[comp.archives] [sci.math] Re: Help: Transposing large m by n matrix in place

kerr@ux1.cso.uiuc.edu (Stan Kerr) (06/14/91)

Archive-name: math/num-analysis/toms/1991-06-11
Archive: ux1.cso.uiuc.edu:/math/toms/toms.ind [128.174.5.59]
Original-posting-by: kerr@ux1.cso.uiuc.edu (Stan Kerr)
Original-subject: Re: Help: Transposing large m by n matrix in place
Reposted-by: emv@msen.com (Edward Vielmetti, MSEN)


larrym@ccwf.cc.utexas.edu (Laurence Macgregor Michaels) writes:

>I would like to transpose a m by n matrix to a n by m matrix in place.
>I would usually just use a seperate array to do this, but m and n are
>large enough that I can't (around 2000 by 3000).

>    Larry Michaels

>larrym@ccwf.cc.utexas.edu

See algorithm 513 from Transactions on Math Software (Jan 1977). You
can get it by anonymous ftp from ux1.cso.uiuc.edu, in file
math/toms/alg513.Z (Unix compressed format). There was also an algorithm
467 published in Communications of the ACM before TOMS began publication.
A remark in the Dec/79 issue of TOMS states that alg 467 is actually
the better of the two. I can fax you a listing of 467 if you wish; we 
don't have it in machine readable form.

   Stan Kerr
   Computing Services Office
   U of Illinois / Urbana
   stankerr@uiuc.edu

-- comp.archives file verification
ux1.cso.uiuc.edu
-rw-r--r--  1 kerr     wheel       31790 May  3 13:27 /math/toms/toms.ind
found toms ok
ux1.cso.uiuc.edu:/math/toms/toms.ind