Date: Thu, 7 Jun 2001 12:52:26 -0400
Reply-To: Michael Palmer <email@example.com>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Michael Palmer <mcpalmer@ZBI.NET>
Subject: Re: matrix inverse without IML
Content-Type: text/plain; charset="iso-8859-1"
>Basically my bigest matrix is a (12,12) multidimensional array and I have
>to calculate its inverse. By hand this means a system of 144 equations with
>144 unknown terms done couple of hundred times...(because I have plenty of
>these systems <:o(((.It is possible to do this without IML.
Assuming that you're dealing with symmetric matrices, there're are probably
lots of ways. "Statistical Principles in Experimental Design" by B. J.
Winer, 2nd edition discusses the Dwyer and SWP algorithms in Section 2.5,
page 120. I once programmed the SWP in an HP calculator so you can certainly
do it without IML.
Zurich Biostatistics, Inc.