Re: [AMBER] MMPBSA.py.MPI Nmode problem

From: Jason Swails <jason.swails.gmail.com>
Date: Wed, 19 Jan 2011 22:21:11 -0500

Hello,

On Wed, Jan 19, 2011 at 8:14 PM, filip fratev <filipfratev.yahoo.com> wrote:

> Hi Jason,
>
> >Other common problems: compiling nab in parallel (parallel> AmberTools)
> >before building MMPBSA.py in serial causes mmpbsa_py_nabnmode to be
> >compiled with MPI support, and its MPI routines clash with the MPI
> >calls in MMPBSA.py.MPI, causing errors.
>
> This was the problem. Thank you very much for your help!!
>

Good to hear. It's tough to *really* avoid this problem, since the
MPI-enabled nab overwrites the serial version when it's built, and I don't
know of a very good way to check it beforehand. mmpbsa_py_nabnmode is only
built during the serial build, so in future releases when this is more
streamlined this error should be more easily avoided.


>
> Now it works fine and the only test error is "possible FAILURE: check
> FINAL_RESULTS_MMPBSA.dat.dif". Presumably the reason is that the standard
> deviation error in my results is not so small. I obtained (by NAB test)
> -10.7398 +/- 4.4178, but the test result was -9.797 +/-2.4051. Thus the
> output massage was "maximum absolute error in..=2.20e+0 at line 10 field 3".
> I suppose that this is not problem?
>

This is not a problem. I've found very strong compiler/system-dependence of
the nmode results. Moreover, since nmode calculations take so long, the
test is a very simple, meaningless system (tri-alanine and di-glycine), so
it's not terribly surprising that these results are so susceptible to this
kind of effect.


>
> However, it is strange for me that I have so high SD values in these
> tests..In the next test (07 test) I obtained the same massage and the
> results were very different than those provided by Amber. Should I worry
> about that?
>

This is the comprehensive test which tests PB, GB, nmode, and the
quasi-harmonic entropy approximation on ALA-GLU-ALA and ALA-ARG-ALA (since a
real system would be too expensive for a simple test just to make sure that
MMPBSA.py works). Therefore, it's not terribly surprising that you'll get
deviations when you don't run on the systems that were used to make the
saved files. For reference, the save files were created on a Mac OS X
machine using the GNU compilers with the Mac Accelerate framework.

Furthermore, don't forget to apply the bug fixes (
http://ambermd.org/tutorials/advanced/tutorial3/py_script/mmpbsa_patches.html),
and I don't think that the test files were updated to reflect changes in the
way some of the standard deviations were calculated.

Hope this helps,
Jason


> Regards,
> Filip
>
>
>
>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jan 19 2011 - 19:30:05 PST
Custom Search