[gmx-developers] stuff in double precision

[David Mobley]

Erik,

> > OK, this makes sense, I think -- although the other thing to think
> > about is that Erik was (is?) working on the ability to evaluate and
> > output the potential energy of a particular snapshot at several
> > different lambda values from one simulation to allow the use of
> > methods like WHAM and BAR for analysis of free energy calculations. We
> > want to make sure that we're able to maintain precision so that we can
> > subtract the potential energies from these after they are output and
> > still have sub-kJ/mol accuracies.
>
> I think (read: hope) that could solve the problem in an easier way
> too, since we'll be adding the free energy contributions separately
> and then merging them with the total energy. In any case it is
> trivial and not performance-sensitive to do that summation outside
> the nonbonded loop in double (it's coming along :-)

OK, that's great -- sounds like that would be perfect, since we don't
really care about having the total potential energy to that level of
accuracy, just the DIFFERENCE in the total potential energy between
different lambda values.

David

>
> Cheers,
>
> Erik
> _______________________________________________
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.
>