[gmx-users] mdrun (5.1.2) restart doesn't recognize pull-code files given default but unspecified names

[Justin Lemkul]

On 4/12/17 2:03 AM, Mark Abraham wrote:
> Hi,
>
> On Tue, Apr 11, 2017 at 7:57 PM Christopher Neale <
> chris.neale at alum.utoronto.ca> wrote:
>
>> Dear users:
>>
>> I noticed some unexpected behaviour with gromacs 5.1.2 and wanted to check
>> here first if it's known and fixed in later versions, or if people consider
>> this behaviour to actually be desirable, before posting a redmine. It's a
>> minor issue.
>>
>> The behaviour might be desirable, since the names do differ, but these are
>> default names, so I would expect them to be accepted.
>>
>
> Indeed. It seems to me likely that the pull code was not implemented (well)
> with appending in mind, e.g. the default file naming is not set up
> properly. It's a good illustration why each mdrun feature should not be
> permitted to write its own set of output files. (I plan in the future for
> GROMACS to use the TNG format to serve as a seekable container for all such
> output, for this and performance reasons. Feedback is welcome.)
>
> The following command works the first time, but gives the listed error
>> message on subsequent calls:
>>
>> gmx mdrun -notunepme -deffnm PULL -dlb yes -npme 0 -cpt 60 -maxh 499 -cpi
>> PULL.cpt -ntmpi 4 -ntomp 6 -gpu_id 0123
>>
>> Output file appending has been requested,
>> but some output files listed in the checkpoint file PULL.cpt
>> are not present or are named differently by the current progra:m
>> output files present: PULL.log PULL.xtc PULL.edr
>> output files not present or named differently: PULL_pullx.xvg
>> PULL_pullf.xvg
>>
>> ###########################
>>
>> I can fix by specifying -px and -pf like this:
>>
>> gmx mdrun -notunepme -deffnm PULL -dlb yes -npme 0 -cpt 60 -maxh 499 -cpi
>> PULL.cpt -ntmpi 4 -ntomp 6 -gpu_id 0123 -px PULL_pullx.xvg -pf
>> PULL_pullf.xvg
>>
>> ###########################
>>
>> Doesn't really make sense to me why this would be the case.
>>
>
> I'm glad you can make it work, but would prefer to fix it, if you can share
> some files on a Redmine.
>

FYI it looks like it's related to this one:

https://redmine.gromacs.org/issues/942

-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

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