I have relied on Gaussian's manual when implementing this:
http://www.gaussian.com/g_tech/g_ur/k_punch.htm
The input file that the interface to Gaussian creates (gau_job.inp) contains "Punch=Derivatives" in the route section. According to Gaussian this should dump gradients to fort.7. Can you run Gaussian with this input file and check any files that are created? Are you aware of any modifications during the Gaussian installation?
All the best,
Andy
PS: I am on travel this week and might not respond immediately to email.
On May 14, 2012, at 12:41 PM, Jason Swails wrote:
> So what is happening here is that fort.7 is not being written but sander expects it there.
>
> The implementors will have to look into this now, but any info you can give to help them reproduce the issue would probably help them.
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
>
> On May 14, 2012, at 9:39 AM, Marc van der Kamp <marcvanderkamp.gmail.com> wrote:
>
>> Hi Jason,
>>
>> Thanks for the reply - I made the change and recompiled.
>> I don't get an error message about fort.7 not existing, but the job stops
>> (abruptly) after writing:
>> --------------------------------------------------------------------------------
>> 4. RESULTS
>> --------------------------------------------------------------------------------
>>
>> Again, gau_job.log is present and shows that Gaussian
>> has successfully finished.
>> The fort.7 file is not present this time (which makes sense), but the
>> underlying problem still persists, it appears.
>> Again, this is the case for tests in:
>> AMBERHOME/test/qmmm_EXTERN/QMMM_MD_GAUSSIAN
>> AMBERHOME/test/qmmm_EXTERN/pure_QM_MD_GAUSSIAN/
>>
>> Thanks,
>> Marc
>>
>>
>> On 14 May 2012 14:12, Jason Swails <jason.swails.gmail.com> wrote:
>>
>>> The problem seems to be that the Amber/Gaussian interface depends on all of
>>> the gradient information being dumped to the file "fort.7", which it claims
>>> will always exist. However, I'm skeptical that this is always true.
>>>
>>> Can you modify the source code and change line 600 of the file
>>> qm2_extern_gau_module.F90 (in $AMBERHOME/src/sander) from
>>>
>>> open(iunit, file=fortfile) !Will always exist
>>>
>>> to
>>>
>>> open(iunit, file=fortfile, status='old') !Will always exist
>>>
>>> This way, if the file doesn't exist, it will bomb out rather than create
>>> the file and then immediately fail to parse it. (Really, I think all of
>>> the file opening in the external module should use the amopen() wrapper
>>> rather than calling "open" directly, since it standardizes most of this
>>> behavior).
>>>
>>> After you do this (and recompile), do you get an error message about fort.7
>>> not existing?
>>>
>>> Thanks!
>>> Jason
>>>
>>> On Mon, May 14, 2012 at 12:58 AM, Marc van der Kamp <
>>> marcvanderkamp.gmail.com> wrote:
>>>
>>>> Dear all,
>>>>
>>>> I have installed AMBER12 a little while ago, with gnu compilers gcc 4.1.2
>>>> and gfortran 4.1.2 (on a cluster running Rocks 5.3, equivalent to Red Hat
>>>> 4.1.2-50).
>>>>
>>>> I would like to try out AMBER/Gaussian, and we have g09 as well as g03.
>>>> I run into problems when running the tests in
>>>> $AMBERHOME/test/qmmm_EXTERN/QMMM_MD_GAUSSIAN.
>>>> According to the mdout, Gaussian is found (in this case g09), but sander
>>>> fails with the following:
>>>>
>>>> SANDER BOMB in subroutine read_results (qm2_extern_gau_module)
>>>> Error reading file fort.7
>>>> Will quit now
>>>>
>>>> gau_job.log is present and shows that Gaussian has successfully finished.
>>>> The fort.7 file in the directory is empty, so it is not a surprise that
>>> the
>>>> job fails.
>>>> But I don't know how to proceed to fix this. Any help is appreciated!
>>>>
>>>> BTW, the same error occurs when using g03 and in the pure_QM_MD_GAUSSIAN/
>>>> tests.
>>>>
>>>> Many thanks,
>>>> Marc
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>
>>>
>>>
>>>
>>> --
>>> Jason M. Swails
>>> Quantum Theory Project,
>>> University of Florida
>>> Ph.D. Candidate
>>> 352-392-4032
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
--
Dr. Andreas W. Goetz
Assistant Project Scientist
San Diego Supercomputer Center
Tel : +1-858-822-4771
Email: agoetz.sdsc.edu
Web : www.awgoetz.de
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon May 14 2012 - 13:00:03 PDT
