GROMACS version:
GROMACS modification: Yes/No
I generated a final tpr file with mdp file in the step prior to start the final production run and got two warning , The first one I understand is because I skipped equilibration step but the second one I am not quite getting , Kindly help .
The warnings:
WARNING 1 [file 0_20ns.mdp]:
You are generating velocities so I am assuming you are equilibrating a
system. You are using Parrinello-Rahman pressure coupling, but this can
be unstable for equilibration. If your system crashes, try equilibrating
first with Berendsen pressure coupling. If you are not equilibrating the
system, you can probably ignore this warning.
WARNING 2 [file unknown]:
There are atoms at both ends of an angle, connected by constraints and
with masses that differ by more than a factor of 13. This means that
there are likely dynamic modes that are only very weakly coupled. To
ensure good equipartitioning, you need to either not use constraints on
all bonds (but, if possible, only on bonds involving hydrogens) or use
integrator = sd or decrease one or more tolerances:
verlet-buffer-tolerance <= 0.0001, LINCS iterations >= 2, LINCS order >=
4 or SHAKE tolerance <= 1e-05
Warnings indicate likely problems with the physics of the simulations. The first is trivial to solve - don’t use Parrinello-Rahman for a simulation in which velocities are generated. The second is more challenging and depends on the force field that’s being used but the suggestions provided should give you a solution.
Hi Justin – we have a question about long (microsecond) peptide-membrane simulations we’ve been doing. We start the simulation off with random velocities, but we do still use Parrinello-Rahman (in other words, we have been ignoring the warning) because membrane runs can a while to equilibrate and we know that the berendsen one, while more robust/stable, is inaccurate so we didn’t want to use it for these systems where semi-isotropic coupling is needed. We therefore don’t do a shorter equilibration step followed by a production run, but we just do one long simulation with P-R. In all of our simulations thus far, the system appears stable and reasonable (meaning it equilibrates fine, visually is fine, and maintains equilibrium such that we can eventually sample the production “period” of the simulation, etc.) even though we started with the P-R barostat. So our question is – if our system appears stable and reasonable and proceeds normally, is there anything really wrong about ignoring this warning and just using the P-R barostat at the start of our simulations? (thermodynamically, if something is equilibrated and at that point you’re using reasonable force field/barostat, etc. when sampling, then it shouldn’t really matter how it got there because of the whole state function thing, but I just wanted to make sure that I’m not missing something because I am in fact ignoring a warning!)
This is fine. But noways there is no reason not to chose the c-rescale and v-rescale coupling algorithms, as they produce the correct ensemble and have no stability issues.