数值方法课件.pptVIP

First, separate different roots with intervals. x0=x0+h Output [x0, x0+h] y n f(x0)*f(x0+h)0 x0=a Input a,h a=x0, b=x0+h x0 x0+h Move it Bisection Method of Bolzano Locate a root with an interval. Divide the interval with its midpoint. Select one of its sub-intervals as the new location. f a r b c + + - [a,b]-[c,b] Second, reduce interval length to infinite small. f a r b c + + - [a,b]-[a1,b1]-[a2,b2]-…-[an,bn] N iterations are needed to achieve the given precision Matlab Program for Bisection method See Page 59 False Position Method Locate a root with an interval. Divide the interval with the point of intersection between the secant line and the x-axis. Select one of its sub-intervals as the new location. f a r b c + - - [a,b]-[a,c] False Position Method f a r b c1 + - - [a,b]-[a,c] c2 c1-c2-…-r Matlab Program for False Position Method See Page 60 Initial Approximation of the Roots for f(x)=0 f a1 r1 b1 a2 a3 b2 b3 r2 r3 How to find the locations [ak,bk](k=1,2,3)? They are the approximations for roots rk. Initial Approximation of the Roots for f(x)=0 f x1 r1 r2 r3 x2 x3 x4 x5 x6 x7 x13 [a,b]=[xk-1,xk] [a,b]=[xk-1,xk+1] Convergence Criteria of the Roots for f(x)=0 f Errors in x are different ! Criteria 1: Convergence Criteria of the Roots for f(x)=0 f Errors in f are different ! Criteria 2: Convergence Criteria of the Roots for f(x)=0 f Combined Criteria: and Matlab Program for Location Approximation See Page 68 Modify the program to find all the root intervals! Newton-Raphson Method We have learned how to solve equation x=g(x). But, we generally have an equation in a form f(x)=0. For f(x)=0, the bisection method and false position method usually are not sufficient enough. How to transform f(x) to x=g(x) and find better algorithm? From Taylor expansion: f(xk+d)=f(xk) + f’(xk)d +O(|d |2)=0 d=xk+1-xk Making f(xk) + f’(xk)(xk+1-xk)=0, we have xk+1=xk-f(xk)/f’(xk), So, g(x)=x-f(x)/f’(x) Geometric explanation for Newton-Raphson Method x1 x2 x3 Tangent line e