[gmx-developers] Removal of -noaddpart option from mdrun

[Thomas Piggot]

Great thanks!

Tom

Berk Hess wrote:
> Ah, my stupid mistake.
> I did not think of the situation where the output file names requested
> from mdrun
> would be different from those stored in the checkpoint file.
> I fixed it.
> 
> Thanks for reporting this,
> 
> Berk
> 
> Thomas Piggot wrote:
>> I thought this would be the case but if I do this then I get the
>> following error:
>>
>> -------------------------------------------------------
>> Program mdrun_git_13_7_10, VERSION 4.0.99-dev-20100713-721405b
>> Source code file: gmxfio.c, line: 527
>>
>> Can not open file:
>> aniso_test_LD2.trr
>> For more information and tips for troubleshooting, please check the
>> GROMACS
>> website at http://www.gromacs.org/Documentation/Errors
>> -------------------------------------------------------
>>
>>
>> My command is:
>>
>> mdrun_git_13_7_10 -s for-md_aniso_LD.tpr -deffnm aniso_test_LD2 -cpi
>> aniso_test_LD1.cpt
>>
>> after having run the following until I got a checkpoint file:
>>
>> mdrun_git_13_7_10 -s for-md_aniso_LD.tpr -deffnm aniso_test_LD1 -cpt 1
>>
>>
>> If I run the mdrun command with -noappend then it works fine but adds
>> the part to the output files (as expected). The only way I can get
>> what I want is if I move all the aniso_test_LD1 files (apart from the
>> checkpoint) to another directory.
>>
>> Cheers
>>
>> Tom
>>
>> Berk Hess wrote:
>>> I don't understand the issue.
>>>
>>> If you request new file names, the will be no old files with matching
>>> names to append to
>>> and mdrun will by default write the requested file names without the
>>> part number added.
>>>
>>> The only thing that is not possible with the new setup, as far as I can
>>> see, is to write to
>>> the same file names as before, without appending, and have the old files
>>> backed up to #...#.
>>>
>>> Berk
>>>
>>> Thomas Piggot wrote:
>>>> This is kind of the reason I asked. For me it would be useful if I
>>>> could easily maintain the same file naming convention in GROMACS 4 as
>>>> I had with GROMACS 3. For GROMACS 4 up to the latest development
>>>> version I could just use -noappend -noaddpart when the run crashed to
>>>> control the naming to be the way I wanted.
>>>>
>>>> The only way I have found to easily do this in the development version
>>>> is to move the old files to another folder (apart from the .cpt) and
>>>> change -deffnm to the new name when restarting (without -noappend).
>>>>
>>>> Maybe the -noaddpart option could be reinstated?
>>>>
>>>> Cheers
>>>>
>>>> Tom
>>>>
>>>> chris.neale at utoronto.ca wrote:
>>>>> Further, some of us have immensely complicated management scripts and
>>>>> programs in which it is important to know the name of the expected
>>>>> output .xtc file. For example, I run my own C-based server that can
>>>>> run REMD
>>>>> on a distributed computing platform (calling gromacs to do the run
>>>>> segments) and my client-side scripts needs to know what the output
>>>>> .xtc will be
>>>>> named. In this case, I always want name.xtc given -deffnm name,
>>>>> regardless of what part number the .cpt file has stored. Also, I
>>>>> can't  allow any appending to
>>>>> any files that may exist. Backward compatibility is, in this case,
>>>>> pretty important to me. Perhaps it will be possible with some
>>>>> combinations of options and I'm just missing it, but resorting to an
>>>>> ls -ltr |tail -n 1 seems like too much of a risk for a distributed
>>>>> environment.
>>>>>
>>>>> Chris.
>>>>>
>>>>> Quoting "Justin A. Lemkul" <jalemkul at vt.edu>:
>>>>>
>>>>>> Thomas Piggot wrote:
>>>>>>> But how can I get the new files, when using -noappend, to not
>>>>>>> have   the 'part' bit in their names but just have the name as
>>>>>>> defined by   -deffnm?
>>>>>>>
>>>>>>> Maybe I am missing something obvious, sorry if this is the case.
>>>>>> Per the documentation, this isn't possible.  With -noappend, the
>>>>>> "part"
>>>>>> number is always written.
>>>>>>
>>>>>> Doesn't this create a problem if a run crashes?  If a file has
>>>>>> ".part000X." and it goes down, starting from a new checkpoint will
>>>>>> use
>>>>>> a different "part" number and thus one cannot append the output?
>>>>>> That's a bit inconvenient, unless I've understood the mechanism
>>>>>> wrong.
>>>>>>
>>>>>> -Justin
>>>>>>
>>>>>>> Tom
>>>>>>>
>>>>>>> Justin A. Lemkul wrote:
>>>>>>>> Justin A. Lemkul wrote:
>>>>>>>>> Thomas Piggot wrote:
>>>>>>>>>> Hi,
>>>>>>>>>>
>>>>>>>>>> I was wondering what had happened to the -noaddpart option of 
>>>>>>>>>> mdrun in the development version (downloaded yesterday). Has 
>>>>>>>>>> this option been removed for a particular reason?
>>>>>>>>>>
>>>>>>>>> Yes, it was removed about a month ago (6/16/2010).  The option 
>>>>>>>>> -noappend takes over.  Now mdrun looks for all necessary files 
>>>>>>>>> and will append to them, unless you tell it not to (with 
>>>>>>>>> -noappend).  If files are missing, mdrun generates an error.  
>>>>>>>>> This should be explained in the mdrun -h information.
>>>>>>>>>
>>>>>>>> ...rather, if some files are missing mdrun generates an error.
>>>>>>>> If   all files are missing, mdrun simply creates new ones with
>>>>>>>> the   desired names.
>>>>>>>>
>>>>>>>> -Justin
>>>>>>>>
>>>>>> -- 
>>>>>> ========================================
>>>>>>
>>>>>> Justin A. Lemkul
>>>>>> Ph.D. Candidate
>>>>>> ICTAS Doctoral Scholar
>>>>>> MILES-IGERT Trainee
>>>>>> Department of Biochemistry
>>>>>> Virginia Tech
>>>>>> Blacksburg, VA
>>>>>> jalemkul[at]vt.edu | (540) 231-9080
>>>>>> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
>>>>>>
>>>>>> ========================================
>>>>>> -- 
>>>>>> gmx-developers mailing list
>>>>>> gmx-developers at gromacs.org
>>>>>> http://lists.gromacs.org/mailman/listinfo/gmx-developers
>>>>>> Please don't post (un)subscribe requests to the list. Use the www
>>>>>> interface or send it to gmx-developers-request at gromacs.org.
>>>>>
> 

-- 
Dr Thomas Piggot
University of Southampton, UK.