[gmx-users] Free energy calculation using inter-molecular bonded interaction

[Justin Lemkul]

On 9/3/15 9:14 AM, Atsutoshi Okabe wrote:
> Thanks, Justin.
> I tried turning [bonds],  [angles], [dihedrals] interaction on one-by-one.
> However, same errors occurred in all cases.
> Also, I tried the calculation on 1 cpu using the following command and the error did not occur!
> (that means I did not use parallelization)
> gmx_mpi mdrun -deffnm equil$LAMBDA  # 1 cpu
>

Well, here you're invoking an MPI-enabled mdrun without specifying a core count 
(-nt) so please make sure (from the .log file) that you are, in fact, only using 
one core.

> So, I am wondering if the calculation using both [intermolecular-interactions] tool and parallelization can not work and that causes the error messages below.
> [mpiexec at leon304] HYDT_bscu_wait_for_completion (./tools/bootstrap/utils/bscu_wait.c:101): one of the processes terminated badly; aborting
> [mpiexec at leon304] HYDT_bsci_wait_for_completion (./tools/bootstrap/src/bsci_wait.c:18): bootstrap device returned error waiting for completion
> [mpiexec at leon304] HYD_pmci_wait_for_completion (./pm/pmiserv/pmiserv_pmci.c:521): bootstrap server returned error waiting for completion
> [mpiexec at leon304] main (./ui/mpich/mpiexec.c:548): process manager error waiting for completion
>
> I used the following command for parallelization
> mpirun -np 8 gmx_mpi mdrun -deffnm equil$LAMBDA # 8cpus
>

Sounds buggy, but I routinely run my calculations with intermolecular 
interactions over 32 cores, so it's not inherently a parallelization problem. 
Could be something unique to your setup.  Please file a bug report on 
redmine.gromacs.org with this information and provide a .tpr file of a system 
that runs on one core but fails in parallel.

-Justin

>
> Bests,
> Atsutoshi
>
> 2015/09/03 5:42、Justin Lemkul <jalemkul at vt.edu> のメール：
>
>>
>>
>> On 9/2/15 10:21 AM, Atsutoshi Okabe wrote:
>>> I mean that the entire [intermolecular_interaction] directive were commented out below;
>>> ;[ intermolecular-interactions ]
>>>
> I tried turning [bonds] or [angles] or [dihedrals] interaction on one-by-one.
> However, same errors occurred in all cases.
> Then, I tried the calculation on 1cpu using the following command and the error did not occur!
> (that means I do not use parallelization)
> gmx_mpi mdrun -deffnm equil$LAMBDA  # 1 cpu
>
> So, I am wondering if the calculation using both [intermolecular-interactions] tool and parallelization can not work and that causes the error messages.
> [mpiexec at leon304] HYDT_bscu_wait_for_completion (./tools/bootstrap/utils/bscu_wait.c:101): one of the processes terminated badly; aborting
> [mpiexec at leon304] HYDT_bsci_wait_for_completion (./tools/bootstrap/src/bsci_wait.c:18): bootstrap device returned error waiting for completion
> [mpiexec at leon304] HYD_pmci_wait_for_completion (./pm/pmiserv/pmiserv_pmci.c:521): bootstrap server returned error waiting for completion
> [mpiexec at leon304] main (./ui/mpich/mpiexec.c:548): process manager error waiting for completion
>
> I use the following command to use parallelization
> mpirun -np 8 gmx_mpi mdrun -deffnm equil$LAMBDA # 8cpus
>>
>>> ;[ bonds ]
>>> ;411 3173  6  0.55  2090
>>>
>>> ;[ angles ]
>>> ;411 3173 3152  1  164.5  20.9
>>> ;409 411 3173  1  122.3  20.9
>>>
>>> ;[ dihedrals ]
>>> ;409 411 3173 3152  2  12.0  20.9
>>> ;407 409 411 3173  2  75.6  20.9
>>> ;411 3173 3152 3144  2  -126.0  20.9
>>>
>>> I checked global atom number in .gro file of protein-ligand complex. It looks no problem.
>>> (407, 409 and 411 are atom number of protein,  3172, 3144 and 3152 are atom number of ligand in gro file of complex)
>>> Curiously, this error occurs when the vdW interaction between ligand and the environment becomes to turn-off(it means Lambda=0.7~1.0).
>>> Conversely, the calculations were finished normally when the vdW interaction remains(it means Lambda=0.0~0.7).
>>>
>>
>> I've never tried using [bonds] in this context; the pull code is a better approach.
>>
>> Perhaps there's an issue with the negative dihedral?  Have you tried turning these interactions on one-by-one to see if there is a specific entry that leads to a failure?  That's the most surefire way to find it.
>>
>> -Justin
>>
>> --
>> ==================================================
>>
>> Justin A. Lemkul, Ph.D.
>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>>
>> Department of Pharmaceutical Sciences
>> School of Pharmacy
>> Health Sciences Facility II, Room 629
>> University of Maryland, Baltimore
>> 20 Penn St.
>> Baltimore, MD 21201
>>
>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
>> http://mackerell.umaryland.edu/~jalemkul
>>
>> ==================================================
>> --
>> 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.
>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================