[gmx-users] Free energy calculation, parallel or sequential lambda runs ?
dmobley at gmail.com
Fri Dec 15 20:03:48 CET 2006
> OK, I do basically the same, but instead of starting the second
> equilibration from the same structure for each lambda, I start from the
> final structure of the previous lambda (still, I equilibrate before
> production for each lambda). I don't see why this strategy should be
> worse the one you outlined, unless there was something strange/bad
> happening at some lambda value, which is already bad on itself.
> Granted, there is the obvious advantage of parallel runs, but when you
> only have one computer this is unimportant (and maybe with the
> "sequential" run you can go with shorter equilibrations).
What I've shown in my work is that at some particular lambda values,
there maybe worse sampling problems than at any other lambda value.
This means that you may need to run longer at that particular lambda
value in order to get adequate statistics. Unless you stop and analyze
each of your lambda values before going on, noticing this later would
mean you would need to go back and re-start the simulation of all of
the subsequent lambda values, since the first time you didn't, you
didn't use the correct ending structure.
> Am I missing anything? When you talk about "sequential" run, do you
> mean no equilibration for each lambda? (there I do see a problem).
I agree equilibration helps. But I think all of the four reasons I
described above still apply even when there is equilibration. I don't
think it is *horrible* to do things this way -- I just think it's
better to do them the other way, because there are fewer pitfalls you
have to watch out for, and less potential for wasted time. If I were
reviewing your paper I would ask you how your results were different
if you repeated some of your calculations using the parallel scheme.
> All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
> gmx-users mailing list gmx-users at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
More information about the gromacs.org_gmx-users