3 2 915b 1 2 b i i i i 916a where l0

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: 9.1.6), we nd 2A3 3 6 r( ) 7 7 6 r( 11 12) 7 7 76 6 7 6 r( 2 1 ) 7 7 76 7 = 6 r( 2 2 ) 7 : 76.7 7 6 .. 7 7 76 7 6 r( ) 7 5 6 17 6 r( ) 7 4 27 5 2c 2L 3 6 d0 6 L01 R1 7 6 c10 6 .. 76 6 7 6 d1 .. .. 6 76 . 6 7 6 .. 6 4 L R 56 4 R +1 6 c N N N (9.1.6b) N (9.1.7) N d N N B Thus, the 2(N + 1) coe cients c , d , i = 0 1 : : : N , are determined as the solution of a block bidiagonal matrix of dimension 2N with 2 2 blocks. This system may be solved by the methods of Section 8.5. We can now ask if there is an optimal placement of the two collocation points on each subinterval that, e.g., minimizes the discretization error y(x) ; Y (x) in some norm. This question was answered in a landmark paper by de Boor and Swartz 3] and we will follow their analysis. Consider the inner product i i (v u) = Z b v(x)u(x)dx: (9.1.8) a Let us assume that u(x) and v(x) are smooth except, perhaps for jump discontinuities at z , j = 1 2 : : : M ; 1. Also let z0 = a and z = b. Consider j (v Lu) = Z M b v u + pu + qu]dx 00 (9.1.9) 0 a where the integral is interpreted as a...
View Full Document

This document was uploaded on 03/16/2014 for the course CSCI 6820 at Rensselaer Polytechnic Institute.

Ask a homework question - tutors are online