Re: Should a pairlist warning concern me?

From: Victor Kwan (vkwan8_at_uwo.ca)
Date: Mon Oct 28 2019 - 09:35:55 CDT

Hi Henrik,

If it only occurs in the beginning of the simulation, try increasing the number of minimization steps.

Victor

On Mon, Oct 28, 2019 at 6:59 AM Henrik Schopmans <h.schopmans_at_gmail.com<mailto:h.schopmans_at_gmail.com>> wrote:
Dear NAMD users,

I setup a simulation with pairlistdist 14, stepspercycle 10, fullelectfrequency 1 and nonbondedfreq 1. I also set outputPairlists to 2000 to receive warnings.

Now there is a single warning at the very beginning of my simulation, "Info: 1 pairlist warnings in past 2000 steps.".

Here (https://www.ks.uiuc.edu/Research/namd/wiki/?NamdPerformanceTuning) it says that the warning does not result in missed interactions and the pairlistdist is adjusted automatically.
But here (https://www.ks.uiuc.edu/Research//namd/2.0/ug/node15.html#section:electdesc) at the very bottom it says warnings indicate that pairlistdist should be increased to conserve energy and momentum.

Is my configuration fine? I had set stepspercycle set to 1 initially, but this resulted in very poor performance. How do you guys determine the right value for stepspercycle?

This archive was generated by hypermail 2.1.6 : Sat Dec 07 2019 - 23:20:52 CST