[SciPy-Dev] Strange results by eigs

Jacob VanderPlas vanderplas@astro.washington....
Fri Oct 21 12:50:13 CDT 2011

Ah, thanks Nils.  I missed that detail when I wrote the wrapper.
It's all a bit confusing, because each wrapper calls one of six arpack 
routines, depending on the type of the input.
Looking at the code, I think everything should work as expected with 
hermitian input for M, but the docs need to be updated, as Pauli suggested.
The bit on M in both eigs() and eigsh() should specify that if the 
matrix A is complex hermitian, then M may be complex hermitian as well.  
It still must be real symmetric in any other case.

Nils Wagner wrote:
> On Fri, 21 Oct 2011 10:12:38 -0700
>   Jacob VanderPlas <vanderplas@astro.washington.edu> 
> wrote:
>>>> The docstring of eigs is misleading in that context.
>>>>        M must represent a real symmetric matrix.
>>> Yep, it should say hermitian instead of real sym. here.
>> Actually, the piece of the documentation quoted here has 
>> to do with the second matrix supplied for the general 
>> eigenvalue problem.  This matrix (M) must be real 
>> symmetric, as desribed in the ARPACK function 
>> documentation
>> http://www.caam.rice.edu/software/ARPACK/UG/node137.html
>>   Jake
> The Hermitian case is here
> http://www.caam.rice.edu/software/ARPACK/UG/node138.html
> Nils
> _______________________________________________
> SciPy-Dev mailing list
> SciPy-Dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-dev

More information about the SciPy-Dev mailing list