Anisotropic pressure coupling with hexagonal box?

[CHANG YUN SON]

Hi all,

Is it possible to simulate a system with triclinic box with an semi-isotropic pressure coupling such as membrane barostat?
I particularly want to simulate a hexagonal box with unknown box dimensions to run a self-assembly simulation of hexagonal phase.
I tried to use the membrane barostat. The simulation runs fine for a few nanoseconds when I initiate the box to be larger than the actual system, but crashes when the system size gets small enough. (The same system in rectangular box runs fine with membrane barostat) I considered using the montecarlo-anisotropic barostat, but it would not couple the xy dimensions, so it would be a bit hard to get the right periodicity for the hexagonal phase. Another minor thing I noticed was that with the python reporters (DCDReporter), the particle coordinates are always aligned with rectangular box, while the box vector shows proper triclinic alignment.
If it's not yet supported by openmm, I can try to equilibrate with other packages, but Openmm is faster in my particular case.

I'm using Openmm-7.0.1.

Thanks,
ChangYun

[Peter Eastman]

When you say it crashes, what exactly does that mean? Does it print an error message? If so, what is it?

Could you try using OpenMM 7.1? That's the latest release, so the first thing is to make sure the problem still exists. If it does, could you post files so I can reproduce it?

Thanks!

Peter

[CHANG YUN SON]

The system blows up and the energy rapidly increases to reach the NaN. The error msg is weird though, since I initially got the usual 'energy value is NaN' error, which is typical for the system that's blowing up, but I tried one last time today and got 'the periodic box size has decreased to less than twice the nonbonded cutoff.' Though the log file tells me the last simulation experienced the same problem of rapidly increasing energy, so I don't know why this strange msg came out this time.

I'll try to use the latest 7.1 version and see if it's still there. I'll put my input files if the problem persists.

Thank you,
ChangYun

[CHANG YUN SON]

Peter,

Updating Openmm to 7.1 fixed the problem, and the simulation is now running fine up to several tens of ns. (The simulation is still running.)

Thanks for the help ! (But I'm going to bug you with another question ... )
ChangYun

[Peter Eastman]

Glad to hear it!

Peter