home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!sun-barr!ames!think.com!news!rlk
- From: rlk@underprize.think.com (Robert Krawitz)
- Newsgroups: comp.sys.super
- Subject: Re: CMSSL routine for solving 2D finite difference?
- Date: 10 Sep 92 15:30:58
- Organization: Thinking Machines Corporation, Cambridge Mass., USA
- Lines: 26
- Distribution: comp.sys.super
- Message-ID: <RLK.92Sep10153058@underprize.think.com>
- References: <JAMARTIN.92Sep5220655@gauss.nmsu.edu>
- NNTP-Posting-Host: underprize.think.com
- In-reply-to: jamartin@nmsu.edu's message of Sun, 6 Sep 1992 05:06:55 GMT
-
- In article <JAMARTIN.92Sep5220655@gauss.nmsu.edu> jamartin@nmsu.edu (Jay Martinez) writes:
-
- I am trying to do a two-dimensional finite difference formulation
- on a CM-200. For the resulting system of equations Ax=B, I get an
- A maxtrix that looks like this:
-
- Lennart Johnsson (Director of Computational Sciences) made the following
- comments:
-
- To: Robert L Krawitz <rlk@Think.COM>
- Subject: Re: CMSSL routine for solving 2D finite difference?
- Date: Tue, 08 Sep 92 22:35:22 EDT
- From: Lennart Johnsson <johnsson@Think.COM>
-
- They can solve it iteratively using the grid-sparse matrix-vector
- routine, or embed it in a block tridiagonal system (not necessarily
- very efficient)
-
- The performance of an iterative solver should be further enhanced
- by the Stencil compiler in 3.1.
- --
- ames >>>>>>>>> | Robert Krawitz <rlk@think.com> 245 First St.
- bloom-beacon > |think!rlk Cambridge, MA 02142
- harvard >>>>>> . Thinking Machines Corp. (617)234-2116
-
- Member of the League for Programming Freedom -- write league@prep.ai.mit.edu
-