Re: liboctave matrix inverse function inaccurate

[Muthiah Annamalai]

On 10/16/07, Alex Roper <address@hidden> wrote:
> I'm attempting to use liboctave to do matrix math for me in my CS171 computer
> graphics class here at Caltech. Everything works smoothly except for matrix
> inversion, which returns a matrix of correct dimension full of 0's. I believe
> it to be tied to some sort of precision issue, but other matrix operations
> seem to work perfectly.
>
> Inverting the identity matrix succeeds, as does (occasionally) inverting
> matrices that have close to integral results.
>
> I understand this may not be an issue with Octave itself, given the same
> matrix inversion succeeds when using Octave interactively, but since other
> matrix operations succeed I'm not sure. I'd appreciate any advice people can
> give.
>
<SNIP>

I may be wrong, but looks to me if the matrices are ill conditioned
for inversion.
Did you check for return values from Octave inverse function, for its success?
-Muthu