home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!keele!nott-cs!ucl-cs!news
- From: P.Samet@cs.ucl.ac.uk (Paul Samet)
- Newsgroups: sci.math
- Subject: Re: Question on numerical integration
- Message-ID: <3082@ucl-cs.uucp>
- Date: 9 Oct 92 18:49:58 GMT
- Sender: news@cs.ucl.ac.uk
- Lines: 20
-
- There are several ways of integrating numerically over an infinite integral;
-
- 1) Truncate the range, because the end s cannot contribute very much (as
- suggested by someone else);
-
- 2) Make a change of variable, to reduce the interval to a finite one, eg if
- you want to go from 0 to oo, put x = exp(-y), so 0 <= y < oo becomes
- 0 <= x <= 1 ;
-
- 3) Use some type of interpolation formula, you get an infinite sum (which may
- not be very different fromfirst method);
-
- 4) Use some sort of gaussian formula, with an appropriate weighting function.
- Look for "Laguerre Integration" for 0 to oo, and "Hermite Integration" for -oo
- to oo in textbooks on numerical methods.
-
- The book "Numerical Integration" by P. J. Davis and P. Rabinowitz, published by
- Blaisdell, 1967, gives a good account.
-
- Paul Samet
-