Re: ABF Non-Uniform Zcount

From: Joey Farrell (jpfarrell_at_uchicago.edu)
Date: Tue May 03 2022 - 19:13:03 CDT

Hi Jérôme,

Your second idea about orthogonal drift seems to be correct.

I am trying to run a separation simulation over multiple windows (each window responsible for ~4 Å of separation). For the first window, where this .count non-uniformity is the most exacerbated, it seems that almost all of the simulation movement is due to drift, rather than helical separation along the axis of interest.

I am going to assume that this is some kind of system set up failure on my part; would you happen to have any recommendations as to how a basic ABF procedure could be modified in order to avoid this issue in the first window?

Thank you,
Joey
________________________________
From: Jérôme Hénin <jerome.henin_at_ibpc.fr>
Sent: Wednesday, April 27, 2022 8:40 AM
To: NAMD list <namd-l_at_ks.uiuc.edu>; Joey Farrell <jpfarrell_at_uchicago.edu>
Subject: Re: namd-l: ABF Non-Uniform Zcount

Hi Joey,

First I would compare this with the ".count" file, to see if there is any major difference in distribution. If that is the case, this can be helped by a tighter extended system restraint (lower extendedFluctuation parameter). However, I doubt this is the case here.
A likely cause is slow relaxation causing a drift in an orthogonal direction to your chosen reaction coordinate, so that ABF is adapting to a free energy landscape that seems to be changing all the time, which defeats the adaptation process.

My recommendation to troubleshoot this is to analyze the slow coordinates in the ABF trajectory.

Best,
Jérôme

----- On Apr 26, 2022, at 6:33 AM, Joey Farrell <jpfarrell_at_uchicago.edu> wrote:
Hello,

I was wondering if I can get any insights into potential causes of non-uniform zcount when running an ABF system. Specifically, my .zcount file has the following (abridged) contents:

 12.05 1233990

 12.15 1248322

 12.25 1168793

  ...

 15.75 121013

 15.85 87096

 15.95 57720

Please let me know if this is an issue and, if it is, potential ways to go about remedying it. I can provide more configuration information ad hoc.

Thank you,
Joey

This archive was generated by hypermail 2.1.6 : Tue Dec 13 2022 - 14:32:44 CST