Previous Thread
Next Thread
Print Thread
Joined: Jun 2010
Posts: 19
J
Forum Member
OP Offline
Forum Member
J
Joined: Jun 2010
Posts: 19
I want to split a production dynamic into several different parallel-run trajectories, each using a different starting point and a different random seed, to obtain a longer time of simulation in a shorter period of time (as calculations can be run side-by-side).

This might sound like a very basic question, but I haven't found in the documentation or forums a clear answer: How does defining a restart file affect the different starting snapshots? I assume for this kind of parallel runs a fresh start is preferred, but I'd like to see what exactly a restart file fixes for the continuining trajectory.


"With magic, you can turn a frog into a prince. With science, you can turn a frog into a Ph.D., and you keep the frog you started with." - Terry Pratchett
Joined: Sep 2003
Posts: 8,606
Likes: 24
rmv Online Content
Forum Member
Online Content
Forum Member
Joined: Sep 2003
Posts: 8,606
Likes: 24
I don't believe it is possible. A lot of the information in the restart file is only available during the dynamics run, such as state data, velocities, and difference coordinates. If you saved a matching velocity trajectory, there is a provision to supply velocities from a file for a STRT run, but that is not completely equivalent to a restart.


Rick Venable
computational chemist

Joined: Sep 2003
Posts: 4,850
Likes: 10
Forum Member
Online Content
Forum Member
Joined: Sep 2003
Posts: 4,850
Likes: 10
I don't quite see the relevance of restart files in the context of running independent replicate simulations (a very good practice!) - restart files (which provide continuity by providing all the data that Rick refers to) should not be used to start independent runs. Just start a set of simulations with different values for the random number generator seed (ISEEd keyword, dynamc.doc).


Lennart Nilsson
Karolinska Institutet
Stockholm, Sweden
Joined: Jun 2010
Posts: 19
J
Forum Member
OP Offline
Forum Member
J
Joined: Jun 2010
Posts: 19
Well, the different starting coordinates for the replicate simulations come from a pre-existing trajectory calculation (of which snapshots are taken at regular intervals to provide the different starting points), hence the possible relevance of previous restarting data.
Thank you both for the replies.


"With magic, you can turn a frog into a prince. With science, you can turn a frog into a Ph.D., and you keep the frog you started with." - Terry Pratchett
Joined: Feb 2010
Posts: 29
Forum Member
Offline
Forum Member
Joined: Feb 2010
Posts: 29
Originally Posted By: JoseLDominguez
Well, the different starting coordinates for the replicate simulations come from a pre-existing trajectory calculation (of which snapshots are taken at regular intervals to provide the different starting points), hence the possible relevance of previous restarting data.
Thank you both for the replies.


I think there's a bit of confusion in terms of charmm's syntax. I think what you're asking is how to start dynamics anew from points along your original trajectory, no?

You've referred to that as "re-starting" them, but strictly speaking to restart dynamics using the DYNA REST command is to continue them, as Lennart and Rick have been saying. In other words, in charmm REST = continue dynamics by picking up where you left off.

If I've understood things correctly, you want to DYNA STARt from the coordinates taken from your original trajectory but with new (randomly assigned) velocities (you could also perturb the old ones, I suppose). You can use the IASORS .eq. 0 to assign new velocities and you can use IASVEL to select how to assign them. I imagine you'd want to use the Gaussian (IASVEL .gt. 0). You'll want to take care that the new velocities are appropriate for the temperature of the simulation (see TSTRUC in dynamc.doc, e.g.).


Intestinal nematode
Bronx, NY

Moderated by  BRBrooks, lennart, rmv 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.5
(Release build 20201027)
Responsive Width:

PHP: 7.3.31-1~deb10u1 Page Time: 0.010s Queries: 24 (0.006s) Memory: 0.7474 MB (Peak: 0.8020 MB) Data Comp: Off Server Time: 2022-10-01 01:52:46 UTC
Valid HTML 5 and Valid CSS