[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [igraph] Trouble with arpack
From: |
Gabor Csardi |
Subject: |
Re: [igraph] Trouble with arpack |
Date: |
Thu, 17 Apr 2008 08:59:25 +0200 |
User-agent: |
Mutt/1.5.13 (2006-08-11) |
David,
you're right, this is a bug. I don't know how much testing
will be done before the next release, but i'll correct the classic
eigenvalue solver for sure (i.e. the mode=1 case). It seems to be
a small bug, we don't copy the output of the internal ARPACK
solver correctly if the eigenvalue is complex.
Thanks,
G.
ps. if you don't write from the address you used to sign up,
then your mails will be delayed.... you can sign up for both
addresses and set one of them to "nomail", then you won't
receive messages from the list twice...
On Wed, Apr 16, 2008 at 06:17:47PM +0200, David Morton wrote:
> Hi,
> I ran into trouble using arpack
> with non-symmetric matrices. E.g.
>
> A <- matrix(runif(16,-1,1),4)
> fct <- function(x,extra=NULL)A %*% x
> arpack(fct, options=list(n=ncol(A), nev=1, ncv=3), sym=FALSE)
>
> generaly outputs eigenvectors that don't solve the aigenvalue equation Av=lv,
> where l seems to correspond to the results obtained with eigen().
> Is a full testing of arpack() planned in a near future?
>
> Anyway, igraph is just a great piece of software!
> Bravo!
>
>
> --
> David Morton de Lachapelle
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
--
Csardi Gabor <address@hidden> UNIL DGM