Re: help about ARPACK solver

andy2O wrote:

- As long as the vectors are parallel then both sets of eigenvectors
can be equally correct. I don't think you can say one is 'better' than
the other in any sense!! Do you have any reason to prefer one to the
other?

As has been written the eigenvectors are undefined up to a scalar complex
factor. Therefore if one set of eigenvectors is 'better' than another
parallel one the reason for that must come from the application and has
nothing to do with the mathematical problem.
In fact it makes sense to always apply some normalization after solving the
numerical problem, then you can compare solutions from different solvers
easily. This normalization could be a pure mathematical one (i.e. scale the
eigenvectors to unity norm), or it could come from the application (e.g. if
the eigenvectors are the modal currents of a multiconductor transmission
line you could make the per mode transported power equal to 1).


Georg
.