Re: [Help-gsl] gsl_linalg_SV_decomp for N=1

[Brian Gough]

Robert L. RAKSI writes:
 > Should gsl_linalg_SV_decomp work for a M x 1 input matrix? I get the
 > "gsl: subvector_source.c:28: ERROR: vector length n must be positive
 > integer" error message. I looked into the source file and there is a
 > line that looks like this: 
 > gsl_vector_view f = gsl_vector_subvector(work, 0, K - 1);
 > The problem is obiously the K-1.
 > Does the SVD decomposition have some special case for N=1?

It's just the same vector, normalized.
Would be reasonable to add it for completeness though--I've made
a note to do that for the next release.

-- 
Brian Gough

Network Theory Ltd,
Commercial support for GSL --- http://www.network-theory.co.uk/gsl/