comp.lang.idl-pvwave archive: archive » Re: Inverting banded-block matrices.

Home » Public Forums » archive » Re: Inverting banded-block matrices.

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Re: Inverting banded-block matrices. [message #36304 is a reply to message #36291]

Fri, 29 August 2003 08:32

Liam E. Gumley
Messages: 378
Registered: January 2000

Senior Member

"James Kuyper" <kuyper@saicmodis.com> wrote in message
news:3F4E8B4D.B90EDAA8@saicmodis.com...
> I've got a problem where I have to calculate g = C D^-1 f, where g and f
> are vectors, and C and D are matrices. C has m by m blocks, each of
> which is itself an n by n matrix. It is banded, with k non-zero
> co-diagonals above and below the main diagnal, both at the block level
> and within each block. C[i,j] ge 0. Every statement I've made about C
> also applies to D.
> For the sake of definiteness, m=10, n=1354, k = 3.
>
> This seems like it should be a pretty common type of matrix structure
> for problems involving 2-D grids. I could solve this by explicitly
> inverting a m*n by m*n matrix. However, I would assume that there are
> existing routines somewhere which can take good advantage of the
> sparseness of these matrices to speed up the calculations considerably.
> Could anyone point me at such routines?

I believe LAPACK routines were incorporated in IDL 5.6, however I have not
tried them.

Cheers,
Liam.
Practical IDL Programming
http://www.gumley.com/

Report message to a moderator

[Message index]

		Re: Inverting banded-block matrices. By: the_cacc on Fri, 29 August 2003 17:30
		Re: Inverting banded-block matrices. By: Liam E. Gumley on Fri, 29 August 2003 08:32

Previous Topic:	Re: map_patch and triangulating satellite images
Next Topic:	Division in a conditional statement

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Mon Oct 27 15:19:48 PDT 2025

Total time taken to generate the page: 2.56112 seconds