C1238 — Gnbro. bo Nemenical Analysi sis Instruction £ Rath Dasgapta This 2 Ani) Kumar Vinod Kaclant Lonit Kagel Sohedule 2 Tue /Fai, 6130-685 Pp Le 002 Topics * > Problem Fon ma lation & Eynoh Analysis a) Linen Eq” (ax=b) 3) Nan ~ linent ey 3 pot Finding 4) opheizakion s) Cuwe fits & gnpenpe teHin y def pormntionen & ‘nkgettn 6) Numenieal 4) ODEs g) PDES Ref enencor :- 0 Numenieah Methods fon Enginees — Chepre 3 Camale . eel, enh Edith Me Go 2 Seventh Computing 2 Are iran oducteny, sunvey eath , SIAM 7 chemi EN ineny 3) Humnenieael Metrads fe Carmonidge, Guiv. Prew — Beers, 4) Applied Numerical Metheds wth Matlab - for engines J ecienks — S.C. Chapre eaten t))pading § > Greet End- gems 50%] id= Sem + 307 of 3 will be daken) 2 Quinze CBr 2 No Rens [Re- en m will be 0 ffened Oven, noblem Rowing proce ve lop a moth medeh tre e3* iy the an odel amt he alges fo soe these. 9"? > Run dha AogZroone a pact & nrolsae! * Mathemaheal pravlen should Nave swell aol” br te aol” epovkd & Langit posed. le He usulls in inpor anoald in owt ( conhrusuily Gener wy Rep dimer \ =z Replace Antegrele 2 Jaca infinite, dimers} arch Byes? finite Sums> Replace dosivehi > fists Aifferoner yasblem —> Linen ones > Law onda me Wa» Dimple aveoley: pogroms) th Arripko fos . Ae sol” ¢ approninae re need fe wbrole AceUnsey & heave o Hey gq s Jemerk cally, improving Sources of error 7 ame dellin > Empl piead am eanmnarents Lie al 7? prewiow cxsen pucoeiers [Dior Aeehion athic C Canpukons woe pure Bite ais) Je dak minimise ( canvengenez.) 7 Nom lineon > High onda & > Complicad ¢ > Genuak amokiien — amotneoo ‘* devined - > Tm cation = Rounding We need argo tm cak™ ennons re praly 4 He a) = ' —~ia«rtl- 4 neo 2nel 2 fs . wan ge 4 Called te Lexitonitg, sonteo “SppMOR + Written by Madhave. (Sadia) pam tHe. Kerala Sehoolof Hstnonemy (4-15 cembuny) [Fun + Reed wiki onkrele on Madore Sanieo’| janine = ge We BR ere Ws a sis F ~ Te ek C . a wei) = 2 V+ 22 — {zo (eee ' * ist oer +28 iE Le ues yee Beet Z ¢ 12 so4o yew Semtes _ ee os mz | 66 ack ° 6ce be 2 6666F ob 5. ecb 0-#3333 nes 6-44401 0: F F038 melo O89 goe Comeck Vere 4, Uh, x. 07 8839 eq guppose we 6w Aictngueh betwen an obyeat hich weighs (gn ve, anotreswhich wer iS 24 ~ OW who hi hs m. % fr st om Aasolutin is 0: 001 ky Ke ponhing a wt ap 1284528 ky is _folse page sion Fixed pt Ae heremtahio dh nwmbow fied 3692319 TITIL.FF FFF H of decimal pleem ie |- 000 12.836 if the ferret i e.4: “abe \onger* ae is 99999: 99 299 dwnalleot # " 5 0900:0000! + Casa Bre Lloakng pl ns pavern tenon Je ig nor pararned ase eae RE aumbon dy digi P on ie J gnsakn port gn = camp uter a heal no is ne patoenied 20 nan (doe bo Be e.) eF gE FP pr P pase Ate spe (_ binony , decal ek.) P* pacasion dai: on ing cue thet of a Se) (Lvl: Exponent Range LZe £0the th x 1456-23 is neparenba ao aces nie 3 eq do dt Eo ds J ea(t+ aw B+ S-% ~ 2. \ai6 sign vo Woe TBS De Normalised rooting powt apn 5 do bs absayy This ig one canvemhor Anothon is aoZo, 4#] mon 3270 gn Be abeve example, tre -mambssa is 48023 Y sno monmalised Ploakag PRE Jem L& UO ond preci MOS. CON woh exponent nomge and lease Pe has Mang da ? be. ap rooenis \% sig ee v do dt 82 Ages exponent 7 Dehercea 1 4 “ \ | Fe u-L*) choles choices eMoiees 1 [2x &! xf? (oi) |* \ mos. co be ne prmronind sypiem Qu the nonmaliced Flooring poms find He Amallest @ve J YS wri aloove, cam be nalparsented for pe lo.“ geo 2% 10° Sev pol Hone geneity "9 mC Mimay” Used system an above, Sm Pe same norm ~ and th Benge num repre fon po 10 w yen “tak cam be C aac) y ads se TH Langer 5 — 2° Q999---- x 10 e dame Mone gen hg ~ amy, poe \2 we eye no. 8 } J ot 5 g-)~ Pel ~ Gola * a + ( j t ee p? er fp dao 5 2 a oe + ga [rr = @-') t(ic 6°) pe. He . - e*) Br a ovenft pomt We wil ee tek flooheg Snbbiem | nos. ane (NOT) omiforviy distr rlouted mw waiton ta number [Ine Dhowbacks © \ooxnw +. rel sponte him 7 Nok all nos: Wave & n 7 Nos: one nor cumi foro ly dighniboted > Ovenflne fe wader Flow eno a> dntenvels peksem nes: incnear® os NOS. an magn stud _——— IEEE Single Precis Preasim * \EEE Double pe 1022 o= Difference be fae precession & accunaey \n meagunemenis oy ayolet accunste. Precis Jou imprecise ‘recone Comp ahrg a port's vain oe precision thay the, Sone will be door not rep! net precsim - accunate tO Exnnons Reletve enncr = ap prox. value — Fi Tone vole,Cokeulate Te of hams Me he lt H+ we 2 6 Relative @wh 4 & f" & no wy 24 fon = os [we = 1.640721 | Comin a f® Ft RoR we the vena of inp £09: we 8 Owtput 43 ay ronal te $ nenact input SR) - Lea) (2) * Ny Depends ™ cheree of alge fon approx: £. Exanple ‘ Approx: Tay 3 Re preoemt ein) yy }s ETON x 2n(2) 2S si(E) @ BL dete enmn—t We need to caladate sin vt pre iI 2) wi denm in Tay tr 0+ 3480 Thue value. ein(Te) RO BAZToh ger = 0-3450 — 0: 3827 = — 0: 00FF be newnrrten 20 TE This “3 Sepa Rr ~o. ~ Vo TE = - fw) _ g(h) “ £ (RK) - fx) - sio(2.) + sin 2) - oF ig 0-3663 \ u Alo 5. oct + OH 90164) \ =z 7 Oo oOtF be Punter enb divided Campabatienck Earnoh ™*4 ynto + 7 DH: been +e = Trance him ennoh poantr & Fe peal) ae «, fo aw gwen algonithr 7 Rewnding enor wor 9 exack athe be ? NS fault produced by on algorithm worry enack onith me he and th, Lome algae uatng fine paediatin J nth me KE Theonem 3— Affe fable wh atte cb- Taylor's Remamda Suppose fn) to (N+) Fimen on te intevh La. bd wipi about tnteh, pended 4 expe’ a et an wo b fix)= $ (xd r Gila) w+ gh xo) Gonele eee Ulm land” ¥ IN pat Fon N=2 wake %- Oo = h g(mern) = £02 * gileyn s $e) phere oe ey xarh] te When WE wonile gile) gas $09 woke h a fe km "(eh 2 We ane aun ecoting the trum coh enon io MB pnee M is an upper bound for 5": gn contnasts if € joy the pewndiny, anor ‘ o, the rng VWON mm neyrrombng gee is ae ce ig mone of = compet pera] . 1. Total compateneyt Endon & Hb + ZE nDuwetve bo = wu - 26 Bose 7° de pp boown oe TOE 3 hophnek = 2 [e es io Double Precision cos Gl) in mathe atic. Catoale | ond woirg He fol aig ea ennetig ota « enone 4 a joe ol) = Choose and pier Leg Natlal Cod Code mn : og(h) Yr ne (Py 1, We * it ono) J ve 1g")Eonwand emoh = §Od- fo Back Wend enon 4 © enue oe OS b pate “a inpet eo thet £8) = (x) = K-% eg We want to evalnole SO ee) at uel (-J= (-<\- or on ®) % I~ om 4 az Foy= oe ore Fonwonk emnoh- = 2 4)" oa = - 00403 2 gack wad oor ! wae Io em % > Re 1 04F2 ge Roms 00472 He. velne she back word error tls 2 nok O-S tre ve codeukate d ten. “ “ " dre exact velae 4 wld we exact velne fh 0 ea C80" “Bock § Aometting akgntg ° oon oe problem Condikm mumbo © = \4 £(R)- BF Tr) 224IL condidined ¢ L . On cendilm no. > Arnall pelative pack wank Ooh leads 40% Longe neletve Formsend Condi hm no. deponda om inpot x a omd +ny te oblein am UpP™ do rot Krowit Bound ™ “the | fonusand enon) = #OD7 gy gQ)- gt) 2 $x AS x) - $s ~ $ (x) © Cm) AX ~ fo) = sae [backward eror| = oe . vim = |¢$ ay yer camdvtin N0- eee] [ #2 = xsd | 5) * of Pine i, He emdihin n0- cce) cach mo: dy He (aver %frat Nao nole = FJ > x= 39) _ 1g) ay 145) $100 dx ne gps! of one prove ms is well emadirined (cw ey f') Ahe west ip WW consume, LA) 77| gemanpe # got = te “ $a) te con) = [s_( We The * © inven prcbtem do 22 w= Y Camditin No. ely) = 24+) _2 y? eg ye Ltx)= tanby Cin) = \% "Gy >| = xen tern) FY |* (aa - es) Fon w= stor? Calakek Cla) % 2748245 x10 (‘23.) te) PA gam bineny (pea, Prd He ment mee atten 2. Assume pAceisin e- [1-6 3 ode * ape (i. D0O- I)x 10" a wpnife ward gne ait yee? lt yen? Le Emeacnine “P mm anbitnany bare Ps Emach = 2 provide a Significonaa of Eemach ‘ band om te Aclahve error in nepaooenbrya Numbrn ™ te floating pt. Syke. dor Ary de Ate 7 ( = » wo dew Apt vo )RB $m denoted athe num ben Let w= The floakny pe represen Re 3. by Ba.Cm) + dstls Sogh+ chops +o p Avgits . ie. gue de» wat ate.) PP fL0n) = on a" 4 sey Relative anor” | ae yen J ay + ated ar)

! ven ply L ate Set ofa & el \eben nelakve chopping enor ‘ _ ce p digit | is bownd a ens emoral i i pl , Roumding Cinskad 2f cho ypna) will wero netanve emeh |e “| £ Emach Tha meng Hmw : ute E(u) = ule &) wih [els d emaen “he precise MV ONAS € depends m He inpot a Sup}ese we Nee wp adh +o 4: Uanally tere ena) mistake > naprrenty 4. $ pos SL = ole ez) oe pew | = allt E)* 1 spe lative ennoh = je - &) | au! alo ar { = bord wal - an, = at! Gn exaek anith mekes Ane ne lahve enone aon be mode Angee b ehoos* we close we ols eo 3Subtrachie_comecttetimn_ ¢ tm (loss of eyntfieont Tloo net: 1-439 x (0 cin a 4, . =1 cigvihiort 1-426 mI? o -1 aigt Subdnee® * 6-0 13% 19 <1 Le piesd = 3 x 19 [only a Ped ts 4 ai 1 is problemas yobee oe b-e : ob eg: 0. ese = 6.3628 [Som as F24—- 35 sob “abou 1. Ne proren, 5 hin Sane compotion witb psu dina 0-31S3 2 & #94 Beq2-3s- 4° Again pawndt ais _ eis Be 4-35 & Age” Oo: 3h = aywisien by ° ae- 36 atbred | alerts | aoe Tay whe Veana® mn | $ . ° * * p’-¢ Epon, prepegee™ We mearura fe Fane velne aw on ©Hoo yuuth error Aaed thin came in fC) 2 Ba) w fH) + £ Glu) sy 4- £ r= ff" rele) anon in we Abels al i) Enrnord bownds aue + anitnme ke dberchis we Br fy 3 = a £2 g% je, Pp , (ald F2 bounds C po Be ow >o) a AddinAbsolulg oven = Jury -&+F)) = (e-+Q-9)] >) e- e2| g tele \ 2) 2 pir P oy te abgolule ener 1% bownded ba He Aum 4 enor bods: a Sublnechoy ce | wy P| = Abs: Erk. = [e&i- é. | & lel * |€2| a pepe Cem ie Try caleukehng the bounds of dk ' on tie, Aclabie cmon DeAatre ennor fon each » Pus faz% Muth plrcebion Rela hve a = yas2 = parse) eglect Rena | ‘ re IN als + ae \o \ 3 ov } cs y dbL \envil ~ [Ene | ™~ ¢ Pov + |Prie Enron abm ex arnple The Vom- ten Waals EOS oo be galved explrartly fon Prarrwe e qr Volume & Hep: NOP p= nRT — no vyonb ven= No d molen = I Ve val. Re Q2-O6 em? ah) | K mol T= 300K a é 2 as VERIO ate (Sam nok. be get ao LR Monten ee 4, Vv oe Mean \ em 190 on and per pecreely ind abcolnk Snore in f (Cts lon expos aro Ye q nT — we P3 3 v2 Yoon ° Dwr (v-‘e ) _nkTt + | met () - nb Vo"indicatyy mean ao < > we mkt — oo” Lepr = Yoo ne ye non 4h V is choosen 6° v vyelue “sel ho a a Lay = WR =nb Vor Vo [x G2106% 300 _ ents 100 — 33°6 a 400: 944 — 1. 249K 10 104 2 = Ao 94k ). Zax 10 = 4001944 - 1244 we 2h6:244pe A ~~ B (v- 67) comhtS VorCP) = Yo [> sv . von oOV- 9) eg vor [Y-42] = 6 var [V1 (\ Fon us 2. (100 3a §) (90? = [-#s3 * anna | - ~653%2: $24 = -3:836 Jae) = G13! = 144iMone qo h(x ) n ON dix ares Gaye h(E gg7) vee " venla)= (2 * Nan (k) * ve a (Sr he csp aired leadin ad He Fonsi wee | (2 ‘ait oy U @le-e)4 @ AN ot nome 1 Some examples Caledote 3 oma? digit 2.001 6662 (sem) pasersion oul 3-01 ~ on @ ae igit_ compo Joo! = 3:002. faerie 3 = O mhis To do batten om this pit {aa+3 9-0) - 3 = = 2 Ol TP _ pare 300+ 3 = 9:5 Oo: 6 3 2 (bFKE