Can't calculate dielectric constant of a big simulation

User discussions
1

GROMACS version: 2020.4
GROMACS modification: Yes/No
Here post your question

Hello everyone!
I’ve been working on a 512 molecules system, I ran 400,000,000 steps with a dt=0.002, rlist=1.2, coulombtype=PME, DispCorr=EnergPres, NPT. I tried to calculate the dielectric constant using gmx dipoles, but the process is killed at 542000 ps. I tried a few options but the process is killed at the same time. Is there a problem calculating properties for a big simulation?

Thanks for the help

2

Hi,
Have you checked that your traj file is not corrupted?
You can use gmx check to do that.
Best regards
Alessandra

3

I used gmx check and got this:

Reading file file.tpr, VERSION 2020.4 (single precision)
Checking file file.trr
trr version: GMX_trn_file (single precision)
Reading frame       0 time    0.000   
`# Atoms  8704`
Last frame     800000 time 800000.000   


Item        #frames Timestep (ps)
Step        800001    1
Time        800001    1
Lambda      800001    1
Coords      800001    1
Velocities  800001    1
Forces      800001    1
Box         800001    1

Does this mean that none file is corrupted?

4

You may simply be running out of memory trying to do that calculation. Does the command complete successfully when limiting the number of frames read, e.g. by setting -e and a short time?

5

Yes, when I set the -e option eg. -e 10000, there’s no problem. I got around 400 GB of free space, is that the type of memory you’re talking about?

6

Disk space isn’t relevant, it’s the amount of memory used in the calculation. Since gmx dipoles is looping over all the molecules and all the frames to compute the dielectric, it requires a lot of memory.

7

So I need to increase the memory, got it. Thank you very much!