home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.fortran
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cs.utexas.edu!sun-barr!ames!news.dfrf.nasa.gov!altair.dfrf.nasa.gov!maine
- From: maine@altair.dfrf.nasa.gov (Richard Maine)
- Subject: Re: inverse matrix
- In-Reply-To: pfennige@scsun.unige.ch's message of Fri, 8 Jan 1993 13:46:12 GMT
- Message-ID: <MAINE.93Jan8090104@altair.dfrf.nasa.gov>
- Sender: news@news.dfrf.nasa.gov (Usenet news)
- Organization: NASA Dryden, Edwards, Cal.
- References: <C0I49C.Jrr@athena.cs.uga.edu> <1993Jan8.100315.5851@edf.fr>
- <1993Jan8.134612.18468@news.unige.ch>
- Date: Fri, 8 Jan 1993 17:01:10 GMT
- Lines: 30
-
- On Fri, 8 Jan 1993 13:46:12 GMT, pfennige@scsun.unige.ch (PFENNIGER Daniel) said:
-
-
- PFENNIGER> Is it true that a Fortran-90 programme inverting matrix A would be:
- PFENNIGER> B = 1./A
-
- No. That would be a legal statement, but it makes each element of B
- the inverse of the corresponding element of A. The f90 arrays are
- better thought of as "arrays" than as "matrices". In particular,
- all the operations are element-by-element. There is a standard
- matmul subroutine for the usual matrix multiplication, but there is
- no matrix inversion.
-
- I wasn't on the committee at the time, but I believe the rationale was
- that there was no single matrix inversion technique good for all
- situations, and that numerically-oriented users were better off
- choosing a particular, appropriate one than whatever the vendor
- happened to implement (which might be different on different
- platforms).
-
- You (or a library writer) can make a "matrix" (or whatever you want to
- call it) user-defined type and define division for it such that the
- above sample line would do the matrix inversion, but such a type
- isn't part of the base language.
-
- --
- --
- Richard Maine
- maine@altair.dfrf.nasa.gov
-
-