Re: Hydrogen Mass Repartitioning in NAMD

From: Brian Radak (brian.radak_at_gmail.com)
Date: Thu Mar 08 2018 - 09:00:04 CST

I might also add that I've also used ParmEd but with PSF format on the
solute only.

I've done a few benchmarks for energy conservation as well. If you want to
be more conservative, you can do a 1-2-4 RESPA multistep integrator
(bonded, nonbonded, full PME) with a 1 fs timestep. It's quite a bit slower
but the energy conservation is much better. Then again, I've heard
anecdotal reports that actual observables like PMFs don't show much
variance at all, just as reported in the JCTC paper for AMBER.

On Thu, Mar 8, 2018 at 9:53 AM, Norman Geist <norman.geist_at_uni-greifswald.de
> wrote:

> I’m using it without problems with a timestep of 4fs for a long time
> already. I’m using Parmed and amber Parm7 files to apply the HMR. I
> recommend to scale the solute only, not the solvent.
>
>
>
> *Von:* owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] *Im
> Auftrag von *Brian Radak
> *Gesendet:* Donnerstag, 8. März 2018 15:44
> *An:* namd-l <namd-l_at_ks.uiuc.edu>; JC Gumbart <gumbart_at_physics.gatech.edu>
> *Betreff:* Re: namd-l: Hydrogen Mass Repartitioning in NAMD
>
>
>
> One potential gotcha that I have not looked at thoroughly is HMR on water
> molecules - fortunately, I don't think most people recommend this anyway.
>
> I remember digging around the code and seeing that a repartitioned water
> molecule might fail to get flagged as such because the *oxygen* mass
> would be too *low*. This could goof up SETTLE.
>
> Another tricky thing that I've seen people run into is that things like
> the Langevin piston timescales may need to be re-adjusted to your timestep
> if you are going up to 4 fs.
>
> HTH,
>
> BKR
>
>
>
> On Wed, Mar 7, 2018 at 6:31 PM, JC Gumbart <gumbart_at_physics.gatech.edu>
> wrote:
>
> Hi all,
>
> As far as I can tell, there should be no problems with using HMR in NAMD
> as long as the hydrogen masses are no larger than 3. Is that right? I see
> in SimParameters.h the following:
>
> > Bool ignoreMass; // Mass < 3.5 does not indicate hydrogen,
> etc.
>
> suggesting that mass = 3.5 is the common threshold used to catch Hs (also
> seen elsewhere in the source code). So using one of the common approaches
> in which H masses go from 1 -> 3, NAMD should still catch them correctly.
> HMR is not applied to water typically, so I don’t foresee any problems with
> the fact that a reduced-mass oxygen (mass = 12) wouldn’t be recognized as
> such.
>
> I just want to be sure I’m not overlooking anything. I tried it already
> and it ran fine, but there’s always the chance of a subtle error.
>
> Thanks!
> JC
>
>
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:54 CST