Re: [eigen] banded matrices in Eigen

[Christoph Hertzberg <chtz@xxxxxxxxxxxxxxxxxxxxxxxx>]

On 13.02.2014 12:26, Laura Flores wrote:
> If I could create my own class, derived from one of the existing ones
> (possibly the Diagonal one?), I could add some methods that suit the
> problem I am working on. Therefore, my question would be whether there are
> any suggestions
> for that.

Looking at the current implementation of ConjugateGradient, it would 
theoretically be sufficient to write a class which efficiently implements:
  VectorXd res = rhs - mat * x;
and
  tmp.noalias() = mat * p;

Basically, you need to make
  YourClass::operator*(const VectorType& v);
return something similar to GeneralProduct.

The same could be done for the preconditioner and
  YourPreconditioner::solve(...)

I'm afraid neither is entirely trivial, nor guaranteed to be future 
compatible.

Christoph


--
----------------------------------------------
Dipl.-Inf., Dipl.-Math. Christoph Hertzberg
Cartesium 0.049
Universität Bremen
Enrique-Schmidt-Straße 5
28359 Bremen

Tel: +49 (421) 218-64252
----------------------------------------------