Topic Options
#37203 - 12/03/18 03:38 PM No dynamics frames in written DCD
jusadi Offline
Forum Member

Registered: 09/19/18
Posts: 2
Hello all,

I recently began using CHARMM, and I am experiencing a probably quite simple problem.

What questions should I be asking to understand why I'm not getting dynamics output? Is there a clear solution here? What additional information would be useful?

- - - - -
The context: I run a CHARMM job with input scripts from CHARMM-GUI with light modifications (to run length). The input scripts are for GBSW on a small protein. My goal is to get a moderate length of dynamics as a test run.

CHARMM version: 43b1 free
OS: Linux-3.10.0-862.14.4.el7.x86_64(x86_64)
- - - - -
The problem: I do not seem to get any dcd frames from dynamics.

- - - - -
What have I done so far:

I have checked a few basic things: 1) the simulation to finishes properly, 2) I do get minimization frames, 3) the logfiles seems to indicate that dynamics is occurring.

I talked with the folks over at CHARMM-GUI to see if the problem is on that end. This does not appear to be the case, and the problem was not seen as obvious.
- - - - - -
I attach the input file and the logfile.
- - - - - -

Thank you very much for your help,
Jacob


Attachments
step2_implicit_membrane-inp.txt (8 downloads)
step2_implicit_membrane-logfile.txt (9 downloads)


Top
#37204 - 12/04/18 02:30 AM Re: No dynamics frames in written DCD [Re: jusadi]
lennart Offline

Forum Member

Registered: 09/25/03
Posts: 4688
Loc: ~ 59N, 15E
You seem to be running this in parallel with a non-parallel CHARMM version.
Try just running it on a single node (or on your workstation)
_________________________
Lennart Nilsson
Karolinska Institutet
Stockholm, Sweden

Top
#37207 - 12/04/18 02:54 PM Re: No dynamics frames in written DCD [Re: jusadi]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8259
Loc: 39 03 48 N, 77 06 54 W
The repeated output suggests it was run with a parallel version.

An unprotected PRNLEV 5 appears to have triggered output from all processes, making the log file hard to read and interpret. For parallel usage, PRNLEV commands should be e.g.

PRNLEV 5 NODE 0

This restricts output to only the initial process, which is usually what one wants.
_________________________
Rick Venable
computational chemist


Top
#37221 - 12/06/18 08:51 PM Re: No dynamics frames in written DCD [Re: jusadi]
jusadi Offline
Forum Member

Registered: 09/19/18
Posts: 2
Hello all,

The initial problem we had has now been cleared up. You both bring up significant points I hope to address.

From what you're saying, I am unsure whether this implementation is parallel or not. Dr. Nilsson, you say that this is run in parallel with a non-parallel CHARMM version. Where should I be looking to verify this?

Dr. Venable, thank for your suggestion regarding simplifying the logfile output. I've done as you suggest, and the logfile is much easier to manage.

Thanks,
Jacob

Top
#37224 - 12/07/18 11:01 AM Re: No dynamics frames in written DCD [Re: jusadi]
rmv Online   content

Forum Member

Registered: 09/17/03
Posts: 8259
Loc: 39 03 48 N, 77 06 54 W
Add the following two commands to your input--

PREF
SHOW

These will produce listings that indicate what features were added at compile time, including any parallel keywords.
_________________________
Rick Venable
computational chemist


Top

Moderator:  John Legato, lennart