[gmx-users] Re: REMD analysis

Mark Abraham mark.j.abraham at gmail.com
Fri May 24 12:05:46 CEST 2013


On Fri, May 24, 2013 at 10:44 AM, bharat gupta <bharat.85.monu at gmail.com>wrote:

> Dear Sir,
>
> Thank you for your detailed response to my query. I understood the concept
> of ordered arrangement of ensembles in replica_index.xvg. But I have a
> doubt, you said that " *At time 4, replicas in ensemble 1 and 2 have
> exchanged. So replica 0 is now in ensemble 2, which is expressed by 0 in
> the third column*
> *of the third row of replica_index.xvg.* "
>
> This is fine , as the output of replica_index is :-
> 4           1    2    *0 *   3    4    5    6    7    8    9   10   11   13
>   12
>
> But, i didn't understand this " The same condition is expressed by the
> first column of the third row of replica_temp.xvg, where you will find 2,
> also expressing that replica 0 is in ensemble 2 at time 4." Here's the
> output for replica_temp . The first column third row is 2, its ok, but, its
> shows that replica 0 is in ensemble 1 instead of 2.
>

No, if the rows of both matrices describe time, and there are two different
matrices for the same exchange set, then the information described by a
column must differ, like I said last email. You are applying the same
interpretation to a column from either matrix.


> 4           2    0    1    3    4    5    6    7    8    9   10   11   13
> 12
>
>
> In addition to this, in my last mail I showed the temp graph for all
> replicas. (https://www.dropbox.com/s/0c8gp584v1hvlbx/replica_temp.png) .
> Not all replicas visit all the temperatures, but some of them visit all the
> temperatures. Is it sufficient to move with the further analysis , as in
> some papers they show that some replicas visit all the temp which means
> that the sufficient sampling has been achieved. In my case this is true for
> some of the replicas and the average acceptance ratio achieved was 0.22 ??
>

I've answered this question several times. Each replica merely visiting
each temperature means nothing for converged sampling. There's lots of
literature here, including stuff by me ;-) A balance of replicas visiting
ensembles is necessary but not sufficient for the kind of replica flow that
would be necessary for generalized convergence. One can shrug one's
shoulders at some point and say things are probably as good as they'll get
for reasonable cost, but your reviewer might disagree with you. Convergence
of sampling at a single temperature can be assessed in a similar way as for
non-REMD simulations, caveat that the exchange events pretty much stop you
using metrics based on correlation time. If you want to know how to do
things properly, you need to do some reading.

Mark


>
>
>
> On Fri, May 24, 2013 at 5:07 PM, Mark Abraham <mark.j.abraham at gmail.com
> >wrote:
>
> > At time 0 we have an set of replicas and an (ordered) set of ensembles.
> We
> > could label these however we liked, but for (in)convenience we use
> 0-(n-1)
> > for both. The rows of the matrices in the *.xvg files change with time.
> At
> > time 2, replicas in ensemble 0 and 1 have exchanged, so replica 0 is now
> in
> > ensemble 1. At time 4, replicas in ensemble 1 and 2 have exchanged. So
> > replica 0 is now in ensemble 2, which is expressed by 0 in the third
> column
> > of the third row of replica_index.xvg. The same condition is expressed by
> > the first column of the third row of replica_temp.xvg, where you will
> find
> > 2, also expressing that replica 0 is in ensemble 2 at time 4. The columns
> > of the two matrices allow you to see either the profile of which replica
> > was in this ensemble at which time, or which ensemble this replica was in
> > at which time.
> >
> > Mark
> >
> >
> > On Fri, May 24, 2013 at 8:43 AM, bharat gupta <bharat.85.monu at gmail.com
> > >wrote:
> >
> > > Dear Sir,
> > >
> > > I tried a lot to understand the meaning and relation between the .log
> > file
> > > and relica_index file, but I was not able to break the code. I tried to
> > > look into gmx forum for some clue, but didn't find any. So, if possible
> > can
> > > you explain it ...
> > >
> > > Replica exchange at step 1000 time 2
> > > Repl 0 <-> 1  dE = -1.067e+00
> > > Repl ex  0 x  1    2    3    4    5    6    7    8    9   10   11   12
> x
> > 13
> > > Repl pr   1.0       .01       .68       .21       .05       .09
> .26
> > >
> > >
> > > Replica exchange at step 2000 time 4
> > > Repl ex  0    1 x  2    3    4    5    6    7    8    9   10   11   12
> > 13
> > > Repl pr        .91       .32       .00       .07       .18       .08
> > >
> > > output of replica_index.xvg
> > >
> > > 0           0    1    2    3    4    5    6    7    8    9   10   11
> 12
> > > 13
> > > 2           1    0    2    3    4    5    6    7    8    9   10   11
> 13
> > > 12
> > > 4           1    2    0    3    4    5    6    7    8    9   10   11
> 13
> > > 12
> > >
> > > On Thu, May 23, 2013 at 11:20 PM, Mark Abraham <
> mark.j.abraham at gmail.com
> > > >wrote:
> > >
> > > > Looked fine
> > > >
> > > >
> > > > On Thu, May 23, 2013 at 4:13 PM, bharat gupta <
> > bharat.85.monu at gmail.com
> > > > >wrote:
> > > >
> > > > > Sir,
> > > > >
> > > > > What about the description of replica_temp file that I posted in
> last
> > > > mail.
> > > > > I think that's correct ... If you can comment on that, I can move
> on
> > > with
> > > > > replica_index file...
> > > > >
> > > > >
> > > > > On Thu, May 23, 2013 at 10:58 PM, Mark Abraham <
> > > mark.j.abraham at gmail.com
> > > > > >wrote:
> > > > >
> > > > > > It's a demux. One might want trajectories to be at constant
> > > > temperature,
> > > > > or
> > > > > > constant replica. The two files define the (mutually inverse)
> > > mappings
> > > > > > between those representations. So one file tells you which
> replica
> > is
> > > > at
> > > > > > each temperature, and the other which temperature holds each
> > replica.
> > > > > > Nobody's ever written down anything about which is which, so
> like I
> > > > said
> > > > > a
> > > > > > week back, look at the first few exchanges, see how those are
> > > > represented
> > > > > > in the files, and decide for yourself which file's columns/rows
> > have
> > > > > useful
> > > > > > information you want to look at. And do write that decision down!
> > :-)
> > > > > >
> > > > > > Mark
> > > > > >
> > > > > >
> > > > > > On Thu, May 23, 2013 at 2:55 PM, simula_460 <
> > > bharat.85.monu at gmail.com
> > > > > > >wrote:
> > > > > >
> > > > > > > I checked the md.log and replica_temp.xvg file , what I
> > understood
> > > is
> > > > > > that
> > > > > > > the 'x' means swapping and replica are written this way.
> > > > > > > For eg.
> > > > > > > Replica exchange at step 1000 time 2
> > > > > > > Repl 0 <-> 1  dE = -1.067e+00
> > > > > > > Repl ex  0 x  1    2    3    4    5    6    7    8    9   10
> 11
> > > > 12
> > > > > x
> > > > > > 13
> > > > > > > Repl pr   1.0       .01       .68       .21       .05       .09
> > > > > .26
> > > > > > >
> > > > > > > output in replica_temp file will be
> > > > > > > 1 0 2 3 4 5 6 7 8 9 10 11 13 12
> > > > > > >
> > > > > > > It means that replica 1 at higher temp. exchange with the one
> in
> > > > lower
> > > > > > temp
> > > > > > > 0.
> > > > > > >
> > > > > > > Replica exchange at step 2000 time 4
> > > > > > > Repl ex  0    1 x  2    3    4    5    6    7    8    9   10
> 11
> > > > 12
> > > > > > 13
> > > > > > > Repl pr        .91       .32       .00       .07       .18
> > > .08
> > > > > > >
> > > > > > > output in replica_temp file will be
> > > > > > > 1 0 2 3 4 5 6 7 8 9 10 11 13 12
> > > > > > > 2 0 1 3 4 5 6 7 8 9 10 11 13 12
> > > > > > >
> > > > > > > [order is from low to high temp]
> > > > > > >
> > > > > > > But I am not able to understand for replica_index file :-
> > > > > > >
> > > > > > > for the above two time steps here's the output :-
> > > > > > > 0           0    1    2    3    4    5    6    7    8    9   10
> > > 11
> > > > > 12
> > > > > > > 13
> > > > > > > 2           1    0    2    3    4    5    6    7    8    9   10
> > > 11
> > > > > 13
> > > > > > > 12
> > > > > > > 4           1    2    0    3    4    5    6    7    8    9   10
> > > 11
> > > > > 13
> > > > > > > 12
> > > > > > >
> > > > > > > The time step four is different here, I don't know why ??
> Ideally
> > > the
> > > > > > > output should be same in both files, I suppose ??
> > > > > > >
> > > > > > > Also, I tried to plot for each column separately , here I want
> to
> > > > > clarify
> > > > > > > that whether each column represents the time evolution of each
> > > > replica
> > > > > > over
> > > > > > > time. For eg. the second column should represent the temp
> > evolution
> > > > for
> > > > > > > replica No. 0 wrt to time. Presuming that I understood it
> > > correctly,
> > > > I
> > > > > > > plotted the temp. evolution over time of all replicas
> separately
> > .
> > > > > Here's
> > > > > > > the replica_temp plot for replicas 0 to 13.
> > > > > > >
> > > > > > > https://www.dropbox.com/s/0c8gp584v1hvlbx/replica_temp.png
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > --
> > > > > > > View this message in context:
> > > > > > >
> > > >
> http://gromacs.5086.x6.nabble.com/REMD-analysis-tp5008199p5008481.html
> > > > > > > Sent from the GROMACS Users Forum mailing list archive at
> > > Nabble.com.
> > > > > > > --
> > > > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > > > > * Please search the archive at
> > > > > > > http://www.gromacs.org/Support/Mailing_Lists/Search before
> > > posting!
> > > > > > > * 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/Support/Mailing_Lists
> > > > > > >
> > > > > > --
> > > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > > > * Please search the archive at
> > > > > > http://www.gromacs.org/Support/Mailing_Lists/Search before
> > posting!
> > > > > > * 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/Support/Mailing_Lists
> > > > > >
> > > > > --
> > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > > * Please search the archive at
> > > > > http://www.gromacs.org/Support/Mailing_Lists/Search before
> posting!
> > > > > * 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/Support/Mailing_Lists
> > > > >
> > > > --
> > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > * Please search the archive at
> > > > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > > > * 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/Support/Mailing_Lists
> > > >
> > > --
> > > gmx-users mailing list    gmx-users at gromacs.org
> > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > * Please search the archive at
> > > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > > * 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/Support/Mailing_Lists
> > >
> > --
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > * 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/Support/Mailing_Lists
> >
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * 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/Support/Mailing_Lists
>



More information about the gromacs.org_gmx-users mailing list