Re: water box and padding

From: Jeff Comer (
Date: Mon May 02 2016 - 11:10:45 CDT


Cutting out pieces of the simulation system can be useful for
improving efficiency, but whether it affects the results really
depends on the problem that you are interested in. For example,
protein subunits irrelevant to the problem at hand are often
truncated. You have to justify this carefully in your manuscripts.

I usually use 8–9 Å or 9–11 Å cutoffs. The improvement in performance
can be as much as 100% versus a 10–12 Å cutoff. Doing this can
sometimes be questioned by reviewers, so I've always been careful to
show that I've validated the choice of cutoff. For example, see the SI
of or Figure 1 of


Jeffrey Comer, PhD
Assistant Professor
Institute of Computational Comparative Medicine
Nanotechnology Innovation Center of Kansas State
Kansas State University
Office: P-213 Mosier Hall
Phone: 785-532-6311

On Mon, May 2, 2016 at 9:54 AM, Brian Radak <> wrote:
> That's unfortunately a highly subjective request. There is no definitive
> answer that I could speak of.
> Some fairly objective experience - reducing the cutoff and switch distance
> while using an LJcorrection can give ~10% speedup without statistically
> changing NpT quantities, at least for small water boxes. I couldn't say what
> this will do for heterogenous protein and/or membrane systems (there have
> been some papers on this in the last decade or so).
> HTH,
> Brian
> On 04/30/2016 11:40 AM, Dhiraj Srivastava wrote:
> Hi all
> whats the minimum padding that I should use without compromising
> with the results of simulation? when I am using 10 A padding then I am
> getting more than 200,000 atoms and in our machine with GPU, I can only get
> 3 ns / day. I want to speed up the calculation without compromising with the
> results. Can I go upto 5 A padding?
> Thanks
> Dhiraj
> --
> Brian Radak
> Postdoctoral Appointee
> Leadership Computing Facility
> Argonne National Laboratory
> 9700 South Cass Avenue, Bldg. 240
> Argonne, IL 60439-4854
> (630) 252-8643

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:05 CST