Re: How to replace NAMD Eletrostatic Forces

From: Brian Radak (bradak_at_anl.gov)
Date: Thu Aug 04 2016 - 15:17:17 CDT

1.) I'm not sure you can turn off the computation of electrostatics
completely, but using "PME off" should bypass a lot of the computational
cost. You can turn off the effect of electrostatics in a number of ways,
for example by zeroing the charges or just using "nonbondedscaling 0",
but the latter also turns of LJ interactions I believe.

2.) TclForces is a good sandbox for trying new models, but I wouldn't
necessarily recommend it for production runs of something like brute
force protein folding.

If you are doing pairwise potentials, you can also try using the
tabulated non-bonded interactions, but these might not be well-typed for
what you want.

Unfortunately (fortunately?) the main purpose of NAMD is exactly to
accelerate the electrostatics, so those routines are a pretty essential
and central part of normal execution. You'd have to tell use more about
what kinds of models you want to use to get more specific advice.

HTH,
Brian

On 08/04/2016 01:15 PM, Emerson Macedo wrote:
> Hello NAMD experts!!
>
> I am new with NAMD and currently working on
> adapt it to our research with proteing folding simulations.
>
> My current challenge involves the possibility of using our
> implementation of electrostatic force calculations instead of NAMD's
> version.
>
> My first thought was (naive I guess) find a way to
> disable NAMD version and then add out implementation.
>
> Here my questions:
>
> 1) How does one could disable NAMD electrostatic calculations or
> replace them?
>
> 2) Does it make sense to use TCLForces API to implement our version or
> would be
> better to implement directly in the source code?
>
>
> Any directions would be very much appreciated.
> Thank you all for any help.
>
>
> My best regards,
>
> Emerson
>

-- 
Brian Radak
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory
9700 South Cass Avenue, Bldg. 240
Argonne, IL 60439-4854
(630) 252-8643
brian.radak_at_anl.gov

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:21 CST