[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: sparse is_symmetric functions
From: |
David Bateman |
Subject: |
Re: sparse is_symmetric functions |
Date: |
Wed, 06 Dec 2006 21:19:17 +0100 |
User-agent: |
Thunderbird 1.5.0.7 (X11/20060921) |
John W. Eaton wrote:
> On 4-Dec-2006, David Bateman wrote:
>
> | While starting to write svds I found that I need to check if a matrix
> | passed to eigs is symmetric, rather than symmetric with a positive
> | diagonal as the cheap test for positive definite matrices does. At the
> | moment eigs uses arpack's dseupd only for real positive definite
> | matrices, where in fact it could use and should use it for the case of
> | [sparse(size(A,1),size(A,1)),A;A',sparse(size(A,2),size(A,2))] as
> | created by svds. I therefore need the is_symmetric function of the
> | Matrix and SparseMatrix classes.
> |
> | However, the current sparse is_symmetric is very inefficient. The
> | attached patch makes this method of the Sparse classes much better.
>
> OK, please check in these changes.
>
> Thanks,
>
> jwe
>
Done.
D.