Scribd
Upload
10K views62 pages

B S Grewal Numerical Method Part 1

Nice

Uploaded by

ਕੁਮਾਰ ਸੰਜੀਵ
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
10K views62 pages

B S Grewal Numerical Method Part 1

Nice

Uploaded by

ਕੁਮਾਰ ਸੰਜੀਵ
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 62
APPROXIMATIONS AND ERRORS IN COMPUTATION 2. Accuracy of numbers i 4, Useful rules for estimating errors Error propagation 6. Error in the approximation of a function | 7. Error in a series approximation 8. Order of approximation 10. Objective Type of Questions 8. Growth of error 1.1, INTRODUCTION ‘The limitations of analytical methods in practical applications have led scientists and engineers to evolve numerical methods. We know that exact methods often fail in drawing plausible inferences from a given set of tabulated data or in finding roots of transcendental equations or in solving non-linear differential equations. There are many more such situa~ tions where analytical methods are unable to produce desirable results. Even if analytical solutions are available, these are not amenable to direct numerical interpretation. The aim of numerical analysis is therefore, to provide constructive methods for obtaining answers to such problems in a numerical form. With the advent of high speed computers and increasing demand for numerical solu- tion to various problems, numerical techniques have become indispensible tools in the hands of engineers and scientists. ‘The input information is rarely exact since it comes from some measurement or the other and the method also introduces further error. As such, the error in the final result may be due to error in the initial data or in the method or both. Our effort will be to minimize these errors, s0 as to get best possible results. We therefore begin by explaining various kinds of approximations and errors which may occur in a problem and derive some results on error propagation in numerical calculations. 1.2. ACCURACY OF NUMBERS (1) Approximate numbers. There are two types of numbers exact and approximate. Exact numbers are 2, 4, 9, 13, 7/2, 6.46, ... ete. But there are numbers such as 4/3 ( = 1.33333 «., V2 (e 1.414213...) and x ( = 3.141592 ...) which cannot be expressed by a finite number of digits, These may be approximated by numbers 1.3333, 1.4142 and 3.1416 respectively. Such numbers which represent the given numbers to a certain degree of accuracy are called approximate numbers. 1 2) Significant figures. The digits used to express a number are called : significant digits (ieures. nee ee numbers 7845, 3.589, 0.4758 contains four ean eae figures while the numbers 0.00386, 0.000587 and 0.0000296 contain onl ee a igni a figures since zeros only help to fix the position of the decimal point. Similarly the numbers 45000 and 7300.00 have two significant figures only. co (3) Rounding off. There are numbers with large number of suelnieah 22/7 = 3.142857143. In practice, it is desirable to limit such numbers toa manegeal le number of digits such as 3.14 or 3.143. This process of dropping unwanted digits is calles rounding off (4) Rule to round off a number to n significant figures : © Discard all digits to the right of the nth digit. (i) If this discarded number is (a) less than half a unit in the nth place, leave the nth digit unchanged ; (b) greater than half a unit in the nth Place, increase the nth digit by unity ; ‘c) exactly half'a unit in the nth place, increase the nth digit by unity if it is odd other. wise leave it unchanged. For instance, the following numbers rounded off to three significant figures are : 7.893 to 7.89 3.567 to 3.57 . 12.865 to 12.9 84767 to 84800 6.4356 to 6.44 5.8254 to 5.82 Also the numbers 6.284359, 9.864651, 12.464762 rounded off to four places of decimal at 6.2844, 9.8646, 12.4648 respectively, A 0 as n — and hence if fix) is approximated by the first n terms of this series, then the maximum error will be given by the remainder term R,(2). On the other hand, if the accuracy required in a series approximation is preassigned, then we can find n, the number of terms which would yield the desired accuracy. Example 1.9. Find the number of terms of the exponential series such that their sum gives the value of e* correct to six decimal places at x = 1 2 43 ant Sol. We have faites e where R(x £"0),a< 2x 10% which gives n = 10. nl Thus we need 10 terms of the series (i) in order that its sum is correct to 6 decimal Places. Example 1.10. The function fix) = tan’! x can be expanded as # 2 ge: 3. 2n-1" find n such that the series determine tan“! x correct to eight significant digits at x = 1 (U.P.T.U., B. Tech. 2007) Sol. If we retain n terms in the expansion of tan x, then (n + 1)th term tan ane = cpt 7 CM oi _c “ 2n+1 for x= 1. Henares vanerine sone To determine tan“ (1) correct to eight significant digits accuracy | a = |2x108 or n>108-2 Hence n = 108 + 1. 1.8. ORDER OF APPROXIMATION We often replace a function /(h) with its approximation 9(h) and the error bound is known to be j1(h"), n being a positive integer so that | AA)— (A) | sp A" | for sufficiently small h. Then we say that 6(h) approximates f(h) with order of approximation O(h") and write fih) = oh) + OH"). For instance, Fp Lt eH AHS + aaa 1 is written as Top lth he +h? + One) eli) to the 4th order of approximation. Reh ee Similarly cos th) =~ rt gy to the 6th order of approximation becomes aD e eos (8) = 1-57 + + O09) Ai) ‘The sum of () and (ii) gives a rt (= fy! +008 (h) = 24h + 7 +h? + Oh + Oh eliid) 4 Since Olh4) + £ O(h) and O(h4) + O(h®) = OU) (iii) takes the form (1—h)1 + cos (h)=2+h+ ne +h + O(h*), which is of the 4th order of approximation. Similarly the product of (i) and (ii) yields we 2! +(-8 2 oe 6 yt “3 ot Se 54 Ae Tene tog tag tog t oy + Oh + OLA) + O14) O14) lio) a-hyt cost acd shat ens 4) +(1+h +h? + h9) O(n) 41 nt «) O(h4) + OCA) OCHS) Since Oth*) Oth) = O(h!9) and = aut at Lee onan canes os = On a a4 Bg * Uh) + OCH) + OCH") = OCH) ; pare he Gv) is redueed to (1 — bY 0s (h) = 1+ A + + + OUh4), whichis of the 4th order of approximation. 1.9. GROWTH OF ERROR Let e(n) represent the growth of error after n steps of a computation process. If | e(n) | ~ ne, we say that the growth of error is linear. If | e(n) | ~ 8" &, we say that the growth of error is exponential. If > 1, the exponential error grows indefinitely as n > , and if 0 <8<1, the exponential error decreases to zero as n ~. | PROBLEMS 1.2 | 1.2 1. Find the smaller root of the equation x? ~ 400x + 1 = 2. Ifr = h(4h5 — 5), find the percentage error in r at h = correct to 4 decimal places. , if the error in h is 0.04. (W.B.T-U. B. Tech., 2005) 3. IfR = 10 x%y%z? and errors in x, y, z are 0.03, 0.01, 0.02 respectively at x = 3, y = 1,2 = 2. Calculate the absolute error and % relative error in evaluating R ? IER = 4xy%/e? and errors in x, y, z be 0.001, show that the maximum relative error at x =. is 0.006. yee 2 5. tv 3 (= e | and the error in V is at the most 0.4%, find the percentage error allow- able in r and h when r= 5.1 cm and h = 5.8 em. 08 si 6. Find the value of /= [ “"* dx correct to 4 decimal places. bx os ee : 7. Using the series sin x =x — +77 —...., evaluate sin 25° with an accuracy of 0,001. 8. Determine the number of terms required in the series for log (1 +.x) to evaluate log 1.2 correct to six decimal places. 9 Use the series log, (#2) =2 (« i-*, correct to seven decimal places and find the number of terms retained. (U.P.T.U., B.Tech., 2003) 10. Find the order of approximation for the sum and product of the following expansions : 2 48 ‘ nh oR the Ze gy FOU and cos hy=1- 2 4A 5 ons . 9 to compute the value of log (1.2) a 1.10. Given the expansions : i 2 2 3 LL 500) and cos (t= 1-H +H + 00%) 3! B! at al Determine the order of approximation for their sum and product. OBJECTIVE TYPE OF QUESTIONS sin () PROBLEMS 1.3 rere Select the correct answer or fill up the blanks in the following questions i. ee ee Seas 10. 11. 12. 13. 14, If both the digits of the number 8.6 are correct, then the relative error is Ifx is the true value of a quantity and x, is its approximate value, then the relative error is (@) | x,-x |/x, (&) | x-x, [ie (©) | xe | @x/| x,-x |. The relative error in the number 834.12 correct to five significant figures is .. Ifa number is rounded to & decimal places, then the absolute error is a 1 = 10h ma @z1 (6) g10* 1 1 - 2 10+ © 3 10 (dF 10+. If ris taken = 3.14 in place of 3.14156, then the relative error is . Given x= 1.2, y= 25.6 and z=4.5, then the relative error in evaluating w =x? +y/z is Round off values of 43.38256, 0.0326457 and 0.2537623 to four significant di Round relative maximum error in 3x°y/z when dx = Sy = & = 0. If a number is correct to n significant digits, then the relative error is 1 1 Lio 2yom (a) 2 @) 2 i 1 <= 10 (d) < = 10"1 Oss (a) <5 10-1. If(/g + V5 + V7) is rounded to four significant digits, then the absolute error is . (Ji02 — V101) correct to three significant figures is .. Approximate values of 1/3 are given as 0.3, 0.33 and 0.34. Out of these the best approximation is ...... ‘The relative error it2 is approximated to 0.667, is .. (U.P.T.C., MCA, 2009) If the first significant digit of a number is p and the number is correct to n significant digits, then the relative error is ...... ANSWERS TO PROBLEMS — cienecinaetatis Problems 1.1, page 5 1. 3.264, 35.45, 4986000, 0.7004, 0.0003222, 18.26. 2. 0.7546; — 0.0002 x 10°; 0.00265. 3. 0.0003, 0.001. 4. 0.077. 5. - 0.0004 6. 4.44%. 7. 0.01%; 10% 8. 1.65. 9. 1.149 x 10+ ; 4.836 x 104. 10. 600.0002. 11. 165.55. 12. 0.17312 ; 0.003178. 13. 0.5. 14. 0.0005. Problems 1.2, pages 11-12 1. 0.0025. 2. 76. 3. 75.6 ; 0.7. 5. 0.23 ; 0.14. 6. 0.7721. 7. 0.423 8. 10. 9. log, (1.2) = 0.1823215; n = 9. . 3 10. os cosh) =2+h +X + O18 seh costh)=1+h—~* + OOF, BP Ba. 11. sin (t) + cos (t)=140-E -O £4, ou) eeu a igientoaie1s0 . 2 2 sin (£) cos (t)=¢- 5 P+ 1s f+ Or. Problems 1.3, page 12 1. (b) 2. 0.000005. 3. (). 4. 0.00049. 5. 0.007. 6. 43.38 ; 0.63264 ; 0.2538. 7. 0.004. 8. 0.0058 9%. (c). 10. 0.0015. 11. 0.0496, 12. 0.33. 13. 0.0005. 14. <1 (px 10), Problems 2.1, page 17 1. xt— 6x3 + Bx? + 42e-70=0 2@-2,143:. (ii) 22 V8,3,-5. 4. a=2,b=1. 5. 1,3, 25. 6. -7, 2,6. #,1,4,7- 9. 473. Problems 2.2, page 21 1. x9 + 6x? - 36 + 27=0. 2x74 85434224 7x-1=0. 3. Ort + 9x? + &?- 7241-0. 4. 2/9, 2/3, - 9/3, 5. 2,3 ,-3,-4; Gi) 2,2,3,2. 674 So.ution Or Aucrpraic AND TRANSCENDENTAL EQUATIONS 1. Introduction 2. Basic properties of equations 3. Transformation of equations 4. Synthetic division ; To diminish the roots of an equation by h 5. Iterative methods 6. Graphical solution of equations 7. Convergence 8. Bisection method 9. Method of false position 10. Secant method 11. Iteration method ; Aitken’s 4? method. 12. Newton-Raphson method 13. Some deductions from Newton-Raphson formula 14, Muller's method 45. Roots of polynomial equations ; 16. Multiple roots 18. Lin-Bairstow's method 20. Comparison of Iterative methods ‘Approximate solution of polynomial equations-Horner’s method 17. Complex roots 19. Graetfe's root squaring method 211. Objective type of questions 2.41, INTRODUCTION "An expression of the form flr) = ay" +02" + vue +O, 1% 40, where a’s are constants (a, # 0) and n is a positive integer, is called a polynomial in x of degree n. The polynomial f(x) = 0is called an algebraic equation of degree n. If ix) contains Some other functions such as trigonometric, logarithmic, exponential ete., then fx) = 0 is called a transcendental equation. Def. The value «of x which satisfies flx) = 0 AD) is called a root of f(x) = 0. Geometrically, a root of (1) is that value of x where the graph of y = fix) crosses the x-axis. ‘The process of finding the roots of an equation is known as the solution of that equa- tion. ‘This is a problem of basic importance in applied mathematics. Iffix) is a quadratic, cubic or a biquadratic expression, algebraic solutions of equations are available, But the need often arises to solve higher degree or transcendental equations for which no direct methods exist. Such equations can best be solved by approximate methods, In this chapter, we shall discuss some numerical methods for the solution of algebraic and transcendental equations. 13 aaa 2.11, Divide 2x5 — 3x4 + 4x? — 5x2 + 6x —9 by x? — x +2 synthetically. Sol. -3 : 2 -1 Hence the quotient = 2x — x? — x — 4 and the remainder = 4x - 1. PROBLEMS 2.2 1, Find the equation whose roots are 3 times the roots of 23 + 2x? dx + te | 2. Change the sign of the roots of the equation x” + rae edoteTet1s0. _ 3. Find the equation whose roots are the negative reciproéals of the roots of x4 + 7x? + Bx? i = 9r+10=0. 4. Solve the equation 81x 18x ~ 36x + 8 = 0, given that its roots are in HLP. 5. Solve : (i) Gx) + x4 — 43x38 — 43x? +x + 6 = 0. (C.S.V.T.U., B. Tech., 2006) Gi) 4xt — 20x8 + 33x? - 20x + 4 = 0. (Madras B.E., 2003) 6. Find the equation whose roots are the roots of: +x?— 3x—x + 2 = 0 each diminished by 3. 7. Show that the equation x* — 10x* + 23x2— 6x — 15 = 0 can be transformed into reciprocal ‘equation by diminishing the roots by 2. Hence solve the equation. 8. Find the equation of squared differences of the roots of the cubic x? + 6x? + 7x +2 = 0. (C.S.V.T.U., B. Tech., 2011) 0, form the equation whose roots 9. Ifa, B, yare the roots of the equation 2x*+ 3 are (1—@)-1, (1-B)? and (1-y. 10, Divide 15x? — 16x° + 30x5 - 3x4 — 5x? — 2x? + 5x + 8 by x? —x + 1 synthetically. 2.5, ITERATIVE METHODS ‘The limitations of analytical methods for the solution of equations have necessitated the use of iterative methods. An iterative method begins with an approximate value of the root which is generally obtained with the help of Intermediate value property of the equation (§ 2.2), This initial approximation is then successively improved iteration by iteration and this process stops when the desired level of accuracy is achieved. The various iterative methods begin their process with one or more initial approximations. Based on the number of initial approximations used, these iterative methods are divided into two categories: Bracketing Methods and Open-end Methods. Bracketing methods begin with two initial approximations which bracket the root. Then the width of this bracket is systematically reduced until the root is reached to desired accuracy. The commonly used methods in this category are : (1) Graphical method (2) Bisection method (3) Method of False position. poem Open-end methods are used on formulae which require a single starting value or tw starting values which do not necessarily bracket the root. The following methods fall undo, this category : (1) Secant method (2) Iteration method (3) Newton-Raphson method (4) Muller's method (5) Horner's method (6) Lin-Bairstow method. 2.6. GRAPHICAL SOLUTION OF EQUATIONS Let the equation be fix) = 0. (i) Find the interval (a, b) in which a root of fix) = 0 lies. ii) Write the equation fix) = 0 as @ (x) = y (x) where w (x) contains only terms in x and the constants. (iid) Draw the graphs of y = 6 (x) and y = y (x) on the same scale and with respect to the same axes, (iv) Read the abscissae of the points of intersection of the curves y = (x) andy = (x). These are the initial approximations to the roots of f(x) = 0. Sometimes it may not be convenient to write the given equation fx) = 0 in the form 6 (x)= (2), In such cases, we proceed as follows : (i) Form a table for the value of x and y = ftx) directly. Gi) Plot these points and pass a smooth curve through them. (iii) Read the abscissae of the points where this curve cuts the x-axis. ‘These are rough approximations to the roots of flx) = 0. M Example 2.12. Find graphically an approximate value of the root of the equation 3-zaer!, Sol. Let fix)=e™4+x-3=0 wld) AD=14+1-3=-ve and f2)=e+2-3=2.718-1=+Vve A root of (i) lies between x = 1 and x = 2. Let us write (i) as e*-! = 3 —x. The abscissa of the point of intersection of the curves yeert ii) and ys8-x lit) will give the required root. To plot (ii), we form the following table of values : 14 A120] 1.3 16 LT 18. 19 20 | | 22 | 1.35 rez | 201 | 223 | 246 | 272 | Taking the origin at (1, 1) and 1 small unit along either axis = 0.02, we plot these | points and pass a smooth curve through them as shown in Fig. 2.2. a ce in any method 3rd method converges faster than the 2nd method. This fastness of convergens is represented by its rate of convergence. Ife be the error then ¢, = 0.~ x; =), —%r : : If ¢;,,/e; is almost constant, convergence is said to be linear i.e, slow. Ife, fe? is nearly constant, convergence is said to be of order p i.e. faster. 2.8. (1) BISECTION METHOD ‘This method is based on the repeated application of th ; Let the function ftx) be continuous between a and 6. For definiteness, let fla) be negative and f (6) be positive. Then the first approxima- tion to the root is x, = y(a +b). Iff(x,) = 0, then x, is a root of f(x) = 0. Otherwise, the root lies between a and x, or x, and b according as f (x,) is positive or nega- tive. Then we bisect the interval as before and continue the process until the root is found to desired accuracy. In the Fig. 2.4, flx,) is + ve, so that the root lies between a and x,. Then the second approximation to the root is x, = 4 (a +x,). If Ff (x,) is — ve, the root lies between x, and x». Then the third approximation to the root is ua at x= 4 (x, + x2) and soon. of the intermediate value property. TiO) fa; | ‘Obs. 1. Since the new interval containing the root, is exactly half the length of the previous Fad 1 one, the interval width is reduced by a factor of > at each step. At the end of the nth step, the new interval will therefore, be of length (b - a/2". If on repeating this process n times, the latest interval is as small as given e, then (b ~ a)/2" (1+ 1.25) = 1.125. ‘Then fix,) = 1.125 sin (1.125) ~ 1 = 1.125 sin (64.46)° — 1 = 0.01509 and f\) < 0. A root lies between 1 and x, = 1.125. ‘Thus the third approximation to the root is (1 + 1.125) = 1.0625. ‘Then fix,) = 1.0626 sin (1.0625) ~1 = 1.0625 sin (60.88) ~ 1 =- 0.0718 < 0 and x2) > 0, i.e. now the root lies between x, = 1.0625 and x, = 1.125. 1 Fourth approximation to the root is x, 2 (1.0625 + 1.125) = 1.09375, Then Ax,) = — 0.02836 < 0 and flx,) > 0, ‘The root: lies between x, = 1.09375 and x, = 1.125. Fifth approximation to the root is x, ‘Then fxs) = — 0.00664 < 0 and flx,) > 0. The root lies between x, = 1.10937 and x, = 1.125. Thus the sixth approximation to the root is ; (1.09375 + 1.125) = 1.10937 += 2.10997 + 1.125) = 1.11719 cae Then fxg) = 0.00421 > 0. But flx;) < 0. The root lies between X, = 1.10937 and x, = 1.11719. 1 Thus the seventh approximation to the root is X= git 10937 + 1.11719) = 1.11328 Hence the desired approximation to the root is 1.11328. Example 2.17. Find the root of the equation cos x = xe* using the bisection method correct to four decimal places. (Mumbai, B. Tech., 2004) Sol. Let f (x) = cos x ~ xe*. Since /(0) = 1 and ft1) = 2.18, s0,a root lies between 0 and 1, First approximation to the root is x, ; (0+D=05 Now f(x,) = 0.05 and f(1) = - 2.18, therefore the root lies between 1 and x. Second approximation to the root is x, a (0.5 +1) =0.75 Now flx,) 0.86 and f(0.5) = 0.05, therefore the root lies between 0.5 and 0.75. Third approximation to the root is x, = e (0.5 + 0.75) = 0.625 Now fiz) = 0.36 and ft0.5) = 0.05, therefore the root lies between 0.5 and 0.625. Fourth approximation to the root is x, . (0.5 + 0.625) = 0.5625 Now flx,) = - 0.14 and 0.5 = 0.05, therefore the root lies between 0.5 and 0.5625 Fifth approximation is x, = 5 (0.5 + 0.5628) = 0.5312 0.04 and f(0.5) = 0.05, therefore the root lies between 0.5 and 0.5312. Now fiz;) seareeeT 1 Z Sixth approximation is x, = 5 (0.5 + 0.5812) = 0.5156 Hence the desired approximation to the root is 0.5156. Example 2.18. Find a positive Sol. Let lx) = x logy, x ~ 1.2. Since f2) = — 0.598 and f3)= real root of x log, x = 1.2 using the bisection method. 0.231, 80 a root lies between 2 and 3. First approximation to the root is x, = + (2 + 3)=2.6. Now ft2.5) ~ 0.205 and f(3) = 0.231, therefore a root lies between 2.5 and 3. Second approximation to the root is x, 3.5 +3) =2.75. Now (2.75) = 0.008 and /12.5) = — 0.205, therefore, a root lies between 2.5 and 2.75. Third approximation to the root is x, = 4 (2.5 + 2.75) = 2.625 Now /t2.625) = ~ 0.1 and /l2.75) = .008, therefore a root lies between 2.625 and 2.75. Fourth approximation to the root is x, = 3 (2.625 + 2.75) = 2.687 Hence the desired root is 2.687. 2.9. (1) METHOD OF FALSE POSITION or REGULA-FALSI METHOD or INTERPOLATION METHOD ‘This is the oldest method of finding the real root of an equation ftx) = 0 and closely resembles the bisection method. Here we choose two points x, and x, such that f(x) and f (x,) are of opposite signs i.e. the graph of y =f (x) crosses the x-axis between these points (Fig. 2.6). This indicates that a root lies between x, and x, and consequently f (xg) f (r,) < 0 Equation of the chord joining the points Alxy, flxy)] and Blx,, flx,)I is a)— flea) ) a Bly foe) . Fig. 2.6 ‘The method consists in replacing the curve AB by means of the chord AB and taking the point of intersection of the chord with the x-axis as an approximation to the root. So the sa of the point where the chord cuts the x-axis (y = 0) is given by fv) wD) x1) ~[(%) which is an approximation to the root. ‘ — If now f and f(x.) are of opposite signs, then the root lies between to and 25 S replacing x, by x» in (1, we obtain the next approximation xy. (The root sopeated til a between x, and x, and we would obtain x, accordingly). This procedure is repeated til the root is found to the desired accuracy. The iteration process based on (1) is known as the method of false position. (2) Rate of Convergence. This method has linear rate of convergence which is faster than that of the bisection method. Example 2.19. Find a real root of the equation x) Qe - 5 = 0 by the mtiod of files Position correct to three decimal places. (Manipal, B.E., 2005) -2x-5 ~1and f(3)= 16, root lies between 2 and 3. "Taking », ft) =~ Lf) = 16, in the method of false position, we get t= Xo 1 . ~ 1 py 22+ + = 2.0588 wl) 0" Fa) flag 180 2 * a7 . Now lx) = £ (2.0588) = ~ 0.3908 ie, the root lies between 2.0588 and 3. Taking xy = 2.0588, x, = 3, flxg) = ~ 0.3908, fix,) = 16, in (i), we get 09412 ; y= 2.0588 ~ FE (0.3908) = 2.0818 Repeating this process, the successive approximations are x, = 2.0862, 2.0941, 2.0943 ete. Hence the root is 2.094 correct to 3 decimal places. @ Example 2.20. Find the root of the equation cos x = xe* using the regula-falsi method correct to four decimal places. (Delhi B. Tech., 2013) Sol. Let f(x) so that FO) sxx —xeT= 0 »f(1) = cos 1 = ie., the root lies between 0 and 1. 2.17798 - Taking x)= 0,x, = 1,f(xq)= 1 and f (x,) = - 2.17798 in the regula-falsi method, we % St Fe) 2 Flxy)~ Fx) (xp) =0+ 317798 * 1 = 0.31467 wl) Now (0.31467) = 0.51987 i, ¢., the root lies between 0.31467 and 1. 31467, x, = 67, x, = 1, Aix) = 0.51987, flx,) = - 2.1798 in (i), we get fe 0.68533 | y= 0.81467 + Fer X 0.51987 = 0.44673 ord a 0.44673) = 0.20356 i.e., the root lies between 0.44673 and 1. 1G Xp = 0.44673, x, = 1, f (x9) = 0.20356, Ax,) = - 2.17798 in (i), we get _ 0.55327 4 = 0.44673 + 57 % 0.20356 = 0.49402 Repeating this process, the successive approximations are 1% = 0.50996, x, = 0.51520, x, = 0.51692 %g = 0.51748, x5 = 0.51767, xy» = 0.51775 ete. Hence the root is 0.5177 correct to 4 decimal places. Wi Example 2.21. Find a real root of the equation x log) x = 1.2 by regula-falsi method correct to four decimal places. (V.T.U., B. Tech., 2010) Sol. Let log yg x — 1.2 so that ve, fl2) =~ ve and f'3) = + ve. A root lies between 2 and 3. Taking x, = 2 and x, = 3, fly position, we get 0.59794 and ftx,) = 0.23136, in the method of false Hq = %q— Faatieo = 2.72102 i) Now lx.) = (2.72102) = - 0.01709 ie. the root lies between 2.72102 and 3. Taking %y = 2.72102, x, = 3,fl%) =~ 0.01709 and ftx,) = 0.23136 in (), we get 0.27898 2.72102 + 995196 + 0.01709 Repeating this process, the successive approximations are 1r, = 2.74024, x, = 2.74068 ete, Hence the root is 2.7406 correct to 4 decimal places. Mm Example 2.22. Use the method of false postion, to find the fourth root of 32 correct to three decimal places. Sol. Let x = (32) so that x4 Take flx) = x4 - 32. Then 2 Taking x) = 2% = 3, fx) x 0.01709 = 2.74021 32=0 16 and f13) = 49, i.e. a root lies between 2 and 3. — 16, flx,) = 49 in the method of false position, we get fal re Pee ey %=%0- Fea) = Flam) Ne) =2 + Ge = 2.2462 Now flx,) = 2.2462) =~ 6.5438 i ‘Taking x, = 2.2462, x, = 3, fly Sa AzABD: |g 335 49+ 65498" ©4538) = 2 the root lies between 2.2462 and 3. 6.5488, flx,) = 49 in (i), we get x, = 2.2462 ~ i... | Now fix,) = (2.335) =— 2.2732 ie. the root lies between 2.335 and 3. 9 + 2.2732 Repeating this process, the successive approximations are x; = 2.3770, x, = 2.3779 ete, Since x, = x, upto 3 decimal places, we take (32) = 2.378. 2:10. (1) SECANTMETHOD This method is an improvement over the method of false position as it does not require the condition f (x,) f (x,) < 0 of that method (Fig. 2.5). Here also the graph of the function y = f (x) is approximated by a secant line but at each eee a recent approximations to the root are used to find the next approximation. Also it is not necessary that the interval must contain the root. Taking xy, joining these as , as the initial limits of the interval, we write the equation of the chord y-flx) (x-x,) _ Flay) - fle) Saal Then the abscissa of the point where it crosses the x-axis (y = 0) is given by %1— Xo 71 0 fx) F(x) — Fl) fe | which is an approximation to the root. The general formula for successive approximations is, therefore, given by %= Xn 7 Xn-1 rr =x, -—— zi 7a Fafa one (2) Rate of Convergence. If at any interation flx,) = flx,_,), this method fails and shows that it does not converge necessarily. This is a drawback of secant method over the method of false position which always converges. But if the secant method once converges. its rate of convergence is 1.6 which is faster than that of the method of false position. @ Example 2.23. Find a root of the equation x* - 2x - 5 = 0 using secant method correct to three decimal places. Sol. Let fix) = x* - 2x - 5 so that fl2) = Taking initial approximations x, Land (3) = 16. and x, = 3, by secant method, we have : ay -2, 2 xy=,-—*1-*0 __ ftz,) = 9- 3-2 16 = 2.058823 a fe=fed * 16+1 Now — flx,) =~ 0.390799 Ax,) = 2.081263 Py and flay) = ~ 0.147204 flxy) = 2.094824 and 3 fix) = 2.0945: fey 4549 Hence the root is 2,094 correct to 3 decimal places. WExample 2.24. Find the root of the equation xe" = cos x using the secant method correct to four decimal places. (U.P.T.C., MCA., 2009) Sol. Let fix) = cos x ~ xe" = 0. ‘Taking the initial approximations xy = 0, x, = 1 80 that f (x9) = 1, f (x,) = cos 1 —e =~ 2.17798 ‘Then by secant method, we have y=, — 2) py 24 A E2.1798) = 0.81467 Foe,)= fa) 2 * 317798 Now [Gx,) = 0.51987 m-% f(x) = 0.44673 and f(xy) 0.20354 a Fay) fa) Fees) = Fm) f (x,) = 0.53171. Repeating this process, the successive approximations are x, = 0.51690, x= 0.51775, xq = 0.51776 ete. Hence the root is 0.5177 correct to 4 decimal places. Obs. Comparing Examples 2.18 and 2.21, we notice that the rate of convergence in secant method: | definitely faster than that of the method of false position. | 2.11. (1) ITERATION METHOD ‘To find the roots of the equation ftx) = 0 i) by successive approximations, we rewrite (i) in the form x = @(x) ‘The roots of (i) are the same as the points of intersection of the straight line y = x and the curve representing y = (x). Fig. 2.6 illustrates the working of the iteration method which provides a spiral solution. Let x=, be an initial approximation of the desired root u. Then the first approximation x, is given by wii) 41-9) Now treating x, as the initial value, the seeond approxi- mation is x, = 0%) Proceeding in this way, the nth approximation is given by %, = 0,4) (2) Sufficient condition for convergence of iterations. Now it is not sure whethe, the sequence of approximations x, Xp. .--»%p, always converges to the same number which jg root of (1) or not. As such, we have to choose the initial approximation x, suitably so that the Suecessive approximations x, xp, ....X, Converge to the root at. The following theorem helps in making the right choice of xp : Mi Theorem. [f (i) « be a root of f (x) = 0 which is equivalent to x = (x), (i I, be any interval containing the point x = a, (ii) | @(@) | <1 for all x in I, then the sequence of approximations xy x) Xy -. initial approximation x, is chosen in I. Proof. Since a is a root of x = 6(x), we have 0 = (a) Ifx, , and x, be 2 successive approximations to a, we have x, = 0(%,_:) : x, — = O64, ) - (0) a) x, will converge to the root « provided the (pa) — O() By mean value theorem, ee = 0 (©) where x, ,<6| ana o-{4} a, by cp 2. d; then the equations (1) are equivalent to the matrix equation AX = D wal) Multiplying both sides of (2) by the inverse matrix A~1, we get ASAK=A4D or IX=A1D fv Aten or X=AD iE an an Ae ‘| ie. = |B, Bp By |x| dp (3) z] Alle, c ¢3} [ds where A,, B,, etc. are the cofactors of a, b,, ete. in the determinant | A |. Hence equating the values of x, y, z to the corresponding elements in the product on the right side of (3) we get the desired solution. Obs. This method fails when A is a singular matrix i.e. | A | = 0. Although this method is quite general, yet it is not suitable for large systems since the evaluation of A~! by cofactors becomes very cumbersome. We shall now explain some methods which can be applied to any number of equations. Example 3.17. Solve the equations 3x + y + 22 = 3; 2x -3y-2=-3:x4+2y4 matrix inversion method. (cf, Example 5 3 1 2) fa bh gy Sol. Here A=|2 -3 -1]=/a, by cy (say) L 2 1) lay 8 op, f] 1 ia fe As) Ta 3 Ss sl = x]d,|=5]-3 1 7|x!-3]= >| Tl CQ G a ie CTs a [3 [4 Hence x = 1, y = 2,2 =-1. (3) Gauss elimination method. In this method, the unknor imi s n . wns are eliminated successively and the system is reduced to an upper triangular system from which the unknowns are found by back substitution. The method is qui i d . Th quite general and is well-adapted for computer operations. Here we shall explain it by considering a system of three e aa for sake of clarity. a Consider the equations artbytez age + byy +032 =dy a ax t+ by tez =d, = aoe — [SoLUTON OF SIMULTANEOUS ALGESRAGEOUATIONS Stop I. To eliminate from second and ti a ; r , ‘Assuming, 20, weep second and third equations. the first equation from the second from the second equation by subtracting (a,Ja,) times by eliminating (ay/a,) times the fi equation. Similarly we eliminate x from the third equation system irst equation from the third equation. We thus, get the new axtbytez=d, a +eyz=d,’ wu) bay + eyz = dy’ ae ts ae a aren the pivotal equation and a, is called the first pivot. Assuming b,’ #0, we ae fr Sera oa i i times the second equation from the thied equation we tana eae eee 4 ax +by+ez=d, bi'y + 6,'2 = dy’ (8) c’2=d," Here the second equation is the pivotal equation and b Step III. To evaluate the unknowns. ‘The values of x, y, z are found from the reduced system (3) by pack substitution. ‘1b,') is the new pivot. Obs. 1. On writing the given equations as [: by ltl is] az bz c2||y|=|d2| ie, AX=D, az & es\(2] [ds this method consists in transforming the coefficient matrix A to upper triangular matrix by elementary row transformations only. ‘Oho. 2, Clearly the method will fail if any one ofthe pivats a,’ or ey” becomes zero. In such cases, we rewrite the equations in a different order 50 that the pivots are nen-2ero. “Oba. 3. Partial and complete pivoting. In the first step, the numerically largost coefficient of x is chosen from all the equations and brought as the first pivot by interchanging the first equation with the equation having the largest coefficient of x, In the ‘second step, the numerically largest coefficient of y is crate from the remaining equations (leaving the first equation) and brought as the second pivot by chose angling the second equation with the equation having the largest coefficient of y. This process is aa ecued till we arrive at the equation with the single variable. This modified procedure is called partial pivoting. Teave are not keen about the elimination of x,y» 2 in @ specified order, then we can choose at each stage the numerically largest coefficient of the entire matrix of coefficients. This requires not only an sage range of equations butalso an interchange ofthe position of the variables. This method of elimination interred complete pivoting. It is more complicated ‘and does not appreciably improve the accuracy. Example 3.18. Apply Gauss elimination method to solve the equations x + 4y 5; s4yn Gan 1878-y-2=% (Mumbai, B. Tech., 2009) Check sum Sol. We have x+dy-z=-5 -1 x+y—Gz=-12 -16 Bx-y-z=4 5 Step I. To eliminate x, operate (ii: and (iti) — Hi): Check sum ~¥y-fe=-7 -15 Sis ~13y +22 =19 8 fe) Step II. To eliminate y, operate (v) - 2 (iv): Check sum 148 sir % — Step III. By back-substitution, we get From (vi) : 2= 18 «2005 7_5(148 From (iv) : 38) rom (iv) 7=3-aa) From (i): x=-5-4(—81),(148)_ 117 _ e479 Vala) a Hence, x = 1.6479, y = 1.1408, z = 2.0845. Note. A useful check is provided by noting the sum of the enelficieats and terms on the right. operating on those numbers as on the equations and checking that the derived equations have the correct sam. 1 4 -Ifx Otherwise: Wehave|1 1 -6|/y|= 3 -1 -1Ilz 14 -1fx] ' o -7 3 lel Laid 13 Operate R,- — R,,|9 es as »3 0 703i ‘Thus, we have z = 148/71 = 2.0845, By = 7 - Gz = 7 - 10.4225 = - 3.4995 ie, y = 5 —4y +z =—5 +4 (1.1408) + 2.0845 Hence x = 1.6479, y = — 1.1408, z = 2.0845. M Example 3.19. Solve 10x - Ty + 32 + 5u= 6, - 6x + By -2 ~ 4u 1.1408 6479 and y+ lle 5x — Gy - 22 + 4u = 7 by Gauss elimination method. CS.V.T.U, B Tech. 2007) Check sum Sol.Wehave 10x-7y +324 5u=6 7 —O — 61+ By-2-4u=5 2 Bx+y4+4241lu=2 2 5x -Qy-22 + 4u=7 5 Step I. To eliminate x, operate («a - (33) a] [aio - 20) [c-Sa): ie 16 a on eee Cheek sum 6 12.2 eal) .2 15.9 olvi) -3.5 endvii) Step II. To eliminate y, operate [wo -3 | ; [os 8 244736842 + 10.315789u = — 6,8157895 will) ~ 2.84210532 + 0.0526315u = 16.447368 i) Step III. To eliminate z, operate [oo -( 3.8y + 0.82 5 3 3\[2/=|7 8 2-4 ilu) (2 Operate R, — 2R,, Ry - 2R,, R,-3R, 1 2 8 —alfe) (10 0-1 -9 1f/y|_|-19 0-5 -4 5]lz|~|-13 0 -4 -13 6][u} |-28 1 2 8 -1fx 10 0-1-9 1lly|_|-19 Operate R;-5Ry,R,-4Ro|q 9 41 Ollz u fo) 0 28) 2 Thus, we have 41z = 82,2 , 46 + Qu = 48, 282 + Qu = 48 i. u -y-9 +u=-19ie,-y-18+1=~-19, y x4 2y+92-u=Wie,x+4+6-1=10, x 21,y=2,2=2,uel ee jordan method. This is a modification of the Gauss elimination method. In this method, elimination of unknowns is performed not in the equations below but in the vquations above also, ultimately reducing the system to a diagonal matrix form i.e. each equation involving only one unknown. From these equations, the unknowns x, y, 2 ean be obtained readily. ‘Thus in this method, the labour of back-substitution for finding the unknowns is saved at the cost of additional calculations. (Anna, B. Tech., 2016) Obs. For a system of 10 equations, the number of multiplications required for Gauss. Jordan method is about 500 whereas for Gauss elimination method we need only 333 multiplications. This shows that though Gauss-Jordan method appears to be easier but requires 50% more operations than the Gauss elimination method. As such Gauss elimination method is preferred for large systems. Example 3.21. Apply Gauss-Jordan method to solve the equations (V.7.U., B.E., 2009) Bx + dy + 52 = 40. Sol. We have x+y+z=9 2x - By +42 = 13 x + dy + 5z = 40 Step I. To eliminate x from (i) and (iii), operate (ii) ~ 2() and (iii) - 30 + xty+z=9 — by +2¢=—5 (0) y+22=18 evi) Step II. To eliminate y from (iv) and (vi), operate (iv) + $ (v) and (vi) + 5 (0): —by+2z=-5 «-Aviii) aa. 12 wedi) 5 ‘Step III. To eliminate z from (vii) and (viii), operate (vii) — 4 (ix) and (viii) — 3 (ix): x —5y =-15 12 z=12 37 Hence the solution is x= 1, y = 8,2=5. X «& Ste 9 Otherwise: Rewriting the equations as|2 -3 4|}y|=|13 S 4 Gils 40. 1 1 Alf 9 Operate R, - 2R,,R,-3R,,}0 -5 2 =|-5 oO 1 4ilz 13 Ae 9 Operate R, += R,,|0 -5 2ilyl=|-5 5 0 0 12/5|]2 12 [ SeLuTION OF simULTANEOUSALSEBANG coUNTONS Operate — R,, 5R, 3 ele 00 12)}|2 60. Operate R, + a Ry 11 fz] [9 1 Operate * Ry 3 18leels 00 ilz 5 10 O}fz] fr Operate R, —-R,—R, 0 1 Ollyfals 00 ajlz} [5 Hence x = 1, y = 3,2 =5. Obs. Here the process of elimination of variables amounts to reducing the given coefficient metric to a diagonal matrix by elementary row transformations only. @ Example 3.22. Solve the equations 10x - 7y + 32 + 5u = 6 ; - 6x + 8y-z-4u=5; 32+ y +42 + lu =2; 5x —9y— 22 + 4u = 7 by Gauss-Jordan method. (cf. Example 3.19) Sol. We have 10x — Ty + 82+ 5u=6 Step I. To ciminate x, operate [9 - (= Sa], [«o-(S)o}, [oo- “Jo ‘ 10x — Ty + 3z + 5u=6 3.8y + 0.82 -u =8.6 (vii) ~-(viii) (aie Step II. To eliminate y, operate [o- (Z ze} {io -(32}o0], [in - (= 2) wo: 10x + 4.4736842z + 3.1578947u = 21.842105 Aix) 3.8y + 0.82—u=8.6 2.44736842 + 10.315789u = — 6.8157895 ~ 234210532 + 0.0526315u = 16.447368 Step II. To eliminate z, operate [io 2 al —————— | (xi) | : 2.4473684 10x — 15.698923u = 34,301075 3.8y ~ 4.3720429u = 10.827957 2.44736842 + 10.315789u = — 6.8157895 9,9247309u = 9.9245975 Step IV. From the last equation u = 1 nearly. Substitution of u = 1 in the above three equations gives x = 5,y = 4,2 =~7. UG) Factorization method". This method is based on the fact that every square ma- trix A can be expressed as the product of a lower triangular matrix and an upper triangular matrix, provided all the principal minors of A are non-singular, i.e. if A = {a,), then a2 ts ay ay 4 a,,#0,| G2 912 |#0,]21 Gee das | #0, ete. @q, Aa2 dy) dye Ose Also such a factorisation if it exists, is unique. Now consider the equations yyy + Oy + yey = Dy Aggy + Ogghy + Opty = by Agi, + Aggy + Aaa y = Oy which can be written as AX = B AV) 1 M2 ay x by where A=|aq1 gq Gog |, X= | X2 | and B= | by 4g, gg yy, Xs 8 Let A=LU, AQ) 1 0 0 Yn Wye Yas where L=|ty 1 OfandU=| 0 Ug, ty 1 be 1 0 0 ugg Then (1) becomes LUX =B AB) __ Writing ; UX=V, — ..(4), (8) becomes | LV=B which is equivalent to the equations 0, = By, Ly, + 0g = Byy Ly, + Lygh + 05 = by Solving these for v,, U,, Us, we know V. Then, (4) becomes Myhy + yay + agg = Uys Upgty + Uggs = Voy Ugg y = Uyr from which xy, x, and x, can be found by back-substitution. eats To compute the matrices L and U, we write (2) as 10 Offun ue ws] fan a2 a, 13 Gi es sess ty bog 1 0 ugg} (a3, agg agg “Another name given to this decomposition is Triangulisation method. Multiplying # : din. es ok ig the matrices on the left and equating corresponding elements from both @u,=a,, uy, =a, i 12=4y yg = Cys (i) lyyu,, =a, 13 = 19 lonley = or ins : a= ay Tyy = Agfa, 5 Uy, = ayy or ly, = yyy, iii) 1 = CG) Dygtyy + thgg = gy OF tpg = gy — 2 ayy ayy Ly = « faittig + Ugg =A, OF Uy, = dy, ~ —*4 ayy ay (Go) Uyrtlis + lguts ays) or yy ola Ea U2 ayy (0) Igyttys + Uggttng + tgs = yg Which gives Ugg. Thus we compute the elements of L and U in the following set order : ( First row of U, (iii) Second row of U, (ii) First column of L, (iv) Second column of L, (v) Third row of U. ‘This procedure can easily be generalised. Obs. This method is superior to Gauss elimination method and is often w linear systems and for fin¢ multiplications for a system the Gauss method. Among the direct methods, factorization me! computers. Example 3.23. Apply factoriz 1 Sol. Let | la Jay so that @R, of U : (i) C, of L = R, of U: (iv) C, of L =: (v) R, of U = Thus ised for the solution of ‘ding the inverse of a matrix. The number of operations involved in terms of ‘of 10 equations by this method is about 110 as compared 333 operations of thod is also preferred as the software for ation method to solve the equations (V.T.U., B. Tech., 2016) Bx + Dy + 72 = 4; Bu + Sy +225; Be + dy +2 =7. 0 Oyfun m2 3] [8 2 7 1 0|| 0 wap us|=]2 3 2 (ie. A), Ig 10 0 ue) 18 4 1 uy Ue Ty, = Igy = 3s Lyyttya + an = 8 Iyyttys + Has = Tyo + [ggttgn = 4 yyttas + Laatlan * Han = 1 0 0) f3 2 7 A=|23 1 O]=/0 B/S -113 1 65 1) [0 0 -86 V, the given system becomes Writing UX = 1 0 O}fy] [4 23 1 Ollv,|=|5 1 65 Il} [7 Solving this system, we have v, 2 guitensd or yet 6 byt Ey ts =7 or at 5 "5 Hence the original system becomes soe) oY len lie 0 58 -113]/y|=|73 0 0 -85 jz} LV. i. 5 ie. Sr +ay+te=4, 3y By back-substitution, we have z=-18,y=9/8 and x= 78. Example 3.24. Solve the equations 10x~ 7y +32 4 5u = 6; - 6x + By-2- du 25; 3x4 y +424 Llu = 2; 5x-9y- 22 + 4u = 7 by Factorization method. (cf. Example 3.19) 190 0 Off, a2 His Mua ly 1 0 Of] 0 typ uo Ma) _|-6 8 -1 4) & Bete | ie 3 0|(0 (a) ee ule 3 ot £1 (ie. A) me i ie ILO © © oes peo a so that (Ry of Ut thy = 10, typ =— Ts tag = Bs tag = 5 (ii) C, of L: = 0.6, Igy = 0-8, Ly, = 0.5 (iii) R, of Us Ugo = (iv) Cy of L: Ig = 0.81579, Iyy = — 1.44787 (v) Ry of U: Ugg = 2.44787, ty = 10.1579 (vi) Cyof LL: lyg =~ 0.95699 (vii) Ry of Uz tyy = 9.92474 1 0 0 10 = 8 5 0 O}| 0 38 08 -1 o|| 0 0 244737 10.31579 ia 8 0 9.92474 S08. 0 Thus A=) 03 0.81579 i 05 - 144737 -0.95699 Writing UX = V, the given system becomes i 0 0 07 fu] [6 -06 1 0 Of _}vg|_|5 0.3 0.81577 1 0] Jos} |2 a 05 - 1.44737 -0.95699 1] |v% Solving this system, we get b, = 6, Vs = 86, 05 =~ 6.81579, v, = 9.92474. Hence the original system becomes 10-7 5 6 0 38 08 -1 86 x 0 0 244737 1031579) | >| =|-6.81579 o 0 0 9.92474 | |u 9.92474 ie. 10x - Ty + 82+ 5u=6, 3.8y + 0.82 -u=8.6, 2.447372 + 10.31579u = ~ 6.81579, u = 1. By back-substitution, we get aeLen | PROBLEMS 3.3 | Solve the following equations by Cramer's rule x+y 4221; 8x+y—82=5;2—2y—52 = 10, (Andhra, B. Tech., 2016) x+y 4+2=66;x-y+2=22; 242432 = 152. x2zHly =e; yale = 04 ; xylet = 1, 2uw — wu + uv = Buvw ; Bow + Qwu + du Bx+2y—2+t=1;x-y—2e4 4t=3; 2x—3y 2-2 =-2; Solve the following equations by matrix inversion method 19uvw ; Gow + Twu — uv = 1Tuvw. ay +24 2020. eee 6. x+2y+B2=4;244y+92=6. 7. jay +82 =6;x+8y +42 =6. (P.T.U., B. Tech., 8. jx—Qy-2=-4;Bxty—42=0. (Bombay, B. Tech., 2005) 9. Ax + Bx, + Bx, — Sx, =— 1; Gry — Gr, + Gry + 120, 10. Inagiven electrical network, the equations for the currents i, iy, iy are Big + i + ig = 8 ; Bi, — Bip — Big = — 5; Ti, + Big Sig Calculate i, and i, by (2) Cramer's rule, (b) matrix inversion, Solve the following equations by Gauss elimination method : Li. bx + dy = 15; e+ Ty = 12 (Anna, B. Tech., 2014) 12, Bx-y + 2e= 1D; x4 By + Be = 11; e—-Yy-z2=2, W.U,, B. Tech. 13. Qx-y4+82=9;xt+yt+z=65;2x-yrz=2. (Bhopal, B.E., 2008) 14, xy + At, +25 = 95 Bx, + 2xy- 2x =—-25x)—% +44=6. (Marathwada, B. Tec 15, xy $y tq $%g= A Gy + Tay Hy tty = 125K, ny + Gey Hy =— 5x, Hy xy Hey 6. Solve the following equations by Gauss-Jordan method : 16. 2x + 5y 4 T2=52;2e+y-z=0;x4y42=9. (W.T.U,, B. Tech., 2013) 17, Wxr+y+z=12; 2c + ly +z=13;x4y+52=7. (Anna, B. Tech., 2016) 18, x+y +z=9; 2+ y—2=0; We + By + 7252. (V.7.U,, B.E., 2009) 19, Qe+y+2=10; Sx + Dy +82=18; x4 4y +92 = 16. (Andhra, B. Tech., 2015) WO, Dey 4 Xy + Bary +2, = 55x, +p — Sty + dey Bx, + Ory — Dr, +2, =8 ; Dry + Dry + Qe — 3x, = 2, ‘Solve the following equations by factorization method : Dh, e+ By 47=95 x4 Dy + 32=6; x+y 42e=8, Qety+4e= 12: &~3y 422-00 p4x4 lly—2 = 33, 2. lrey+2 3 33x + ly +2 = 14; 24 By + 10 15, 1; 2e,-x5+2,= Lyxy +x, 42x, (V.T.U., B. Tech., 2015) 33, +2xy-x, 3.5. _ITERATIVE METHODS OF SOLUTION ‘The Preceding methods of solving simultaneous linear equations are known as direct methods, as these methods yield the solution after a certain amount of fixed computation, On the other hand, an iterative method is that in which we start from an approximation to the {rue solution and obtain better and better approximations from a computation eycle repeated as often as may be necessary for achieving a desired accuracy. Thus in an iterative method, the amount of computation depends on the degree of accuracy required. For large systems, iterative methods may be faster than the direct methods. Even the round-off errors in iterative methods are smaller. In fact, iteration is a self correcting process and any error made at any stage of computation gets automatically corrected in the subse- quent steps. Simple iterative methods can be devised for systems in which the coefficients of the leading diagonal are large as compared to others. We now describe three such methods : YP Jacobi's iteration method. Consider the equations ax + by +ez gx + bey + 69 gx + byy + 3 Ifa, b,, ¢, are large as compared to other coefficients, solve for x, y, z respectively. Then the system can be written as aa (1) ds (d,-b, y-¢z) (dpa, x-c,2) (2) x 25 (dy—ajx—by) Let us start with the initial approximations x9, ¥, 29 for the values of x, y, z respec- tively. Substituting these on the right sides of 2), the first approximations are given by x ra (d, - by, — 25) 1 3, (2-0-0) 1 5 (43 Sato Bay) Substituting the val are as lues x, ¥,, 2, on the right sides of (2), the second approximations 1 Ye By y= ax, 032) 1 Baa ee to trea) This is i i eens process is repeated till the difference between two consecutive approximations is ‘Obs. In the absence of any better estimates for x9, Yo, Zo, these may each be taken as zero, M Example 3.25. Solve, by Jacobi's iteration method, the equations 20x + y—~22 x + 20y - 2 x - Sy + 20z = 25. (Anna, B. Tech., 2016) ‘Sol. We write the given equations in the form 1 X= 9p (IT -y +22) 1 Y= og 18-3t +2) onli) 1 Jp (25- 2x + 3y) We start from an approximation x9 = ¥ =) = 0- Substituting these on the right sides of the equations (i), we get 25 0.9, 2,= 59 = 125 1 M1=- 39 the right sides of the equations (i), we obtain Putting these values on 1 gp (IT 91 + 2) = 102 1 a= Gg © 18- Bx + 2)) = - 0.985 - 295 5p (25- 2m + By) = 1.03 Substituting these values on the right sides of the equations (i), we have 1 y= Gp (17-9 + Bea) = 1.00125 1 5 = Gp 18 Brg Hq) =~ 10015 00825 1 255 Gp (25 2e2 + S90) Substituting these values, we get = 1.0004 1 X4= 9g (IT 95 + 225 1 gp (— 18 Sty +24) = - 1.000025 3p (25 ~ 2xy + By,) = 0.9965 Putting these values, we have : 1 i (-17-y, + 2z,) = 0.999966 3p (— 18 - 3x, +2,) =~ 1.000078 75> z (25 - 2x, + 3y,) = 0.999956 Again substituting these values, we get 1 30 (-17-y, + 2z;) = 1.0000 1 3p © 18 - 8x, +2,) = - 0.999997 1 24 = 5p (25 — 2x, + 3y,) = 0.999992 The values in the 5th and 6th iterations being practically the same, we can stop. Hence the solution is x=ly=-Lz=1 Example 3.26. Solve by Jacobi’s iteration method, the equations 10x + y - 2 = 11.19, x + 10y +2 = 28.08, x+y + 102 = 35.61, correct to two decimal places. (Anna, B.Tech., 2007) Sol. a the given 22 as Fg (MA9-y +29 = 55 F, (28.08 “~ 2,25 7p @5.61+x- =) We oe from an sl %)=Yo 0. First iteration = 3.561 19 x)= N28 1.119, y, = 2808 = 2.08, spi Second iteration x)= + (11.19 -y, +2) = 1.19 2 + (28,08 - x, -2,)= 2.34 a" + (35.61 + x, ~y,) = 3.39 Third iteration 1 y= 7p (UL19 yy +25) = 1.22 =F 02.0 Ya = Zq (28.08 — xp —29) = 2.95 24 = Gh (85.61 + 3-92) = 845 Fourth iteration x= 5 (11.19 -y, +24) = 1.23 90> (28.08 - 5-24) = 2.34 1 ' = Gh (95.61 + 25-95) = 8.45 Fifth iteration X= $ (11.19 -y, +24) = 1.23 Ye= Gh (28.082, -2,) = 2.84 45 > (35.61 +x,-y) = Hence x = 1.23, y = 2.34, 2 = 3.45 @ Example 3.27. Solve the equations 10x— 2x, -x,-2,=3 : = 2x, + 10k, —%y— x4 = 15 =x, — ty + 10x, 2x, =27 =x, —Xp- 2x, + 10x, =— 9, by Gauss-Jacobi iteration method. Sol. Rewriting the given equation as ' nya Gh Gt ante ted mye (15 +e +2544) m= i OT 4m +m, +20) y= 9 4, +, + 2x) ~ 10 We start from an approximation x, = x, First iteration ‘3 x, = 0.3, x = 1.5, xy i EM nec nen nen scene] Second iteration 1 0 (3 +2 (1.5) + 2.7 + (—0.9)] = 0.78 a (15 + 2 (0.3) + 2.7 + (—0.9)] = 1.74 a (27+ 0.3 +15 +2(—0.9)] =2.7 ¥4= Gp 694034154209) =-0.18 Proceeding in this way, we get Third iteration 0.108 0.9, x, = 1.908, x, = 2.916, Fourth iteration — 0.036 .9624, x, = 1.9608, x, = 2.9592, x, Fifth iteration 2%, = 0.9845, x, = 1.9848, x, = 2.9851, x, = - 0.0158 Sixth iteration %, = 0.9939, x, = 1.9938, x, = 2.9938, x, = - 0.006 Seventh iteration x, = 0.9939, x, = 1.9975, x, = 2.9976, x, = — 0.0025 Eighth iteration X, = 0.999, x, = 1.999, x 2.999, x, =- 0.001 Ninth iteration %, = 0.9996, x, = 1.9996, x, = 2.9996, x, = ~ 0.004 Tenth iteration x, = 0.9998, x, = 1.9998, x Hence x, = 1, x, = 2,x,=3,x,=0. *@) Gauss-Seidal iteration method. This is a modification of Jacobi’s method. As before the system of equations : a,x + by +02 =d, a,x + by +0 = dy =) age + bay + C42 =dy 1 is written as x= > (d,-by-c2) 1 1 Y= By G,-a,x—e2) (2) £ 2= 5 (dy—ax— by) Here also we start with the initial approximations Xpy Yor Zo for x, y, z respectively which may each be taken as zero. Substituting y = y,, z= z, in the first of the equations (2), we get 1 ay (ba e180) eo) Noxt nubatituting x = x), y = y, in the third of the equations (2), we obtain 1 21s, (dy age, — by) and 80 on i.¢, a8 Soon as a new approximation for an unknown is found, it is immediately used in the next step, ‘This process of iteration is repeated till the values of x, y, z are obtained to desired dogree of necuracy, Obs. 1. Since the most recent approximations of the unknowns are used while proceeding to the noxt step, the convergence in the Gauss-Seidal method is twice us fast as in Jacobi's method. Obs. 2.Jacobi and Gauss-Scidal methods converge for any choice of the initial approximations if in cach equation of the system, the absolute value of the largest co-efficient is almost equal to or in atleast one equation greater than the sum of the absolute values of all the remaining coefficients. Example 3.28. Apply Gauss-Seidal iteration method to solve the equations 20x + y ~ 22 3x + 20y 2 =~ 18; 2x—3y +202 = 25, (V.T-U,, B. Tech., 2016) (cf. Example 3.25) Sol. We write the given equations in the form Fp ty +20) oD) 1 S = (25 - 2x + By) ii) 20 “ es First iteration Putting y = yo, z = 2p in (, we get Putting x = x,, 2 = 2 in (ii), we have Putting x =x, y =; in (iii), we obtain Second iteration 5 a Putting y = y,,2 = 2; in (i), we get a2 2g (17 ~y, + 2z,) = 1.0025 1 Putting x = x», 2 = 2, in (ii), we obtain %2* 39 (- 18 ~ 8x, +z,) =- 0.9998 = $ (25 — 2x, + By,) = 0.9998 Putting x = xy, ¥ =» in (i), we get Third iteration, we get oy 1 3 a5 (17 = yy + 22) = 1,0000 1 ‘ Yg= Gp © 18— Bey #2) = ~ 1.0000 22 a (25 = 2x, + 3y,) = 1.0000 ‘The values in the 2nd and 3rd iterations being practically the same, we can stop. Hence the solution is x = 1,y=-1,z=1. W Example 3.29. Solve the equations 27x + by — 2 = 85, x+y + Ode = 110, Ox « +22 = 72 by Gauss-Jacobi method and Gauss-Seidel method. (Anna, B.Tech., 2012 Sol. Rewriting the given equations as x= (85-6y +2) fi) (72 - 6x ~ 22) ii) 15 1 io =< = siti) ay (1N0-2-») (@) Gauss-Jacobi’s method We start from an approximation xy = ¥y = 2) = 0 First iteration 110 148, y, = 2 = 48,2, = U1? =2.087 Second iteration y= Fe (85 ~6y, +2,) = 2157 %* + (72 - 6x, - 22,) = 3.269 z)= + (110 -x, ~y,) = 1.890 54 Third iteration 2.492 y= Gy 65-6, +2 Por (72 — 6x, — 22) = 3.685 47 (110 - x, -y,) = 1.937 Fourth iteration = py (85~6y, +25) = 2.401 (12 = Gx, ~ 224) = 9.646 2, = + (110 -x,-y,) = 1.923 54 a) = 1.928 Fifth iteration Xs 1 = Oo (85 - Gy, +24) = 2.432 3.583 Ye= Gp (12 Gr, ~224) ae 4 (110 ~ 2, ~yg) = 1.927 Repeating this process, the successive iterations are. 4 = 2.423, yg = 3.570, 2, = 1.926 Hq = 2.426, yq = 8.574, 2, = 1.926 ty = 2.426, y, = 3.573, zy = 1.926 Hence x = 2.426, y = 3.573, z = 1.926 (6) Gauss-Seidal method First iteration Putting y = y) = 0,2=25=0in(, ase (85 - Gy, +25) = 3.14 Putting x =x,,2=2, in (i), ne ea (72-6, — 22) = 3.541 Putting x =2,,y =, in (iii), ane (110-2, -y,) = 1.913 Second iteration 1 Ya= Gy (12 bt Bey) = 8512 = Fy (110 mp y9) = 1.926 Third iteration m7 Fe 5-6, 42, 426 n= Fy (V2 Gry Bey) = B.573 926 1 257 gq (M0-%,- 9) = ENGINEERING AN sorence | Fourth iteration 1 2.49 (85 = By, + 24) = 2.426 a7 yy + 23) 1 eas -92,) = 3.573 y= yg (2 By ~ Beg) = 3.078 1 ' 245 py (110 x4 —y4) = 1.926. Hence x = 2.426, y = 3.573, 2 = 1.926, x," ‘Obs. We have seen that the convergence ix quite fant in Gauns-Seidal method as compared to Gauss. Jacobi method. ample 3.30. Apply Gauss-Seidal iteration method to solve the equations : 10x, - 2x, Be 2x, + 10x, ~ xy =x, = 15; ~x,—x, + 10x, + 2x, (Bhopal, B.E. 2009) (cf. Example 3.27) Sol. Rewriting the given equations as $0.3 + 0.2x, + 0.1x, + 0.x, x, = 1.5 + 0.2x, + O.1xy + 0.12, 7 + OAxy + 01x, + 0.2%, = 0.9 + O.1x, + Oley + 0.2 a First iteration 0, x, = 0 in (i), we get x, = 0.8 0, x, = 0 in (ii), we obtain x, 1.56, x, = 0 in (iii), we obtain x, = 4s 1.56, x, = 2.886 in (iv), we get x, Second iteration Putting x, = 1.56, Putting x, = 0.8869, x, Putting x, = 0.8869, s Putting x, = 0.8869, = 1.9523, x, = 2.9566 in (iv), we get x 2X5 = 2.886, x, = ~ 0.1368 in (i), we obtain x, = 0.8869 2.886, x, = — 0.1368 in (ii), we obtain x, = 1.9523 1.9523, x, =~ 0.1368 in (i), we have x, = 2.9566 0.0248. Third iteration Putting x, = 1.9523, x, = 2.9566, x, Putting x, = 0.9836, 2.9566, x, Putting x, = 0.9836, x, = 1.9899, x, = Putting x, = 0.9836, x, = 1.9899, x, Fourth iteration. Proceeding as above 2%, = 0,9968, x, = Fifth iteration is x, = 0.9994, x, Sixth iteration is x ~ 0.0248 in (i), we obtain x, = 0.9836 ~ 0.0248 in (ii), we obtain X= 1.9899 — 0.0248 in (iii), we get x, = 2.9924 2.9924 in (iv), we get X 0.0042. 1,982, x, = 2.9987, x, = - 0.0008. 1,9997, x, = 2.9997, x, = - 0.0001, 9999, x, = 1.9999, x, = 2.9999, x, = — 0.0001 Hence the solution is x, = 1,x, = 2, x, = 3, x, (3) Relaxation method*. Consider the equations ax thy tee =d, ‘This method was originally developed by R.V. Southwell in 1936, for application to structural engineoring problems, SOLUTION OF SIMULTANEOUS ALGEBRAIC EQUATIONS yx + by +o =A, Wek by + eye =, . ; A We dofine tho residuals i, R,, I, by the relations R,=d,~ayx~by oe Ry = dy a,x ~ by ~ eye AD) R, age ~ by ~ eye To start with we assume x = y = z= 0 and calculate the initial residuals, Then the residuals are reduced step by step, by giving increments to the variables. For this purpose, we construct the following operation table : BR, aR, oR, feral =a, ifr =a, ay byt ~b, ~by nb, &=1 =e ~ey = 6s We note from the equations (1) that ifx is increased by 1 (keeping y and z constant), R, R, and R, decrease by a,, a,, a, respectively. This is shown in the above table alongwith the effects on the residuals when y and 2 are given unit increments. (The table is the trans- pose of the coefficient matrix). At each step, the numerically largest residual is reduced to almost zero. To reduce a particular residual, the value of the corresponding variable is changed ; e.g, to reduce R, by p, x should be inereased by p/a,. ‘When all the residuals have been reduced to almost zero, the increments in x, y, z are added separately to give the desired solution. Obs. 1. As a check, the computed values of x, y, 2 are substituted in (1) and the residuals are calculated, If these residuals are not all negligible, then there is some mistake and the entire process should bo rechecked. ‘Obs. 2. Relaxation method can be applied successfully only if the diagonal elements of the coeffi- cient matrix dominate the other coefficients in the corresponding row i.e, if in the equations (1) Ja, (21 1+lel 1b, 12lazl+lerl Leg le Lay 1 +1 a1 where > sign should be valid for at least one row. W Example 3.31. Solve, by Relaxation method, the equations 9x —2y #2 =50; x+5y—Sz= 18; —2x + 2y +72 = 19. Sol. The residuals are given by R, = 50-9x + 2-25 R, = 18 -x~ By +32; R, = 19 + 2v-2y Te The relaxation table is ae Rg, R, 205 ie ob S 194 160 43 4 179 97 40 5 as 60 19 t 0 25 15, 4 5 ae 8 9 <2 10 -1 0 0 0 ZSx = 82, Thy = 26, Br = 21. Hence x= 82,y = 26,2 = 21. PROBLEMS 3.4 1. Solve by Jacobi’s method, the equations : 5x—y +2 = 10; 2r+4y=12;x+y+52=—1, starting with the solution (2, 3, 0). 2. Solve by Jacobi’s method the equations : 18x + 5y —8z +u = 18; 2x + Ly +2—4u = 18 ;x—4y + 10z += 295; 2 ty Bz + Qu = 31. 3. Solve the equations 27x + 6y 2 = 85;x +y + 542 = 40; Gr + 15y + 22 = 72 by (a) Jacobi’s method (6) Gauss-Seidal method. (V.D.U., B. Tech., 2015) Solve the following equations by Gauss-Seidal method Ae det Qy tz = 14x + 5y—z= Osx ty + B= 20. (Anna, B. Tech., 2015) 28x + dy —2 = 32; x4 By + 10z = 245; Be + Ty + 42 = 35 (Anna, B. Tech., 2013) lx +y +2 = 12; 2x4 ly +2= 18; 2x + 2y + 102 = 14. (V.L.U., MCA, 2007) Bx + Qy +22 12;x44y + 22=15 5x + By + 52 = 20. (ethi, B. Tech., 2015) Ber — O. Lrg — 0.2 = 7.85 5 O-Lr, + Tay — 0.8rg = 19.8 5 0.82, ~ 0.2x, + 108, = 71.4. (Mumbai, B.E., 2004) pues Solve, by Relaxation method, the following equations 9% 3x4 9y—22 11; 4x + Dy + 182 = 24; 4x — dy + 32 . (Bhopal, B.E., 2002) nxt Wy-22=73—-2-y + 10z=8. (P.T.U., B. Tech., 2001) 10. 10x — 2y — 2 LL. 9 4 By + dz + 100 = 052 - Ty + 82 + 80=0; 2x + 3y—5z + 60=0. 12, Bax +y +2 = 110; 2x + 1Sy + Ge = 725 —x + By + 272 = 85 (Bhopal, B.E., 2003) 3.6. (1) ILL-CONDITIONED EQUATIONS ‘A linear system is said to be ill-conditioned if small changes in the coef ‘equations result in large changes in the values of the unknowns. On the contrary, a system is well.conditioned if small changes in the coefficients of the system also produce small changes in the solution. We often come across ill-conditioned systems in practical applications. Ill- 3.9 Ere EAA pean es "NUMERICAL METHODS IN ENGINEERING AND SCIENCE - Use Newton-Raphson method to solve the equations x = x? + y2, y = x2 — y® correct to two decimals, starting with the approximation (0.8, 0.4). ; Solve the non-linear equations x? — y? = 4, x? + y? = 16 numerically with Xo = 2.828 using Newton-Raphson method. Carry out two iterations. (V.T.U., MCA, 2007) OBJECTIVE TYPE OF QUESTIONS or fill up the blanks in the following questions As soon as a new value of a variable is found by iteration, it is used immediately in the following equations, this method is called (a) Gauss-Jordan method (b) Gauss-Seidal method (c) Jacobi’s method. (d) Relaxation method. The difference between direct and iterative method of solving simultaneous linear equations is In solving simultaneous e quations by Gauss-Jordan method, the coefficient matrix is reduced to .. . matrix, ‘The condition for the convergence of Gauss-Seidal matrix is that in each equation of the system A matrix in which a, 0 for i #j is called .. Solutions of simultaneous non-linear equations can be obtained using (a) Method of iteration (6) Newton-Raphson method (c) None of the above. To which form the coeffici elimination method. ient matrix is transformed when AX = B is solved by Gauss- Guass-Seidal iteration converges only if the coefficient matrix is diagonally dominant. (True or False) What is ‘partial pivoting’ and ‘complete pivoting’ in the solution of linear simultaneous equations. ‘The convergence in Gauss-Seidal method is..... than that in Jacobi’s method : (a) more fast (6) more slow (c) slow (d) equal. (Bhopal, B.E. 2007) By Gauss elimination method, solve x + y = 2 and 2x + 3y = 5. (Anna, B.Tech., 2007) Problems 2.9, page 61 1 1. @). 2 x, 412 Oat Nix,). 3. ‘The convergence is linear i.e, the convergence is of order 1. x1-% 4: 82=%0~ Fea) Flag) 0 1 5. (). 6. Chord AB. Te Xn = BOnt Nie. 8. 1.79. 9. Initial approximation is chosen. sufficiently close to the root, 10. © 11. Newton-Raphson method. 12. 0.657. 13. (c). 14. | o(@) | <1. 15. Ifwe start with a smaller interval for the root. 16. True $7. 21. 18. (2, 3) 19. x, (2—Nx,) 20. if we start with a smaller interval for the root. Problems 3.1, pages 66-67 5.0,-2. 6. 5. Problems 3.2, pages 75-76 & wo woe wow dice 7. (i) 2; (ii) 2; Gii) 3. & 0 l.x=2y24,z=1Lw=3. 4/3 05 oT 4] 1-1 -5|;)-14 -V4 5/4 5 1-13)| 54 V4 -13/4 ao Of 6 2 6.|7/5 1 O/]0 195 -325 35 4v19 1)|[0 0 32719 8. @ Inconsistent (ii) Inconsistent (iii) Consistent, x 9. @A=3,n#10; (ii) #3; 10. 4=1,-9.For4=1, sol. isx =k, y=—h, For A =-9, sol. is x = 3k, y = 9k, z =— 2k. Problems 3.3, pages 87-88 [ a4 4 vey 2. x=12,y=22,2=32, 3. x=y=2 5. x = 19/50, y = — 29/50, z = — 51/50, t = 0. 5,z=5. : 9. i, = 1.5, ig = 2.5. 10. i, = 1.5 ; ig = 2.5. 11. x = 2.478, y = 0.652. 12, 2=3,y 18. x=1,y=2,2=3. 2, 15. x, 2, x,=- 1% 2-2 Oe 19%. x=7,y=9,2=5, 21, x = 35/18, y = 29/18, 2 = 5/18, 23. x=1,y=3,z= 99 problems 3.4, page 56, y = 1.722, z = — 1.055. 2.103, 2 = 2.845. 10. x= 2.57, z = 2.43, Problems 3.5, page 101 1. Ill-conditioned. 2. x=2,y=-1,z=1. Problems 3.6, page 103 1. x=2,y=1;x=-1.683,y = 2.164, 2. 3.x=3,y=4. 4, x=0.7974,y 1.853, y = - 1.927. .4006. 5. x =3.162,y = 6.45. Problems 3.7, page 104 1. &) 2. §3.5 8. Diagonal 4, the absolute value of the largest coefficient is greater than the sum of the absolute values of all the remaining coefficients. 5. diagonal matrix. 6. (b) 7. Upper triangular matrix. vergence of Gauss-! ,-Seidal method is twice as fast as that 8. False. In fact, the rate of con of Gauss-Jacobi method. 9. §3.4(3), 10. (@) u. x= Ly=h Problems 4.1, page 113 1/2 -1/2 1/2 -1 3 [7-8 geese 3.|-12 3 -1 -2 4 pee ea 5/2 -3/2 1/2 fs ¢ 2-2 2 02 -16 ied fg ce 5 [3 5-4 6. 02 04 Ht 08 [ 1-2 3 1 ellis x 206 -17 -24 14 02 -04 © eae An ee eel 6 soz 22 -15 0 05 ae en ® S078 34 11-02 -01 -15 14 12 18 0.429 2429 - 1429 | wi} 3 7-6 -2 ws [208 so 1a] 0143 0.143 ~ 0.143), 9} 9 -3 0 -3 02 ‘ -0857 -3857 2857 “6 -1 38 1 bh

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy