On Thu, Apr 24, 2014, newamber list wrote:
>
> I want to see temperature influence so want to increase the temperature
> slowly (say 0.5 K increment) like 320K, 320.5K, 321K, 321.5K and further
> upto 350K and planning to run 1ns or so for each temperature.
>
> In the output file one can see, in almost all simulations, temperature
> fluctuations are nearly +/- 0.5 K over the run. So is it wise to choose dT
> as 0.5 K increments or I should increase it say 1K or 2K?
Be sure to distinguish the "instantaneous temperature" (which is basically
the kinetic energy at each step) from the mean temperature (which is what
would be compared to an experimental or "thermodynamic" temperature. You need
to estimate how much uncertainty you have in the mean temperature, in order
to know how precisely you know the mean temperature.
Gratuitous suggestion: start with bigger temperature increments, say 10
degrees, and make sure you can interpret the results, and are sufficiently
equilibrated and so on. You can then go back in fill in intermediate
temperatures. OR, consider using temperature replica-exchange: it is
designed to study the effects of temperature.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 24 2014 - 05:00:05 PDT
