Re: [AMBER] entropy calculation starts from where it stops

From: <psu4.uic.edu>
Date: Tue, 25 Jun 2013 15:00:23 -0500

Dear Jason,

    Thanks for the detailed explanation! Yes, most of the articles use
around 40-48 frames for normal nodes. We try to increase the number of
entropy frames (larger than 40-48 frames) to see if the entropy & dG
results will be improved significantly. Since the entropy frames are
minimized and increase the number of mmP(G)BSA frames doesn't improve the
dG (dH) prediction results significantly, using even more entropy frames
might not bring significantly better dG (&entropy) prediction as well.

    Cheers,
    Henry

On Tue, Jun 25, 2013 at 1:59 PM, Jason Swails <jason.swails.gmail.com>wrote:

> On Tue, Jun 25, 2013 at 2:09 PM, <psu4.uic.edu> wrote:
>
> > Dear Jason,
> >
> > Thanks. Wonder if the restart function will come in the next version
> of
> > MMPBSA.py / AmberTools? Just curious.
>
>
> This is unlikely. Checkpoints for these simulations would be incredibly
> complex (especially in parallel!). Unlike MD or QM simulations, MM/PBSA
> calculations are not well-ordered, so you can't simply store a single
> restart file that captures the 'current' state in a checkpoint file. You
> would need to figure out how many threads were running, how far _each_ one
> had gotten in the 'last' calculation, redistribute the remaining work
> across available threads, etc. Furthermore, normal mode calculations
> currently performs the minimization and normal mode calculations in one
> shot without printing any information (like minimized snapshots), so this
> code would largely need to be rewritten to really make checkpoints
> worthwhile.
>
> Given that MM/PBSA calculations are intended to be 'quick' post-processing
> analyses, the importance of having restart capabilities is somewhat
> limited. These two factors provide a significant impetus to implementing
> such functionality.
>
> Sorry...
>
> FWIW, I don't think you need anywhere near 100 frames for normal mode
> calculations, especially since the structures are all minimized to the
> nearest local minimum.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jun 25 2013 - 13:30:02 PDT
Custom Search