dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Thu, Nov 25, 2010 at 06:44:36PM +0100, Benjamin Kehlet wrote:
> On 25 November 2010 18:34, Anders Logg <logg@xxxxxxxxx> wrote:
> > On Thu, Nov 25, 2010 at 05:07:40PM +0000, Garth N. Wells wrote:
> >> There is a dependency on
> >>
> >>    dolfin/ode/SORSolver.h
> >>
> >> in dolfin/common/real.cpp. This seems odd to me. Is there a reason for it?
> >>
> >> Garth
> >
> > real.h/cpp provide some basic algorithms that work on high precision
> > floats. One of these is the matrix exponential which is computed by a
> > rational approximation, hence the need for solving a linear system.
> >
> > The SORSolver is the only solver in DOLFIN that handles high precision
> > floats (uBLAS etc can't be used). It would be more natural to put it
> > under dolfin/la, although it would still be a bit non-optimal to have
> > a dependency from common to la. Any other suggestion?
>
> Should the implementation of the matrix exponential be put somewhere else?
> (I mean, if common/real.cpp is reserved for the most basic
> functionality on the high precision floats)
> In dolfin/la together with the SORSolver?

Yes, perhaps that is the best solution.

--
Anders