Fi8.6.?. Nururical (6.I ) $c folloeing /' =0 25 vrlus ol invEirnts @ 3hown in Tablc 6.9, initi.l cdtditid (6.U0) !t!t prcduce ab@nd q=24'. 4=12 elhm if of P . fi. droctic.l lolutio is troM (s.e Mil.s tl55l). The slltion is.o{ Bble it 4 >3. w. F.fom M NMicd sibulatron usiig MQ ndial basis tuictioi in thc tu8c -m<r<m bling /'=0.0625,&=0.001. TlEc ard t@r Fo. { bound sote of solibns at dly d =3 and t =4 epccdwly ri@s of sihuhdon. TIE plo$ ol nEdul6 of elution d shown in Fig.6.8 aid bound solilons of snall nmw sElcruEs dcv.loFd fo. 4 TlEs 6ulll e dc s.m 3 dcvdopd by Clder {1451 ald KottDaz I l52l-TlE datlic v.h6 of inv&iurs for initid coidition {6.20) e 6 fouoss: fi9.6.9. Thc sl!6 of boii invdimL @ v.ry closc !o rh.@rical Emin coened 6ult! a showtr i. by our hdhod for 4 = TabL 6.10. 18 and 4 er al. = 32 , drd LtL Cl: Si!8h $lit . .t t = I t MQ 1,ttt33 t.33333 IMQ |MA TrhL al CoIp6ri$ oldld. slitor !r t= | ehh @!, Fe* in lildr@ i.:riiixid irQ oll2J IMQ 0.1125 cDQtl'l 0.3125 -t2136&to, l -5,a9l2oxlo' {,flm3 109 T.bk 6l Rr( or converg.nc. 'n ,p6e rtr sin8l. eliM dt=I (Ma) 15.7139 0Mq) T.bL 6a: (MQ) OMQ) Ro& of ony.r8.M in 6ft fq singh eliton for N = 160. T.bh 6& Consrv.d qdi.i.s fq t*o $lnos :'.:') 3199999 T.bl. 66: Consncd qMiiri.s ftr rwo slihs .l t = 2 5 .5.41r3x10' -:65so'r0' tMO lMA .r.ol3&4^ro" cDQt1521 T.bL 6.7: Bini of EEdinS l.576rb/roi $lib i-I l.25llll o367rir6 r.2J3314 0.57157 r 253ll. 0,167037 0167174 l.25r]|,r 0.36?lao 0.167142 1.253314 0.36?109 0.167r{15 Trbl. 6J: Binh of ibdins sohor ,Ma Dra b.ir'3.e'rrdb I ro 39rmo 2.0 39?rdl0 .ae2t6it 3ti0o0 l-aert6r7 4.9256,17 r 9710m 4,925621 -4.925656 3.9?10@ -4.925625 I r.0 t.971000 -{ 92J65r l.97rrm -4.925562 3.97100) ' .{.9256?3 -4 95549 t.o t.97tw1 49u1L5 l..tttu2 49uqn 6.0 1.97r0r3 4921521 t 19@93 j.gt otl L .,::::: --:':::: :::: A= 1,73, Anall4 cv.lues:4= 3.97100, /:=-4 92562. T.bL 6.9: Biih of nEbrle sli@, A MA T.t'L t l;+--t 6.10: Bound sbE of : | 73. --'i i elution I iia .1r.113333 1{ImD r!2 tl|E6.r:[email protected]. 113 ito 0,5 o1 oo15 o,@ o.@5 Ilcal! Ra 06 0G6 0.0,t o0a5 0.o5 (t65 ordvd3@ i!:pc. 1L It|Er 6l! li!C, O) &116 ibl n-1, (r) i= lr(.) lr2r (rt) r... t- 0; It6 6lr llbdbr of ts.obts, 115 tgG 6J: Sddji8 6Uh , A=I tL1 t|ra.4 rdt-bq A- l'r fla ,l.r a': ||.rab.ra A= l.7l. 119 FlsBt d& Bo&ul.[b ol&lud4 q = l3 121) choEtE t rpd 6J: B.md N ol bl{id. q - !2 121 6,4 CoDclusiotr In this wo* we have appli.d G!hl.$ nonline$ Schr6dingd .quatiotr. Fou mcthod qT.! of of lir6 ror numdol slutiotr of Prcblens includiry Doton of a sin8l. lolitlly wav.s, Max*elie inilirl @ndition and bound !t!td of solitoN have hen studicd. Th. nuitric.l 6ult! fc @r |ll]m .nd inwi{tg of DotioD e .nom in @npdio eith drtid drnods .EilabL i! rh. lLiilr .Id it h fould ltnt tlc Fcent rlhod tiB !u! .cqrr!r. Bdts i! tI truEicd siEultlions. solitary wav.. inlcnclion of two r.l! of cdvergEc in stc &d tn it ds d.eodnE4 In dl NtEiical cxFdrcna N mdEd F.fot$ vcry s.ll. TIE 122 Chqpter 7 Future Scenario Th. resedch pr.senEd in using radial bsb the disenaion is f@used on rhe use furctions to d.velop nututical schem€s lir.d dm dcFndclr partial difreEntiol equadons have prcduced encouFeing Nme.ical resh td l€ss mcthod of the soluion oI lin€s no.lirc& iiv.$i8rdotu cdicd out al$ iniriarcd a tumb.t of oFn (PDES). The Bults but have reseeh prcblcN. The ue of globdlly supponed RBFS Esults in dcnsc coefficienl ill condilioned wheo nunber of coU@ation points n increa*d and rn marix becones n.slvablc if nlmbet of poifts exceds 1000 lhis difficuhv hiod.s thc ue of RBF n€thods for ld8c eale ptoblems Another probbm n rhat of matrix which becomes sbbiliry whilc usinB explicil tine intcgaion $henes These difficultjca can b€ avoided by lollowing appr@chcs: .:. De.ohposing domdn in Io smallfi subdo@ins insrcad of glob.l lbtge donaan. Only very few tesulrs de av.ilable fo. couPling donain deconposnion lechniquo wi$ RBF nethods. A very useful work an I can bc done by extendine lhis ide! 1o ecl insigh inro rhis technique s4 Dubal 11561and BeaBonll5?l Ue of inplicil scheGs for timc iilegotion in od{ to orercom. lh. problem of References ul A. R. Mirchcn, D.F. cdfrirh (1980), The Fini& DitreEM Mcrhod in Panial Fqnarions. lohn WikJ & Son. I2l L. D.mkowicz, J. T. Odcn, w, R&bovicz ud O. Hardy (1989), Tovard a univeNbl h p ad.ptive finire ele@nt straleCy, Pan l: consutined appoximtion rc. ^dd^r^sruc []l w. Racho*icz, a&Fivc finnc J. T. @en 6nd ebMt s[ M.ch. Enere,77. [4] H Power, CMp& M.th,Appl. Mech. Enery,71,W.19-112. L Dcntwicz (1989). Tosard a oniv€Ml lrp .8y. IIL Dcsig of h-p m*h.s C@pr,, M.t . Appt w. r8t-212. V. Bd@ and lJniv.sita D.gli Studi di Paledo conpdson Analysis b€tween (2002). A UnrynncFic md Symeric Radial B$is function Collocation Mcthods for thc Nu@ical Solulion oI Panid Dificrcntial EquatioB. CMW| [5] Math. Appl,43, W. 551543. A. Zaf@llah (1970). Applicdion ol the ftthod of lircs ro pebolic pdtial difcEndal equtions wirh e@t estu.G../. ACM, l? (2), [6] M. B. Caner aod H. w. Hiids (1978), cquation. .tmllart r,3l .Schi6s (l9l), [?] w. I8l R. PEgla aid E w. Peltl (2), fte pp. 29,L302. nethod of lincs aod thc adyecrion pp.5H9. The Numeri.al Methdl of (19E9, ft. und lA.adeni. PNs, Su tuthod or lifts, ii T. Irolr (.d.), Nuftncal T.chniques for Micrcqave and MiUimtcrNave Passiv. Suucrnt s. John Wiler, N.w York, W, 381446, I9l R. PEgl! (1986), Analysis or planar dicrcwave sducor$ on naSrEriz€d reniE tubs@c. An It Ll.lL Ub.nrue@e,40. W 21O 27 3. trol R. Pregh (r98?). About rhc natuc of the ne6od of liBs. U b. n rc8 u n s, 4 | (6), Illl tl2l R A. ELeL W. 36v37o M- N. O. Sadiku and C. N. Obiozd {2000) A sinPle ol li..s. Int. J. Arh Eltt. En|. Eat.3l iibdrcion to thc rrhod (7), PP 2E2'2%. CinSold and J. J. Monashan (197?), smoorhed particle hvdrcdvntnicsl theory and .pplicarion to Altrnonical S@ary, lal, pp. 375'3E9 tlll L B. re!, A n@dctl non-sPherical aPprech !o Oe t stts Monthl! Notice! Roral sriog of tusion V@$. Atrotutcol J ou tltol, 48. pp. 1013-1024. I14l R. L. Hrdy (1971), Multiquadric Equations of Topography and otnd lft8uld Surtaces.I 6€"pn$, A.t, ?6, pp. 1905-1915 ll5l E. J. KlNa .pplicatio.s (1990). Multiqu!.lrica-a scauercd d1ta apPrcximation scheme ro €onpuiational ntin dyn nics-l Conp - Mdth. *ith ApPl l9 $19) p9 121-t45. tl6l E. J. Klns (l9m), Muliqu!&ic$a $andEd .Lta aPPrcximo(o applicarions !o comPuladonal sheme sitn flrid dyuni.s rr' Ctuput Moth APpl.19 (Elg), pP.l4?-161. Il?l nuftncal ftrhod ro' n a' rrMsret radi.l b4is lurcriotr /d ./' tsu.| Meth Ensq M. Z.doukat, H. Powet, c. s. Cnen ( 1998), A poblcbs uring €ollMlion and 42,W.1263-1214. tlsl E. trFson and B- Fombcry (2003). A nuu@cal sndv or soDc rldial bsis function base<l solution 891-902. n thods for clliptic PDES Cotpd M4th Apll 46 pP I l9l Li. Jichun, A. H. D. cheng and .C S chc! A conp&isn @nErgcnce of multiquadric coll@lno! @rhod and Ele[, Anal. Bolnd. t20l c. A 38, pp. 18l-2(x). ol eleEd d'b: di$ame mriss fomnons CNn Apprcr'2'Pp ll-?2' Micchcui (1986), lnt ryolaion EnSiiceB, 2d Ed, ISBN 0-8ll6-4323-Q' Bnkhaui.r Eotton (192), The $@ry of Bdid bais fun tion aPpoximtion m 1990, in: Will Light (Ed.), Warclcts. subdilision alSornhm ed Edial basis turcrioos. Orlo Uniw5irr, otfot|,2. PP to5'203 i. D. M. PowcU R. Madych dd S A Nclson (l9o) Multveiatc i eaolation ondiioDlfy pcirivc definire tu$tj6'J, Math CtuP.54'W.?ll-30 t24l w. t25l md Lokenath Dcbnarn (2005), Nonlines P.nial DiffeEndal EqlarioN for Sccntisrs and I2ll mcthod- Eng 27, pp.251-251. coldiljoElly positive definik t22l linirc.lcrcnr R. Fra.ke (1982), scane8t .tata intctlolalion: test of some mc$ods Ma'ft Carpr, t2ll oi efficiencv ud edot R, Schab&k, Emr esliDats and condirioD and nmb.A for odial basn tunction i.reryo].'rion. Adv. Conput. Math. ), pP 25l'264 126l z. wu (1998). SolvinS PDE vith tadial b.sis function a.d thc cde esrinaton ln: Z Chen. Y. Li, C. A. Macchelli. aii t2?l c. opp l ie d Y Xu. M Dctlet Edaro6 &c,ut tut's d Put' tu the wt i. t, 2O2, C @ 9&ou. FEnre, R. Schabact (1998), Coiveteence odes of tushless coll@aton ne$ods using ndialbdis tunclols Adv ConPut Moth'a.PP.381'399 I28l Matherutic! (t99), ss Gal.rtin mthods of conputation, 6a, pp l52l_l5lI H. w.ndland Mesl using tsdid batis rurclons l29l M. A. colb.rg and C .S. Chen (t9??), Dhcrcte prej4tio. nethods for inregRl .qtario.s. Conpu| Mech, Publ. Botton, MA. [30] M. A. colberg and c. s. Chen ( 1994), The th.o.y of ndal bdis futctions aPplied to l|le BEM for inlFmcpn.s panial diff.rrtial eA@tioB- Bound Et n CM,5,W.5I4l. ISll B, Fodb.rg and T- A. tlar t32l Drisll (2m2). Interpoladoi in the linit of imE6ingly ndi.l b6is tuncti@s. Cd'p! t. M.th. Y.c. Hon and z 43, pp 4lA22 M. wu (2000), A qusi-itrt tpolalion mrhod lot $lvins stiff ordinaryditr@htial cquations. /rt,lJ, I33l Ad, NwL Methodt EnBi4a,W llal ll97 W, Cho. md Y.C. Hon (2003), NuDerical convctgence of boundatv knol melhod in the ,Mlysis of Helnhollz, nodifi.d Helt$oltz, ed convection niffusion ptobl,6. Conp6. Mc'h. Appl. M.cit, ,"8.8r 192, pp. lE59 l8?5 l34l w. Chen sd M. A. Tae*a (2002), A nBhles, i egnn@ r4e o'rly RBF rechniqE. CanP!. Ma tL AWl,43, W. 319 [35] c. s- chen, C. A. BEbbia 6d fo' Herrnholtz 136l ryPe oF ato6. t, Elcn, 20, PP. boundary_ i9r H. Powcr (1998). Dual und. dd riPciry nethod us'n8 495 5M G. E. Fashauer (199?), Solvingpanial diffcrcnlill e{uations bv collcation wnh ndial basis tunctions, Io Sutfa.c Fitri.g and Muld_esohrion Methods, Mchaulc AL. Rabur C, S.hunaks LL Gd6l. Vanderbilt Unitersiry Pressi USA' Pp 13l 138. I3?l Y. C. Hon iridd t38l ed X. Z. Mao (199?), A mulriquaddc inte.polation val@ prcblens. Sci C@p!r, 12, pp ftlhod fot 5l'5. A..fficicd nurcricnl schetu rc Burs.n AppL MdttL Con pa,95,l9a. W 37-50 Y. C. Eon dd X. z. M&. solvinS equaron t39l Y. C. non, K. F. ChcunS, X. Z. Mao and E J r@s! (199), A muliquadlc elution for shfllow {at r equ.lion. AscE J Htd4ulie Engtg l25 Pp 52'L533 {4ol Y. C. Hon dd x. Z. opriohs pricin8 I4ll Y. c. Mao (1999), A 6dirl model,, Fr'en., Er8r8,8, Hon and x. aou (2m0), A PP. basis iuiction m$od lor $lving 3l-49. companson on usins larious ddial basis funclions for options pricinS, /t t, J. AppL Sci Comput, l , W 2941 [42] a I43j ud A. J. Khaltat s. Islm (2008), A conpeativ. sludy of olnerical solut ons or cl6s of Kdv equiions. Appl Math, Conput,99, W. 425434 S. H.q, s. klan and M. Uddin (2009), A msh{@ ftthod ror the nu@ricar soluton of the Kdv-BurScr cqution Awl. Mah. Mod.l, 33, I,t4l W 3442-3449 s. lstm, s. Haq at|d M, uddin (2008), A msh fec i.Lrpoladon rethod lor rhc nnrcncal $lution of tlE c@plcd nonliicd pafiial difcEntial Equations. Ez8 AMt- Bourn Elen, 33, w. t45l Y. Dercli, I- Bdirl t46l basis Drsui dd 39HW. fumdotu. Crraor, M. Dchghd dd A. rd Nt.s eqution sins 5,/noa a'd Ftctals,42,ltp. l2!l-123) l. D.e e009), Sohon solutiotu Shokti (2@7), A numdical @thod for rwoiimensional SchrodinSer equanon usint coll@etron and mdia! bash funcdons. Appl. 54, [4?] w. 36-146. M. Dehghai and A. Shokri (2006), A nlmtical mcthod rot solution or thc No_ dinensional sine-Gordon equrtion usin8 Sinll. t48l CMpu| Moth 19, I'p. t. ndial basis funcrions Math Cmput 1W-115. Arshed Ali (2009), M€sh f@ coU@ation mcthod fo. num€ncal soluion oi inirial bounddy value pobllDs using iadial b6h tunctio , PhD tnesn, Chulan Ish.q Kha. lostiture of Engineering Sciences and T.chDology. Pakhld. l49l N. nycr .nd W. B. blsis l50l turctios., B. Fomb.rS !d Wridt P/rrr, Conplr. P. Ceilc patlmar *h.. elvint a (2@'l), TnEPotl sclEncs on t sPhcE using nditl ,W. t059-&4. (2008), Or choosiry a 6dial bslis fultd,on .nd @Netivc PDE on I sh.F ! 3PhF., Co,9!,. Pi'rr,22?, P? 275&-nW, t5tl M. D. Buhlnlnn (2m3), Rrdi.l BGis Functions Tl'cdy [email protected]. I52l G. E. Uninrsi,t Prctt, Catnb 8.. Fsh.ud CmD, M6hfie Apptorimtio M.thodr witli M!ll.b. Sci. rlc I5tj ttd lhPl@nutions, Prblishi'tg Co, Pt .Un. L J. S.h@nb.€ (193E), M.dic sp*as lJd compldcly ,ttarlr, 3q pp. 8l wa.rd notolo!. futttions. ?tdr. l-{E4. polffd.r nudy of tlrdyS nomquadric mahod fd $lrrccd dlta inrdpofrtior. t ed@ Uv.tun National Laboruory. T$hnical TN.ar 15!l A. E l55l M. A. Golb.rg, C. S. ClEn (1985), A dd S. R. X!ff (1996). lmptotld nultiauld.ic f/4, lpP'oiitrflion fq p9nid diflGMtrl q$tioN. E4. AMl. 8oud. la, PP. 9-l?. I56l J. Hick ftll lld Y. C. Hon. Rdid brsis funtid approrimtio wird li.ld fmn sca!.ad t5?j dlt . l,t !. Appl, of th. sutf&c S.i. Conpn. 4, 1998,W. B. FmbcrS ,!d G. WriSnl (204), S!Dl. cmPutltio of int ryolac to dl vrl6 nl-247. nultiquadric otdF shrF p@n :t r. Coflp. Uztll lppl,48. PP.853- 867. l5El H. wcdhld (205), S.lndd D!t! Apprdimtdon. Koln &d M. wtr81 , Erfs. P. Ci.d.l, A. ls.d.s.R. C!nbridg.,V.K: [email protected]|. uair.Pt $. 129 l59l H. wendland (195), Pi@wis. Polynomial P6i!re .lcfi.i|c dd conpacdv supponed Bdilr funcrios of mi nin l d.s@. Atb ConPut. M4th'4, W- 389 t60l M. D Buhha.n (2001), A nc* class of ndial bash funclioos wirh conpact sur{r]n, Math. ConPb,7O, t6ll c. W 3A-318 E. Fashau.r (1998), On sn@rhing for oultilcvcl aPPreritufion enh mdial b6is tunctions, h Apprcximtion Th.ory Ix, 2: conPutational Aspec6 c K' Chui and L.L. Schunaker (eds.) , vonn ftiL Univenit, Pt.s pP 55'62 t62l z. wu (1995), Competly supPoned Posirve definne Edial rlNrions Ad' Co,tpa. Math,4,W.2a3. 163l R. K. Bea$on, J. B. chdie, finctions: Mctbods bed and C- T Mouar (199), Fasl fininB of Edial on PEconditio.ed GMRES bsis iteffiot Adv C"lP!' Math. Il,pP 251'21tJ t64l L Ling and E, J, (as (2o{,4), A Llst-squ@s tulcoons coll@tion relhods L65l i. Li and Y.C. A d! P€ondiioer for 6dial Co,tpu! Moth.23'PP 3l'54. Hon (2004), Don.in Dccompcirion for R.dial Basis Merhods, NuneL Meth. Pan. 166l L. Lins and E.J. Kosa basis Diff Eq.2O, Pp. Mshl6s 450462 (2004), PGconditio.ins for radial bdsis runcions vith domin dccoEp6idon ftthods. M4th Cohlpt. Model.4o 9F t4tu427 Y@ (20G) r,cal Edial blsis function bed difierental quadnt@ melhod and i6 applicatot lo solvc two-dinensional incompFssibrc Navi.rstok€s.q! tions Conw| Meth AWl M4h Enery' 192 16?l C. PP- t6Sl Snu, H, Ding sd KS. 941-954. G.R. Liu .nd Y.T. CD (2ml), A l@al Edial poim intcryolation mthod (LRPIM) for IEe vibndon analyes of 2_D 6olids. J so@d vib 246 PP 29 46 t69l Y.L. wu and G.R. Liu (2003), A mshfiEe romulalior of local ndial Point inre+oldtion melhod (LRPIM) for incoDPrssible flow sinulalion conput M..h. 30,pp,355-365 ocshl.s mlhod €dial brsis furctions. Lr. 1 ]V!tut M.tMs. Ene.54, pp 1623-1644. t?01 J. t?ll J. c. Wang and O- R. uu (2002), A Point inl.rpolation based on Uu em2), On $e opdnal shape P&ahe|es or ndial basis tu.ctions used for 2-D neshless ffirhods. Conput Methods APPL Me.h EnS c. wart and G. R. l9l,pp.26112630 l72l C. R. U!. K,Y, D!i.ICM. Un dd Y, T Gu (2003). A radial point inte.polation mrhod for sinulation of rsoninensioMl pic@l4Iric sltulw. Smd Matc. s,ru.r, 12,PP, l7t-180. {?31 N, Flyr and w. B, wriehl vaGr cqlaions on t?41 R. Plarc and T. di$rcua ons for a (2C{)9), A splEE. DiM[ Pd, ndial basis tuiction method fof the sharbw i 5e A,465.pp 1949-19?6 Om6), ErSenvale stability ol Bdi.l bdis hnaions [email protected] d€it Poblcft- Co,tpst. t dh Appl. 5r W. ' l25l 1268. L?51 s. A. Ssa (2008), A nunerical study of $c accu.acv md stabilitv of and .symm.ric RBF coUmatiot D.Uods for htF svmetric bolic PDES Nuner' Mdh Pa . Dif, Eq,24\2),670 46. t?61 S. C. Reddy and L. N. T@felhn Moth,62, I7?l PP, 234-24 (192), Shbilitv of ln. ftlhod of lites N@ei . G Ddzin md R. s. Johnsn (1996), SolitoB: ,n inttodrction- cMbridse U n i re B it, P re s, Cmb.i&e. P. 13r l?El ft$od of liEs tor thc.utEricll $ludon ol Kdv cqufon using ddi.t b$k tomnons. Eni. Atul gornd Et.n,33, W tlllI l?91 tu$l* Q. Shcn (2009), A lE0. S. Haq. N. Bibi, M. fd thc Usnd lid S. I. A.l-ntrtzi (2010), M.shlcs nclhod or lircs nurrric.l solutid of gcrcnlizcd Kl6mdosivashi||3l't c4t M at h. C M pu t, 2 11, W ti6 APPI 2404-2413, ii .nd Nrbulc@ of w.w frents ditfusion Eeriotr iy$em. P.o8i ?fuofu . t/,F. 63. PP. 1885.1903. lEol Y. KMndo lEll c. sivashinsty (19?7), No.lin{ lnaly.is of htdlDdtnrnic itu|'bility ir nam. lEzl A. P. ts [83] (1980). lnstrbility Prn I. D.dvrton of brsic cqutioN. A.|. At H6pd, R. GriNhlw (1985), vis@s fluids. Pr)r Filrdr,2E, @!t. 4, hhinr pP I | 17.1206. Nonli@ iGtlbiny .r th. i.Erf!c. belvdn pp. 37-45. T. Tsozlli (1976), P.Risrctrt Fopr8rdon or comcnuaton wavci in disipatirc m.dio ld fm 0Fm.l equilibrilm, Pre& n!.oi Pirrt 55, Y. Kunnolo and 1976, pp. 35G369. lE4l ,. M. Hynd md 8. Nicola.nro (1986), Th. Kunnoto-SieahiMty cqlatid: blidgc bctvtcn PDES &td l85l dFrnical 3y3rcN. P/,rrid A. P. H@F. rnd R. Gdmhas (1988). E, p?. | | 3- 126. tO. pp. tlE 405429. T. S. Yrng (19?). On tnvcllint -wavc elurio6 ol thc K@notcshiv6hinskt cqunon. PtFi.d t87l I TdEllins s.v. soludds or Kr6no{Gshiv.5hi.sty cqudoa. w@. Molioa, 186l a , R. Glibshlv .d a I 10, pp. 25'4?. A. P. Hoopd (1991). Th. mdisrcre of a eltlir clNs or u.rcUiry wavc dunons of tlE Kuramb-ShivNhisky .qua!on. Phyica D. 50. PP.23l-23E. I8El G. l. Sivashinsky (1983), i.u fl.nes. AM, t89l 15, pp. 179199. Y. Xu dd C. w. Shu (2006), Led di$ontiruou! calc.kjn ncthods for 0'c Kulanoto'SivlshnNky cqultions rnd 0E ltorr?. .ouPlcd XdV cqualiois. Conput. I90l FiidM.c/r, brtlbiliiics, patEm-fomation, lnd turbul.'c. in Metwt AwL Mah. A. H. Kharcr .nd R.S. En8ry, 195,W. 343G3447 T.s!h . (2OE), Nutuiiclt sol[ti4s ol th€ lpneolizd cqutid by Ch.b'.!hcv tpetr.l coll@tion Kuranorcshilashinsty @lhodr. Conpu| MathAWl. 56, pp. t465-1472. I9li c. Akivis .d Y. S. Smyrlb (2004). lmplicn+rplich BDF mdodt rm ln. W l5l 169. C. Ma t2@9), Lrni@ BoltznDd ncthod fd thc KubmorGsir.shi$ty cqodion- Appl,NueL Mch, I92l H- Lri dd 51, 8*mli4d K|Jt|lmlo.sivlshirsti .{u!tid. Phrsico A,388, W. t&5-1412. l93l M. Uddin, s. Hlq rld S. lil.n (2009). A msh-fE nu'Eic dlnod ror slutior of thc fuily of K||molcsivr$hituky cqotiom. Arpl. Marr. C@p,r. 212, [94] E. W T.dtuc SIAM J. I95l 45&-469. Mah. Anal, Z. Chlia, B. Shib so@ s.ll-po6dFis or $. K@[email protected] eqution. (1986). Th. 17, W. 88/-893. 6d Ur ZHtb (200E), A ui8cd tlltie Bol|zhon nodel rd mnliielr pcnirl difi.Enti d .qraion . Ch@t Solito^ at l Fmctals. 36. W a1+aa2_ t%l E. Fan (20(J0), .q$ri.ft. l9?l Ext.id.d lod{urction m$od ed iti applic.lions to ndlin@ Phts. I2tt. 4.2:11 , pp. 212.218, C. Y. zEng, H. L. Te, M. R. Uu lnd L. J. KoiS (2004), A lattie BolaDann Do&l and rinul.rion of KdV-BurgcR .quatior C@nu. ft.a^ Phts, 42, Pp, 2at-2u, 133 t98l M. A, Chug-F.ng (2005), A tt w latd@ Bollzrllm dodcl ror Kdv Butc6 t eg@non. ChiL Phrs. [99] n.22,Qp.23l3-2315. H. wendand (2001). Grusian inEryolaioi M. Nemru (Eds.). ?i.,6 d App tutitutio. evhi€d, i.i X. m.ory, Joh^ WileJ, W. {1001 A. H. D. Chene, M. A. Colb.ry, E. J. Kans sdG. convergence and eqvno$. Nuer Uoll L. Kopotun, T. Lych€, zinnio I lO- (2m3), Exponeiial h{ dultiqua.ldc [email protected] rethod fo. P.nial difie€nrial Meth. P4tti4l. Diff. Eqnt, t9, pp 57 t-594. Collatz (1968), Ix. Numcrical TE tm.nl of Difleettial Equations. 3rd .dinon, Sp rinqe L V..la8. [02] R. t. t veque (2007), Finn diff.dnc. Mctho& fot Ordin'ry od Partial Ditf.EnrialEquations.S'AM, Prblicarionr. ISBN 9?8{-89871G29-0. 0031 T. Kawafid (19?2). Osciltatory Soliury Wav4 in DsFEivc Media. J. PrrX. soc..top@, 33, w. 2&264. U04l T. Bridg6 and G. Deilc (2002), trE Kawan@ Eg@to ad tiFd lnst2bil,ty of Soli6ry wrve solutions or its A.rctuliz2'ioni SL4M. J. MatL Anat, 33, PP- 135G1378. tl05l .1. K. Huntc' md L Scheu.l. (1988), Exisi.@ or Penutu€d solikrt wave Solulions lo a Mod.l Equltion for Wal4 Wav.s, P/t 1106l D.J.Bemy(1966),t .8 waves Il0?l S .P. Lin (19?4), Fluid. Mech. 63, ll08l on Uquid Finit Adplitude 191 4, pp, 4 11 D, 32, pp. 253-268. J. M4rft. Pnts, 45, Sidc-Band Slrbilny pp 150_155 of a Visous Film. L -429. Beki (2008). Symbolic ComPut tion dd New Fmilia of TFvelling Solutions for tn Kawah4a and Modified X.wahd Equatrons E. YusufoSlu and A. Exact filn. ! C@put. MotlL Appl. 55, W- | | l3-ll2l. I l09l D. Kaya (2003), Ar Explicir and Nuncrical Solutions of som€ Fifth-Ordet (dV Equtioo by Dsompcirion Metlbd. Appl. M4th. Conpd. 144.2.lp. 351 363. lltol E. Yusufoslu, A. Bekn arld M. Alp (200E), Penodic and Solitary wave Solurioro K!*and of KnwaIaE md Modificd SolitN Equations by & [email protected]. pp, I ll93-l191 Itlll L Iin (21D9), Application 01 Sire{osine Method. Craor, . VriatioNl hehtion Merhod ad HomoloPy Penubaior Merhod lo rhc Moditlld Kawahm Equarion. Moth.Nti.al and cohpskr Mo.Lllkg,49, w. 513-578. lll2l A. Korrmz od L Dag (2009), Crank-Nicolson' DjffeEnnal QuadratuF Algori$ms for $e Karahm Eq\sriot. Chno! SoLnoN & Fndah,42, W 65-13 {ll3l K. Djidjeli, w- c. Pri.c. E. H. Twi@lllnd Y. wang (195). Nu@ncal M.thods for tlE Solutaon of lhe Third- od Fitd-O.d.. DnFEire Koneve8 de vrics E4rari6s. J. Conprt. Appl. M4rr,58, pp- 30?'336. lll4l J. M. Yu$, l. Kawahm-TyF tlr5l Wu (2m8), A Du.l Petrov-Calerkin Medod for $e i. Shcn ond Equtid' J. tci Conpu,34, W 4443, S. Hrq and M. Uddin (2011), RBFS oppoxinarion @thod for r$wonda e4nanil. Ene. AnaL Boufu.l ELn 35, pp.5?5J80. ll 16l A. M. weNaz (2007), New solilaty wsve soluions to the oodified Kawahe equaioh, Pn rica. tllTl lztl,rs ,4,360, pP, 586-592. $lhos elutions for sevenl fms of tb€ fifth!sin8 th. tuh netlid. Appl. MdL Conry\ la2, W. A. M. wazvaz (2006), Abundant order KdV equation by 283_100. Il l8l J- C. Ceballos. M. S.pulved. arxl O. P. v, villaSru (2mD, The Konewee_d. vriesKaealtm equario in a bound.d domin ud sorc MdlL Conp6. 19O, nuftnd Esult5. , PPL W.912436. r35 i. Momsoa l. D. Meiss {d J. R. Cey *aws Phrsica 1I D, pP.1U 336. tltgl P. tl2ol L. R. T. Gddner ed G. A. GaitE (198a), S.attedng of RLw solihry (199t), soharv waws or$c cqual width wlvc cquatron. ./. Catp!. Prrr ic, l0l. pp 2la 223. Il2ll B. G. Archilla 0996), A spectralDedod for Phrtb, t25, Il22l pp. J C,n/!r' equal width equaion 395402. S. L Z.la (2mO). A lsr squ@s CMpa. Math. AwL Mah. Ea.. tl23l tii finit eleftnt shem 189. K. R. Rbhn i2005), [email protected] for th. Ew cquarion. pp.58?-94 ft$od using quanic B{pline fot equl widrh (Ew) .qo tj6. AwL MB|L ConPlr, l6E,Pp. 795'805 tl24l J, I. difietne @thod for Ramos (2006), Explich linitc eqn{rio s. the EW aod RLW Appl. M4th. Co,9ut, 119. W. 622434- t1251 8. Saka em6), A finne cLmenr m.thod for cqual width qn{rion APpl Math CMput. tl26l 175, pp. 13GI41 B. Srkn, L Dtg, Y. . Detli and A, Korrnaz (2008), Tlree diffeant mcthods for num.rical solurion of the EW tt2?l L Daig and B, Srr. (2004), egnaton, M4th. Conput. cq !rrio.. Ene. A cubic B{pline colleaioo method tor lhe EW Ap,q,Pp. tt28j K, O. AMull@v, H. Bogolubsty 38192. stld crand. of imldtic sfiion intctuiib. tl29l v- C Mathar*ov (1967) Oe Dm Phys. Iztt. A. 56, pp 421424 L. R, T. Oadhea C. A. Cardnel md T. C.yikli (194), Tnc bounddv lorced MKdv equ.tion. Il30l AMl. BoBd Eletu,3z, pp 556'66 A. Esen and J. comput. Phys, I I, Pp 5_12 s. Kurluay (2008), Soliu'y wlve soludotrs of $c modified equal width wav. cquation. Comlr rvotiizar S.r'. &'.r,stnd,13.PP 1538 1546 ll3u A. Een (2006), A lump€d Gale*in helhod for rhe ouensl soloion of $e nodified equal-width wdve equation using qua&aic B{PIines lnt J CMput Math,83, pp, a49459, Il32l ed K. D. J. Eves R. wadi (CEW) eqution. It331 B. Sa*a Rdrd @005), Soliraty waves for th€ generalird cqlal r,r. J. Cdpu| Math,82. W. 445455 (20('). Algodlhm fot tumricd sludon of rlE trFdit d m$od. Mdh- Conp Mod2ll,45, p?- 1096'l cquarion using colldation l!341 S. L Zti (2000), Sohary eq! ron, CotAut, tl35l Y, D.rcli (2009), Phrs. equal widlh wa!. intfactons for the modined Couu@, equal I 11 wid$ wave 126, pp. 219-231. Radial bash funoions nethod for .unencal solution of $e nodified equat qidh eqution, ft t. J. Codput,Pp. l-9, iFitsr. whithd (1974), Lirce ud Nonlincu waves wit YlnE^ci.ac. N.|| lf36l G. B. ll3?l R. Xncbel (?000). An IntlDducton to mllEtutical tD@ry of vaves- Moth.tutiol Soci..t, I6titub Iot advanccd uliet: Ptutin4rct, R. Au.fl:dn I lllSl L. Dcb.alh ( 19??), Nonlined Panial Difieenrial Equarions Bi*hause. Botton Il39l R, E. Mickens (1943), Matn nddcal Methods For Scieoces. Wo.id uDemltics ll40l L ArSyris Sck^tifE Ptblithing Co, Pte, ll4ll Ikl Engineering 6enes on advances in for applied scien@t, 65. aid M, Has lppfication of tlE finirc w7l $c Na$ral md (l9E?), An c.gineR goide to soliton phe.omc.a: dtunr @$od. Conptt. Methods AWL M..h. En| 6l. 122- L Da! (199), A quad!"!c B-splinc linic.lcmt method fd elving nonl'n..r Schrodingereqution, Conr'li Mrthodt APPI McclL Ets,114.pp 241-54 tl42l L. R. T Gadner. G. A. Gardner atd A. ll- eletu nsinS qladratic spli.e finile A Ali ( 1991), SimulatioN or sohons rs, ConPut Metholls APpl Mech Ene 92. pp.231 .43. tl43l tl44) L. R. T. Gadner, c- A- cdrlrF!, S. L zali, E I shadwi (1993), B{plinc rc nd Schrcdinget €guadon. Cr,tPlr M.tlrdr finire clenc AppL. M.eh E48, loa,pp. 303-la, studiB of z iie L. R. T Gardier, G. A. GddDer, S I Zaki.nd z E I Shsra*i (1993), A l.apfos slgorithD bnd stlbilny studies for $€ .onlircd Scltodi n8.r .qu ation Arabiaa J Sci. Lla51 En..pp.21 32. M. P. Robi$o. and c. FaiNater (l9a), OtioBonal sPlin coucation Dethods for Scttroding€r-rr?c cquaion in one sPace latiabl. Muer Marlt 68 (3), pp. 355-76. tl46l M. P. Robiso! (19??), Th€ slution or tonliftar Sch'odingct equtio.s usins onhogonar sprine ll47l col@tion. M. Dclrour, M. Forrn linat SclediDgcr tl48l E. H. Twtdl, A. ed Co nput. Math ApPt.33\7), PP.39-51 . G. Payne (1981), Fmit€'difcEnc. elurions ora non €quation. I c. Bntros, Cotuput Phtt,44, W 217-48. J. C. Neqby (1997), A finite'difi.rcnce method ror elving lhc cubic Schrodinger eqlltion. Malh conPut sinul, 43 pp 67 Il49l B. M. Hcrbst,l- Li, Motris dd A. R. Mitchcl (1985), Nuftrical cxpcrince wnn rlBnonun.dSchrodnger.qualion. tl50l 75 t C@prt Phtt,60,w 282-305 T. R. TahaaDdM. J. Ablovitz (1984), Analyrical and nlncicdl4pects orcertar. norlinear .volution eguations, IL Nu€ncd, nonlineat SchrddinSer eqlations ./ Conput. Phrs, 55, pp, 201230, tlsll A- Kodoaz, t. Dag em8), A diff@ndal qladraoa alsorithm fo. simtrlations or nonlinee Schrcdingd equion, ConPuL MatL Appl,56. PP. 2222-22U tl52t Y. DeFli, L Dudun dd l. Da! (20091 Soliton $lutios fd NtS.quation usins ndial bsh furcfioN. Cria,s S,litont ann Ftuctib,A2. pP 1227-1233 tl53l S. H!q, S. IsbE a M. Uddin schrcdinrer (2009), Nutrncal Solution of oonlin@ Equtiqs by Colloc.tio. Mcthod 6int Rdid Eais Furtids CnE,l4.pp.lt5-l35. tl54l v. E. zrlhNved A. B. shabat (19?2). Ex4t thory of iwo dinensional elr fdsint and one dinnsional stu wlvcs iD mtlin€d medi.. J,ti4 Pl''! IEPT. 34.W.62'69. t1551 J. M, Mif.s (f981), A! dvol@c lolilon poblem- SAnt l APP| Math 4r. W 221-230. 11561 M. R. Dlbd (1994), Dotnrn d4onposiriot.nd locd tctioemnt muliqladft appoiidarions, ApPi S.i. Cotpu, tl5?l R. K. Bcltloi (2000). w- A. ught tmtion i .ryolrtid 22,W llll-1740. cqoadom: 1, pp 146' l? ror l. od S Billi.es, Fsr $lutior of the ndial bais iLcotrP6i!@ tuthods. tlAM J- S.i. Conplt'