[petsc-users] computing matrix exponentials

[Jose E. Roman]

El 18/01/2014, a las 18:11, amlan barua escribió:

> Hi Jose,
> I could run the slepc matrix exponential but I would like to clarify one thing with you. Suppose I need to compute exp(At)*v where the vector v keeps on changing but exp(At) is unchanged and therefore should be computed only once at the beginning of the problem. Will slepc do this or during each iteration it will compute exp(At) fresh?
> Amlan 
> 

Please do Reply-all so that the answer goes to the list. 

The current method builds a Krylov subspace with v as starting vector, so a new v implies rebuilding the subspace.

Jose