20.40.2044 x a A $02 = 1,05, 2,53 < 5.66 = 329 - 18,55 Ae 4,33 0,56 hte 200.50 269,26 669,94 -0.4d -O,.54 AGE 36.85 “102,65 “255.43 Alles cocs nome’) 6.60, 6.67, 2.96] pa absavic roud sum VA rwo nance Oxf HE» 136,53, 495.28] cond A]. YAIWAT > 6.69 4438,69 +595) 1, L itl cond systemt Methods For Selving Linear Syst. of Eqns. (Ax =b) T) Direct Metheds eae: 1. Graphical Method 2, Cromer’s Rule 3. Elimination of: Unknwions (Gauss el of Gaust-Jerden ei.) #1, LU Decomposition 3S. Direct Invession T) Indiceck Metheds (iterative |. Methods ) epee Cee Ee Cee eee ate eee Peeeetoerece Er v1, Gavss ~ Jacobi Method %2. Gauss ~ Sersel Method |eo sae eee ee. LU. Factorization. .{ Be compost onl ae i Ax = b However KAS ;then Gavss LU factorization if you need. to Gauss - Elimination con be used to - elimi nation Seperate | the Solve becomes solve Atzb. che some. syst. with. different insufficient fie — Comturning — elienination of enatrix A from the maripuiatnns of RHS Once A is decomposed into A=LU mvltiple RHS vectors Con be evaluated ef fidertly. FACT: Any matrix thot hos all its cliagonal elements nonzero Gate writes a5 9 product of @ Jower riangviee © an Upper tripeguior rratrix in an infinity of wow. Choose. diagonals did) Gee a ( eee Gorp!) as 5 4 Vy. Sean Dag. Tobrtagd whe ft ts whet Axe b | 68 tf af GY Jolol lu! Us | a hen | ayy an 4 ty ty DLO 0 ua J i) @vu225 0925 bart @ byvgtad=8 UKs @arvu = 64 B56X5:4 0 6% 4oq= £4 2256 ya 4d 25@ Uavus 4 eet @ hvu tu Ob 0a. beevan = 42 D564 4 Un5 of. tg, = [uh 25.96 day23,5 25 UO), 2-41.96 @ bar Pg bay Yay + Yas! bag-= 0.2 4 Deo eee eer) dela 4 usfo -ba [sae as «| Gee oeeieoa General Dolittle Algorithm @. First row of Ut Ug eej jet to N b. Fist col of 45294 tag ig N ua €. Second row of U! = A233 ~ eq 4. v2 to N uy 4 d d. Secons col, of Li: fig = lan = ty Ug} dae mee AA ARARSHRABARARARKHAHATLSKAEOHRHHLTCOKHKREPARVWTSS &:OXOMple 8 a “Solve the following aysiem with LU factortzaten : 2505 4) f 7) fF doe uo 8d mw fos | i722 Ax day 4 % 239.2 25°45 4 fa ° olf 235 4 Az | 4 3 1] = | 256 4 @ O “We -156 E tay 12 4 Ste 4 o 8 97 A 7 a u LU x eb my Hab ° Yi f 106.8 Yr= 106.9 ° Yel = [4992 [ 25+ =e wy Ad Ys 1279.2 | S264 Ssy24¥3 = 2792 4 i yg0385, p 106.8 ) [2% 5 4 Vf) fob } yr (96.2 fe 9-49-86] ) x2) =] - 952 0.735 ° ° oF FAxg 78S x3 =. 1.05. i i Gata 156K3 = = 962 x 2199 29x + 5 xy ty 2MObB LM Hy, 3 0.29 ' (0.29TE «Indirect. ( tterative) Methads for Selving Linear Systems i 4 Are based on an initial guess fer the soln vector X (AR = b) Gnd produce beter approximations by sequenté’ of Heradtions. erations. stop. when error QSy ON of iter 10x # of iter, roleronce. @® Govss -Jocobi Metined @ Gouss - Seidel Method 4. Eavss Jacobi Method: Ax = b Apply elementos row operations fe & e Git diagonal elements of A to be ft, thal matrix A® aed corresponding & as b* pick X° Cinitiel guess vector! X= bt -Bx es ben Be aie iie Cela oe ext. x BK ot We proceed until < Bjoerance oR for each iteration. j} and for all X;'s coleulaie; #102. £ Fyerance SHSSSSHHRHKRARROSHSRASRSSKRAKRKOKRABKLOKRAASKRSKHEHHKARAARAAEAHE.Wa ET allgorithen + (yan) : KEY be4i2e-u0 voriable AUMmbEE (Re ayn Me Gauss Jacob! 2. Gauss ~ Seicel Method : Use the updeted (most recent ) valves..of X/"" in the iterahorty ey kat 2 & rT be - 2 Qj Xi - EB aij Xj Xr bas Z of fet2nn Qn Convergence Criteria for Gavss-Jacobi © Seidel Method DIP A con be reartonged such thet foe eath row of & Wiagonal elements have magrifules that ore greater than the som magnitudes of “remaining elements ia the Corresponciing Cts 5 a lay /> = lal P42, .. yn (BMA GONALLY | jee \ BOM NANT F . vi YF Diagonal deminanty if @ SOFFICIENT Cont of Convergence for 67° 6S methods. - Ee eres eee eeOt Ad. 2044. CHAPTER 4 SOLUTION OF SYSTEMS ON...NONLINEAR EQNS Important: i /j) = unknown ~ariavles ke iWeration # © Graphical Seta; eqn 9 ys F2K 2 ean 3d Xe tay=i DRAWBACK, A)H's not always so easy do plot fhe eqns: 2) Gono! agpty his ae reelhod for more thon eee eqns. @ Fixed Point eration 3 Ch 25 FRET — rewrite fix) =0 ~ xe glx) such that \gbil <4 ¥ Herae. Ch 3% FRI ~% Gouss Jacobs ¥ Gauss Seidel x ab een HE age (H jedaeun (jad FOgonaliy dominant. Chu: FRE. Pull one unkown out of each equation . such +hat = 9G) jetpun (fet) plow MAL Some peobiem with EPI 5 “SY Find. the root ia the first = cont. ye ( ere 3x72 (0.5/4) or (Xy ye? = 10.51) gisterie@i@ere TSSSSSMARAAARNHSSKARAPSPHKRAKHHKRAKRTCHSLKHLKHHHLHOKOKRANDRAATADGweuveuvususevueuUuL » ey Y co ” k o 0.5 1 _ 4 Os 0.966 0.433 0,866 0.433 0,940 ) , . Improve 1 gong stewwly ! Fea? i Ay = Yea Yee = (gE 0 Os q 4 0.5 0.8660 a 0.49330 00,9096 3 0.4548 0.9945 4 O.4u74 0, 8995 we To improve the convergence. of FPE for. syst. of nontiaeer eqns. . use the Updaled values. instead. of the. older. ones whenever posible. (Just ike 65 imeeees es)15: Ad.A014 CHAPTER. S : APPROXIMATION OF FUNCTIONS dear Averoge Temperglur ex : (yeor) “iso bod og ~——-8000 ~) Why Approximation of a function? o> Te etropvlate @ dose point “inter poloie a alata point Blocaton isiraiorma. © dodo gunts (nA) polynomial! 1) lagrage interpolation : has collocation property a dota point fa-4) Porater pocind 2) Cubic spline excel.) » has collocation property 3a cube function senuten every 2 data points. ~> (n-1) piecewise cuore fonctions 3)Least squares 2 No collocation rope mt fégre ssion, \ propery 4 siraight tine or 4 non-linear function FSESSHHHRHFRAAKSSSHAORHHNKNTKRAERTCHKCHEH HK HSASNKNHERAARBRAVE