Re: PME Grid Settings and its significance relative to a cutoff value

From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Fri Aug 19 2011 - 13:05:26 CDT

On Fri, Aug 19, 2011 at 1:38 PM, Erik Nordgren <nordgren_at_sas.upenn.edu> wrote:
> Andrea,
>
> You may find some useful info in the "non-bonded interactions" section of
> the NAMD User Guide.
>
> The efficiency of the electrostatic energy calculation, when using PME,
> should be fairly insensitive to your choice of cutoff.  The main relevance

that is not entirely correct. the shorter the coulomb cutoff, the faster the
real space part of the electrostatic interactions is computed. however,
that should also increase the computational effort in reciprocal space.
since the shorter the real-space cutoff is, the larger the reciprocal space
and PME grid has to be to maintain the same accuracy. depending on
the hardware and system, it may be possible to gain performance by
increasing the coulomb cutoff in order to change the balance between
the real-space and kspace effort (since the real-space calculation
parallelizes much better than the PME part).

that being said, for typical biological systems, it is usually not worth
the effort and one should stick with the recommended value for the cutoff.

cheers,
    axel.

> of the cutoff distance is in the computation of the VdW forces; the larger
> the cutoff, the more VdW interactions will need to be computed (and the
> slower the simulation will run).  But if you decrease the cutoff too much,
> then accuracy will suffer.  The optimum choice of cutoff may be dictated by
> the details of your system and/or the force field you are using.
>
> -Erik
>
>
>
> On Tue, Aug 16, 2011 at 4:07 PM, Andrea Holfelder <andreaholfelder_at_gmx.com>
> wrote:
>>
>> I need a little information about PME Grid Setting .
>>
>> I am using a default cut off of 12A . Is it a little too high ? Can I cut
>> it down a bit further to improve computing time ?
>> My Pairlist dist is also set 14A .
>>
>> PME divides the whole system into a system grid pattern of Electrostatic
>> interactions and it means that the entire system/PBC unit cell will be
>> considered so does it make a point to still use a cutoff value ?
>>
>> I read that using a cutoff along with PME grid does not mean the
>> disfunctionality of either one. I guess it simpy means that PME grid will be
>> set weaker beyond 12A and will be stronger till 12A.
>> Am I right ? Can somebody explain me the significance of using PME Grid
>> along with setting up a cutoff distance?
>>
>>
>> Any help/information will be great
>>
>>
>>
>> -Andrea
>>
>>
>
> --
> C. Erik Nordgren, Ph.D.
> University of Pennsylvania
>

-- 
Dr. Axel Kohlmeyer
akohlmey_at_gmail.com  http://goo.gl/1wk0
Institute for Computational Molecular Science
Temple University, Philadelphia PA, USA.

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:43 CST