Do you have all bugfixes applied? Early versions of AmberTools 12 could have SCF convergence problems for larger systems. If you have all patches applied, can you send me the input files so I can have a look at what might be going wrong? Thanks.
All the best,
Andy
On May 3, 2013, at 6:14 PM, me, Nancy wrote:
> Now I have tried DMSO system with PM6 in serial. It can run without
> termination by previous d-orbital error, but after it begins to run, I see
> the warning message:
>
> QMMM: WARNING!
> QMMM: Unable to achieve self consistency to the tolerances specified
> QMMM: No convergence in SCF after 1000 steps.
> QMMM: Job will continue with unconverged SCF. Warning energies
> QMMM: and forces for this step will not be accurate.
> QMMM: E = -0.2942E+07 DeltaE = -0.1385E+04 DeltaP = 0.1311E+01
> QMMM: Smallest DeltaE = -0.2096E+02 DeltaP = 0.5187E+00 Step = 235
>
> When I run the same system with PM3, it's no problem. Also I tried water
> system, PM6 works well.
>
> So anyone knows if PM6 works for system that contains DMSO? Thanks.
>
>
>
> On Thu, May 2, 2013 at 7:34 PM, me, Nancy <bwan0408.gmail.com> wrote:
>
>> Thanks, that was my question. The parallel speeds up the MM calculations
>> even though it is not parallelized well for QM. PM3 works well in
>> parallel. It just cannot run with PM6.
>>
>>
>> On Thu, May 2, 2013 at 6:25 PM, Jason Swails <jason.swails.gmail.com>wrote:
>>
>>> On Thu, May 2, 2013 at 5:48 PM, me, Nancy <bwan0408.gmail.com> wrote:
>>>
>>>> When I run QMMM simulation for DMSO system using PM6 in AMBER 12, I got
>>> the
>>>> error message "using d orbitals but the d orbital code is not
>>> parallelized,
>>>> please run in serial".
>>>>
>>>> I am wondering if there is any solution now for the problem? Thanks.
>>>>
>>>
>>> If your real question is whether or not PM6 runs in parallel now, I think
>>> the answer is no. Of course the `solution' to the problem causing this
>>> error message is to run in serial (as indicated).
>>>
>>> Semi-empirical QM does not parallelize particularly well, anyway (due to
>>> the fact that the diagonalization of the Fock matrix is rate-determining
>>> and not easily parallelized). So even if/when it is parallelized, I would
>>> not expect impressive speedups.
>>>
>>> HTH,
>>> Jason
>>>
>>> --
>>> 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
--
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 Sat May 04 2013 - 19:00:02 PDT
