Processes that occur on the msec to seconds timescale are not likely to be
accessible to direct simulation. There is not a simple way to increase the
time step much more than you have done already.
For slow processes, other methods tend to be used, such as biased MD,
umbrella sampling and so on. Most of these methods require you to have
some advance knowledge of the slow motion (defining what are called
"collective variables"). Some groups have used machine learning to identify
slow motions. Another possibility is to look into methods such as Markov
state models.
This is an active research area and you would probably want to read the
literature on the topic.
On Thu, Jun 10, 2021 at 11:05 AM Syeda Amna Arshi <seyydaamnaarshi.gmail.com>
wrote:
> Hii all,
> I am running a long simulation in which the changes that will take place is
> of the order of msecs or secs. We already increase the timesteps by 2 folds
> by using hydrogen mass repartitioning. Is there any idea how we increase
> the timesteps or speed up these long simulations.
> --
>
> Sincerely,
>
> *Syeda Amna Arshi*
>
> *Ph.D. Scholar*
>
> Multidisciplinary Centre For Advanced Research and Studies
>
> Jamia Millia Islamia
>
> Delhi-110025
>
> Ph.no: 7737638065
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 10 2021 - 09:00:02 PDT
