[gmx-users] g_wham gets stuck

[Jochen Hub]

Justin A. Lemkul wrote:
>
>
> Amir Marcovitz wrote:
>> Thanks for your answers,
>>
>> I tried to struggle a bit more with that today.
>> my input dat files listings (i.e., tpr-files.dat, pullf-files.dat and 
>> pullx-files.dat) are fine and consistent with other in terms of file 
>> numbering.
>> i use gromacs 4.0.5 on Linux with gcc 4.1.2 compiler.
>
> Time and time again, we have to point out the bold warning on the 
> Downloads page:
>
> "WARNING: do not use the gcc 4.1.x set of compilers. They are broken. 
> These compilers come with recent Linux distrubutions like Fedora 5/6 
> etc."
>
> Buggy behavior that is difficult to diagnose is often due to these 
> faulty compilers.
>
>>
>> i run: *g_wham -it tpr-files.dat -if pullf-files.dat -o -hist -unit 
>> kCal -b 1000    *(my data is 16ns long and i ignore the first ns)
>>
>> and it starts reporting that the file are read:
>>
>> Reading file topol742.tpr, VERSION 4.0.5 (single precision)
>> File topol742.tpr, 1 groups, geometry "distance", dimensions [N Y N], 
>> (1 dimensions)
>>         grp 0) k = 1000.000  inittial distance = 2.41331
>> Reading file topol748.tpr, VERSION 4.0.5 (single precision)
>> Reading file topol761.tpr, VERSION 4.0.5 (single precision)
>> Reading file topol792.tpr, VERSION 4.0.5 (single precision)
>> and so on..
>>
>> it reports that the boundaries are found and continue to read until 
>> it stucks..
>>
>> However, when i do the WHAM with the pullx files (i.e., -ix 
>> pullx-files instead of -if pullf-files.dat) the wham converges within 
>> a reasonable time to a PMF profile which not so smooth.
>>
>> i therefore have some questions:
>> 1) what is the difference in the profiles for using  pullx or pullf 
>> files?
>
> In principle, there probably shouldn't be any, but if there is perhaps 
> someone else has seen that.
It should not make any difference at all. It it does, there is something 
wrong with the data or g_wham is buggy. With the pullf files, g_wham 
simply computes the pullx values from the force and the force constant. 
This is not really required since it is more natural to just work with 
the pullx, but I added to for the case that someone forgot to write the 
pullx and only has the pullf files.

>
>> 2) suppose that my histograms overlap is poor for some locations 
>> along the pulling vector, how one can solve that?
>
> Better sampling (more time) within the sampling windows, or more 
> closely-spaced windows.
And maybe stronger force constants. g_wham can combine the runs with 
smaller and larger force constants, that is no problem.

That is probably the reason for the weired PMF!

Jochen

>
>> 3) To generate the input configurations -  what is the ideal pulling 
>> procedure?  ( i used     pull = constant_force, with a small value of 
>> K1)
>
> There isn't one "best" method, per se.  The only criterion is that you 
> have generated configurations from which you can derive 
> reasonably-spaced windows in which you do sufficient sampling.
>
> -Justin
>
>>
>> Again, Thanks a lot for the quick reply
>> Amir
>>
>> On Thu, Apr 29, 2010 at 10:33 AM, Jochen Hub <jochen at xray.bmc.uu.se 
>> <mailto:jochen at xray.bmc.uu.se>> wrote:
>>
>>     Amir Marcovitz wrote:
>>
>>         Hi All,
>>
>>         I have some problems with g_wham, and i already gone through all
>>         the postings and didn't find a hint..
>>
>>         basically, I'm trying to calculate PMF between two charged
>>         plates. I've performed a pulling simulation between the 2 plates
>>         according to Justin's UMBRELLA tutorial in the website (all
>>         steps, i.e., minimization, equilibration etc. up to that point
>>         work fine)
>>         from the pulling i generated input configurations for the
>>         umbrella sampling runs (pull=umbrella , rate=0.0), which are 15
>>         ns long
>>         and collected all the output pullf.xvg and *.tpr files.
>>
>>     You could also run g_wham -histonly to get the histogram file. Then
>>     check with xmgrace -nxy histo.xvg whether the histograms properly
>>     overlap.
>>
>>     But if they do not overlap, I would rather expect g_wham to give a
>>     zero PMF or to iterate forever, so not sure what is wrong.
>>
>>     Jochen
>>
>>
>>
>>
>>         i then run g_wham (with -it and -if) and it works fine at the
>>         beginning, but then the computer simply gets stuck (!?) and the
>>         calculation is killed -  with no error massage.
>>
>>         what is it that I'm doing wrong?
>>         it looks like my output data (pullf and tpr files) are fine, but
>>         is it possible that some of them causing the problem?
>>
>>         this is really frustrating..
>>         need your help,
>>         Amir
>>
>>
>>
>>     --     ---------------------------------------------------
>>     Dr. Jochen Hub
>>     Molecular Biophysics group
>>     Dept. of Cell & Molecular Biology
>>     Uppsala University. Box 596, 75124 Uppsala, Sweden.
>>     Phone: +46-18-4714451 Fax: +46-18-511755
>>     ---------------------------------------------------
>>
>>
>>     --     gmx-users mailing list    gmx-users at gromacs.org
>>     <mailto:gmx-users at gromacs.org>
>>     http://lists.gromacs.org/mailman/listinfo/gmx-users
>>     Please search the archive at http://www.gromacs.org/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
>>     <mailto:gmx-users-request at gromacs.org>.
>>     Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
>>
>


-- 
---------------------------------------------------
Dr. Jochen Hub
Molecular Biophysics group
Dept. of Cell & Molecular Biology
Uppsala University. Box 596, 75124 Uppsala, Sweden.
Phone: +46-18-4714451 Fax: +46-18-511755
---------------------------------------------------