[gmx-users] Re: Simulations backward in time

[Vitaly Chaban]

>
>
> Vitaly Chaban wrote:
>>>  I wanted to perform some simulations backward and forward in time (for
>>> transition path sampling ). If I specify a negative value for 'dt' in the
>>> mdp file, would that work for backward integration of the equations of
>>> motion ?
>>
>> Time does not run back. :) It's a bad idea.
>>
> I think it will work, but I'm not sure whether this is the best solution
> to the problem of transition path sampling.

No doubt, the integration itself will work. But will all the attendant
algorithms work correctly while the system moves from more favourable
state to less favourable ones?


Just for interest, I started such run (dt = -0.001) for liquid
acetonitrile box and after a few picoseconds obtained a series of pdb
files and a complete crash.

About transition path sampling... I don't know. :)


-- 
Vitaly V. Chaban
School of Chemistry
National University of Kharkiv
Svoboda sq., 4, Kharkiv 61077, Ukraine
email: chaban at univer.kharkov.ua
skype: vvchaban