[gmx-users] Calculate interaction energy dynamically

[Justin Lemkul]

On 7/26/13 5:21 AM, Davit Hakobyan wrote:
>> From: davhak2000 at hotmail.com
>> To: gmx-users at gromacs.org
>> Subject: RE: [gmx-users] Calculate interaction energy dynamically
>> Date: Thu, 25 Jul 2013 09:13:14 +0000
>>
>>
>>
>>> Date: Wed, 24 Jul 2013 09:35:26 -0400
>>> From: jalemkul at vt.edu
>>> To: gmx-users at gromacs.org
>>> Subject: Re: [gmx-users] Calculate interaction energy dynamically
>>>
>>>
>>>
>>> On 7/24/13 9:25 AM, Davit Hakobyan wrote:
>>>> Thank you again for your time and help.
>>>>
>>>> Performing "rerun" on the original system passes without warnings:
>>>>
>>>> The
>>>>    subsequent g_energy also runs smoothly. One gets no warning also if
>>>> passing the ordered *ordered.tpr and the new index (*ordered.ndx) files
>>>> to the "mdrun -rerun" of the original system. This indeed leaves with
>>>> the ordered trajectory file as the only problematic point which might
>>>> cause the warnings.
>>>>
>>>
>>> Good to know.
>>>
>>>> Interestingly, the ordered *.trr file
>>>> (product of trjorder command) is nearly twice as small as the original
>>>> *.trr file. Since the coordinates in the ordered *.trr file look fine
>>>> then probably it is the velocities which might be missing in it. Can the
>>>
>>> There's no "might" about it - your gmxcheck output from before shows that
>>> velocities are definitely absent, which should account for the difference in
>>> file size.
>>>
>>>>    warnings come from this issue ? In this case the g_energy should not be
>>>>    able to calculate the kinetic energy. If this is the case may one still
>>>>    rely on the numbers obtained by g_energy for the coulomb and vdw terms
>>>> (despite the warning messages) ?
>>>>
>>>
>>> This is a very weird issue.  The .trr file clearly has step and time information
>>> (from your earlier post), but the resulting .edr file from mdrun -rerun does not
>>> have the correct step number (though it appears to have time).  Your problem
>>> does not derive from anything to do with velocities, as far as I can tell.
>>> Proceed carefully and scrutinize the output to see if it indeed makes sense -
>>> you can verify energy values from snapshots that have not been ordered by
>>> creating index files that correspond to the same lipids, for instance.  This is
>>> a very non-standard exercise, so it's probably never been tested and debugged.
>>
>> The energy calculation between the nearest neighbors at a given time frame based on the original trajectory file (with help of new index file) indeed gives the same values as for the ordered trajectory file. So one can probably rely on the energy output from the ordered trajectory despite the warnings.
>>
>> In case of more information about the warnings I will write in this thread.
>>
>
> It seems that the problem appears when for the trjorder command one indicates *.trr instead of *.xtc as an output.
> When *.xtc is mentioned like "trjorder ... -o output.xtc" both the "mdrun -rerun" and g_energy subsequently pass without warnings.
>

Interesting.  That leads me to suspect that there is something wrong with the 
way trjorder writes .trr frames.  It writes .xtc files correctly, but incorrect 
step information in .trr files.  Thanks for following up.  I'll look into it and 
probably file a bug report on it.  What Gromacs version are you using?  Maybe 
you said that before, but I've forgotten.

-Justin

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

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

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