Re: [AMBER] Generation of velocities depends on conformation?

From: David A Case <>
Date: Wed, 7 Nov 2018 07:52:24 -0500

On Wed, Nov 07, 2018, Nicolas Cheron wrote:
> I plotted the temperature after the first step of MD and observed something
> that puzzles me: the initial temperature is configuration dependent. By
> that I mean that for a given conformation, all 200 temperatures are around
> 400K. For another, the temperatures are around 350K. For others, around
> 300K. After 1ps of simulation, all temperatures are randomly distributed
> around 300K.

To add to what Tom wrote: if you plot temperature vs. time, you will
see that it takes far less than 1 ps for the temperatures to come into
line with the thermostat. Are you using SHAKE? If so, what I suspect
is happening is this: the algorithm in sander and pmemd to set initial
velocities according to a Maxwell-Boltzmann distribution doesn't know
about SHAKE, so you will get a spike in the temperature at the very
beginning; as Tom pointed out, this goes away very quickly.

If you are concerned about this, you can use the "setvelocity" action in
cpptraj to make the proper corrections. [Developers: why don't we port
this ability into sander and pmemd?]

[Ignore the above if you are not using SHAKE.]


AMBER mailing list
Received on Wed Nov 07 2018 - 05:00:05 PST
Custom Search