[gmx-users] REMD: A small bug in repl_ex.c and a related question

[Mark Abraham]

On Fri, Jun 27, 2014 at 5:27 AM, Suman Chakrabarty <
chakrabarty.suman at gmail.com> wrote:

> Hello!
>
> 1. It seems I have encountered a minor bug (?) in repl_ex.c for
> version 4.6.x (for NPT simulations only):
>
> Line 880 (in version 4.6.5):
>             fprintf(fplog, "  dpV = %10.3e  d = %10.3e\nb", dpV, delta +
> dpV);
>
> should be:
>             fprintf(fplog, "  dpV = %10.3e  d = %10.3e\n", dpV, delta +
> dpV);
>
> The extra "b" results into lines like this in the log file:
> bRepl ex  0    1    2    3    4    5    6    7    8    9   10   11
> 12 x 13   14 x 15   16 x 17   18   19   20 x 21
>
> While the REMD itself runs fine, the lines containing "bRepl ex" are
> never parsed by the demux.pl script. So, only half of the exchanges
> are analyzed by the script.
>

Yes. This will be fixed in 5.0. You can safely hack out the excess "b" in
the source if you want a fix in 4.6.x


>
> 2. The related question: In the log file, the dPV corrections are
> reported only every alternate exchange attempts: (I have kept only a
> relevant trimmed part below)
>
> Replica exchange at step 2000 time 4
> Repl ex  0    1    2    3    4    5 x  6    7 x  8    9 x 10
>
> Replica exchange at step 3000 time 6
> Repl 0 <-> 1  dE_term =  6.670e-01 (kT)
>   dpV = -3.918e-05  d =  6.669e-01
> bRepl ex  0 x  1    2    3    4    5    6    7    8 x  9   10
>
> Why is this necessary? I hope it doesn't mean the correction is being
> applied every alternate attempt (I have not checked this part of the
> code yet).
>

Such output is only done from the lower-numbered replica, and the sets of
replicas that exchange alternate between subsequent exchange attempts. You
can only see the full situation by considering all the .log files.


> Is it safe to use NPT REMD with the current version of the code? If I
> just apply the above correction, and use demux.pl to analyze ALL
> exchange attempts, is it still valid? Please confirm.
>

I expect so. Note that all simulation users should be doing their own
sanity checks, such as confirming that your settings produce statistically
indistinguishable results on (say) some water boxes with and without
-replex on. Nobody else has tested your exact system ;-) Or
http://dx.doi.org/10.1021/ct300688p

Mark


> Thanks,
> Suman.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>