RE: Cuboid water box issues

From: Raman Preet Singh (ramanpreetsingh_at_hotmail.com)
Date: Tue Aug 11 2020 - 13:28:55 CDT

Dear Peter and Josh,

In fact, I tried all the steps that the error tried to suggest. As Josh mentioned, useFlexibleCell was set to ”no” but was of no help.

I tried the following steps which resolved the issue. Thought of sharing my experiences so it might help others in the NAMD community. I made the following changes:

  1. Equilibration step was performed with NVT – 10000 minization steps followed by a run of 125000 steps (2 fs/step).
  2. Production run using NPT (2 fs/step).
  3. Changed useConstantArea “no” to useConstantRatio “no” in equilibration and production runs.

Regards,
Raman

From: Peter Freddolino <petefred_at_umich.edu>
Sent: Saturday, August 8, 2020 1:46:50 AM
To: <namd-l_at_ks.uiuc.edu> <namd-l_at_ks.uiuc.edu>; Raman Preet Singh <ramanpreetsingh_at_hotmail.com>
Subject: Re: namd-l: Cuboid water box issues

Did you try the other things suggested by the error message? That is, restarting from a recent checkpoint or increasing the margin?
Best,
Peter

On Fri, Aug 7, 2020 at 3:22 PM Raman Preet Singh <ramanpreetsingh_at_hotmail.com<mailto:ramanpreetsingh_at_hotmail.com>> wrote:

Dear All,

I am trying to simulate a linear protein fragment in water box. Using the solvate module in VMD, I generated a water box measuring 238x45x45 A. I have followed the steps mentioned in NAMD tutorial (Ubiquitin in water box).

After a minimization of 10000 steps, production run of 10 ns follows. However, after 1000 steps of the production run, the following error occurs:

FATAL ERROR: Periodic cell has become too small for original patch grid! Possible solutions are to restart from a recent checkpoint, increase margin, or disable useFlexibleCell for liquid simulation.

The conf file is attached for your reference.

I tried increasing minimization to 25000 steps but the error still appears in production run after running for a few thousand steps. Manually changing PMEGridSizeX/Y/Z offered no advantage.

When I increased the water boxsize to 235x235x235 A and minimize for 10000 steps, the simulation runs fine.

My question is: is this error occurring due to the shape of box (a long cuboid compared to a near cubic box usually employed)? The bigger boxsize has resulted in several fold increase in number of water molecules resulting in a painful increase in runtime.

Please help resolve the issue.

Thanks!

Raman

From: Peter Freddolino<mailto:petefred_at_umich.edu>
Sent: 08 August 2020 01:47
Subject: Re: namd-l: Cuboid water box issues

Did you try the other things suggested by the error message? That is, restarting from a recent checkpoint or increasing the margin?
Best,
Peter

On Fri, Aug 7, 2020 at 3:22 PM Raman Preet Singh <ramanpreetsingh_at_hotmail.com<mailto:ramanpreetsingh_at_hotmail.com>> wrote:
Dear All,

I am trying to simulate a linear protein fragment in water box. Using the solvate module in VMD, I generated a water box measuring 238x45x45 A. I have followed the steps mentioned in NAMD tutorial (Ubiquitin in water box).

After a minimization of 10000 steps, production run of 10 ns follows. However, after 1000 steps of the production run, the following error occurs:
FATAL ERROR: Periodic cell has become too small for original patch grid! Possible solutions are to restart from a recent checkpoint, increase margin, or disable useFlexibleCell for liquid simulation.

The conf file is attached for your reference.

I tried increasing minimization to 25000 steps but the error still appears in production run after running for a few thousand steps. Manually changing PMEGridSizeX/Y/Z offered no advantage.

When I increased the water boxsize to 235x235x235 A and minimize for 10000 steps, the simulation runs fine.

My question is: is this error occurring due to the shape of box (a long cuboid compared to a near cubic box usually employed)? The bigger boxsize has resulted in several fold increase in number of water molecules resulting in a painful increase in runtime.

Please help resolve the issue.

Thanks!

Raman


2A497EA073BD45AB9E8027E345A093AA.png

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:13 CST