Subject: Re: [ublas] [bindings] Workspaces
From: Karl Meerbergen (karl.meerbergen_at_[hidden])
Date: 2008-12-12 03:12:48
Rutger ter Borg wrote:
> Could it be that the option for 3 workspace arguments is missing? E.g.,
> AFAICT, cgelsd could be called with three workspace arguments (WORK, RWORK,
> IWORK). At the moment, this is not supported through workspace.hpp.
Yesterday, I suggested to add workspace3. No reactions so far. I have
started the implementation in boost.sandbox, but I want agreement with
the ml before continuing.
> Perhaps it's safer to select the number of workspaces by the number of
> workspaces defined in the Fortran code :-).
That is the case for the bindings I am familiar with. It is determined
by the value_type.
> Another question: are there cases where the formulae for the
> minimum_workspace requirements differ for each value_type? (I.e., 4 cases)?
> Or is two formulae the maximum (real vs. complex)? I remember I saw
> something along the lines of template specialisations based on
> real/complex/mixed scenarios, which would make it three, or is this based
> on something else?
Often it is different for real and complex, but not always.
The mixed case you refer to is probably the Geev case where you can pass
real or complex eigenvectors in the real case. But it should not have an
impact on the workspace.
> ublas mailing list