| Re: [eigen] Problem with LU and Cholesky inversion |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
Gael, you beat me to it, I too experimented with replacing A.transpose() by A
and making A be nxm instead of mxn, it works.
Did you forget a nestByValue() ?
In any case thanks for investigating this :)
And many thanks to Timothy for the bug reports!
Benoit
On Friday 10 October 2008 14:51:28 Gael Guennebaud wrote:
> Hi Timothy,
>
> thanks for finding another bug. In fact the bug you hit was marked as a
> FIXME in TriangularSolver, so I know what have to be done, and I'll do it
> as soon as I have time.
>
> in your example, if you first evaluate A to:
>
> DMatrixType B = A.transpose();
>
> and call
> V2 = hessianChol.solve(B);
>
> then it is OK.
>
> cheers,
> gael.
>
> On Thu, Oct 9, 2008 at 6:44 PM, Timothy Hunter <tjhunter@xxxxxxxxxxxx>wrote:
> > Hello list,
> >
> > after running into numerical trouble when implementing a convex
> > optimizer, I wrote this small test (see attachment). It inverses a
> > positive semidefinite matrix using all the methods available in eigen.
> > Can you tell me if I am doing a mistake here or if it exposes a bug in LU
> > and Cholesky modules?
> >
> > Thanks in advance
> >
> > Tim Hunter
> >
> > --
> >
> > Timothy Hunter
> >
> > Student (Stanford University)
> >
> > T. 404 421 3075
---