On 9/19/12 5:10 PM, Ladasky wrote:
After weeks of trying various conditions, I found my problem. Here, from my
position-restrained MDP file, are the two relevant lines:
; Temperature coupling is on
tcoupl = V-rescale ; Weak coupling for initial equilibration
[snip]
; Pressure coupling is on
pcoupl = Berendsen ; Pressure coupling on in NPT, also weak
coupling
This is a change that I forgot to mention in my list of changes, but in fact
I did post the full MDP file. I switched to using two DIFFERENT algorithms
for temperature and pressure coupling when I performed the initial
position-restrained equilibration. This frequently sets up an INSTABILITY
which manifests itself anywhere from a few picoseconds to a full nanosecond
into the standard equilibrium MD run.
So the initial equilibration was NPT? Did you ever try simply running NVT with
either Berendsen or V-rescale before applying any type of pressure coupling?
Immediate application of NPT is often unstable for a variety of reasons, not
necessarily the temperature coupling algorithm.
I am giving this number in real time, rather than in compute cycles, because
I tried cutting my step time to 1.0 fs from my usual 2.0 fs -- and I still
had a water molecule dislocate abruptly, at nearly the same physical
location and simulated time point.
I will share my full debugging process if anyone is interested. It took me
several runs, all of them over 24 hours long, to figure this out.
Now, WHY would I have switched from a Berendsen temperature-coupling
algorithm to the V-rescale algorithm? Because of this cautionary note that
I started receiving in GROMACS 4.5 when I started the position-restrained
mdrun:
NOTE 1 [file pr.mdp]:
The Berendsen thermostat does not generate the correct kinetic energy
distribution. You might want to consider using the V-rescale thermostat.
I took this as a warning that a deeper study of the algorithms had revealed
a flaw in Berendsen, and so I sought to use what GROMACS was advising me
would be a better choice. This is apparently MISLEADING advice when doing
the initial equilibration. If you are using the Berendsen pressure
algorithm, you must also use the Berendsen temperature algorithm.
I know that mdrun cannot be sophisticated enough to know which phase of my
run I am performing, but I hope that the GROMACS documentation will be
revised to discuss the appropriate use of the V-rescale thermostat. I just
misused it,and wasted a lot of time as a consequence. I hope that some
discussion of its proper use would be written and summarized on a page like,
for example, this one:
http://www.gromacs.org/Documentation/Terminology/Equilibration
It already states: "Using, i.e., the Nosé-Hoover thermostat for initial NVT
equilibration can lead to wild oscillations of the temperature, with the
system ultimately blowing up." A similar warning should be added about the
V-rescale thermostat in NPT equilibration, ESPECIALLY if GROMACS is going to
issue messages that recommend that you use it!
There is a page for the Berendsen barostat/thermostat:
http://www.gromacs.org/Documentation/Terminology/Berendsen
There is apparently no corresponding page for the V-rescale thermostat.
Perhaps one should be written.
Anyone is welcome to contribute to the wiki; it is a community project, after
all. I would also be interested to see if it can be demonstrated that NPT with
V-rescale + Berendsen is unstable after an independent NVT simulation with
either method.
-Justin
--
========================================
Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
========================================
--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
* Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-requ...@gromacs.org.
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
