Some methods have error expansions that only contain

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: successive column of the extrapolation tableau. If the trapezoidal rule is solved exactly at each step, then it has an expansion in powers of 2 ( = 2). Extrapolation with rational functions is also possible ( 5], Section II.9). The basic idea is to approximate ( ) by a rational function ( ) and then evaluate (0). Bulirsch i and Stoer 1] derived a scheme where q ( ) is de ned as the rational approximation that interpolates ( ) at = i i+1 i+q for = i i+1 i+q . The values i i of q = q (0) can be obtained from the following recursion h zth Rh R zth R h h h ::: R h h h h >h > ::: > h R i R;1 R i q = i+1 q 1+ R ; =0 i R0 =( i) (4.1.8a) zth i+1 i q 1 ; Rq 1 i i i i i ( hh+q )2 1 ; Rq+11 ; Rm 1 Rq+11 ; Rq+22 ] ; 1 i R ; 7 ; ; ; ; ; (4.1.8b) when the method has an error expansion in powers of 2. The computation of cording to (4.1.8b) is equivalent to interpolating ( ) by a function of the form h i j (h) = R ( i q R ac- zth a0 +a1 h2 +:::+aj hj b0 +b1 h2 +:::+bj hj a0 +a1 h2 +:::+aj;1 hj;1 b0 +b1 h2 +:::+bj;1 hj;1 if is even if is odd j j : (4.1.8c) There are several enhancements to the basic extrapolation approach. Some of these follow. 1. A variable-order extrapolation algorithm could choose the order (i.e., the number of columns in the tableau) adaptively. Error estimates c...
View Full Document

Ask a homework question - tutors are online