Hi all,
I’m working on a screened coulomb’s law pair style like coul/debye, and I’ve successfully produced the expected energy and force vs. r table with the potential, but I’m still having a bit of trouble with it and I’m hoping for a debugging tip or two. When I try using the Nose-Hoover thermostat, the temperature control is wacky (e.g., a run that ought to end at 7000 K ends at 12000 K, albeit with a relatively monotonic increase), but the Berendsen thermostat works fine.
However, I don’t understand the Nose-Hoover integrator well enough to know what kind of problem with my potential might show up with Nose-Hoover and not with the Berendsen thermostat. Can anybody who’s familiar with Nose-Hoover suggest other values that I might check beyond energy, force and temperature to try to understand what might be going wrong?
Thanks!
Tyler Harvey
The NH thermostat knows nothing about your pair style. If you can
run your pair style stably with NVE, then you can run it with fix nvt.
Steve
Hey Steve and Maxim,
Thanks for the tips! I’ll definitely try running in NVE and playing with the tchain parameter to see if I can diagnose what’s going on. Steve, do you have any suggestions of what to look for in NVE that might be signs of the same kind of instability that’s causing wild temperatures in NVT? Wild temperatures again, perhaps … maybe melting crystals at too-low temperatures?
Tyler
Steve, do you
have any suggestions of what to look for in NVE that might be signs of the
same kind of instability
In NVE you should just check that you are conserving energy.
Steve
Hah, right. Sorry for the dumb question. Turns out energy conservation was awful, but nearly as bad with the tabulated form of the potential–so it looks like I’ve got bigger fish to fry with this potential beyond subtle mistakes I made in writing the pair style. Thanks for the debugging help, both of you!
Tyler