[R] Generalized singular value decomposition

[Oana Tomescu]

Am 30.11.2011 16:13, schrieb Berend Hasselman:
>
> tom wrote
>
> I don't think you have to install "something".
> You have not told us how you are trying to load the  dll's.
>
> You can use the GSVD function from
> http://r.789695.n4.nabble.com/problem-with-generalized-singular-value-decomposition-using-LAPACK-td834930.html
> but you will need the exact path to the Lapack dll in the dyn.load call.
> Something like dyn.load("C:\\R\\.....\\libRlapack.dll").
This was just what I was looking for, thank you very much! When I read 
the post before I did not notice the line with the dyn.load.

> You do not need to dyn.load the Blas dll.
>
> I have tried this GSVD function with Mac OS X with success.
I tried the GSVD with the folowing matrices:
A = matrix(rnorm(100,100,20),10,10)
B = matrix(rnorm(200,10,3),20,10)
res = GSVD(A,B)

The resulted U and V have all entries equal to zero. This seems a little 
wired since A=U*E1*Q' and B=V*E2*Q'. If U and V are zero then A and B 
should also be zero. Are my example matrices a special case for the 
decomposition? I just started working with it, so I am a completely 
newbie ...

Thank you again very much for your help! (I was already thinking about 
doing the decomposition with Matlab and loading the result into R.)

Regards,
Oana

>
> Berend
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Generalized-singular-value-decomposition-tp4122542p4123335.html
> Sent from the R help mailing list archive at Nabble.com.
>