I actually made a mistake with my input file. I did not want to select an
acetonitrile for the QM region, that was a relic of my original system
where there were two separated species in solution. In theory, I should
have only input qmmask = ':1'. However, I made the same mistake with my
methane example (and received no errors). I'll investigate further with QM
settings and see if I can get better convergence. It seems like there's
"something" going on with d-orbitals and separated QM masks, perhaps not a
bug though. Thanks for the help!
Dan
> Hi Dan,
>
> In principle there should not be problems with large separations between
> molecules in the QM region. But Qiantao is right, if either of the two
> molecules (chloromethane or acetonitrile) is alone in the QM region the
> SCF converges quickly and the forces look OK. Also, if you choose to put
> an acetonitrile in the vicinity of the chloromethane into the QM region
> (e.g. qmmask=':1,271'), everything looks good.
>
> It is not a problem with PM6 parameters, MNDO/d shows the same problem.
> However, PM3 seems to behave well. So it might be related to the presence
> of d orbitals.
>
> What really happens is that the SCF converges to a wrong state in the
> first step. This gives very bad forces as a result of which nasty things
> happen (see also the vlimit error) and with a bad structure in the second
> step, the SCF fails to converge. As to why the SCF converges to a wrong
> state in the first step for your example, I do not have an answer.
>
> All the best,
> Andy
>
> On Jul 27, 2012, at 10:31 AM, Qiantao Wang wrote:
>
>> Hi Dan,
>>
>> I noticed your QM region includes two far apart (~10 Angstrom) molecules
>> (chloromethane and acetonitrile). This, from my experience, may cause
>> problems sometime (I haven't investigated much about this issue; other
>> QM
>> developers may know more). Is there particular reason for you to do
>> this?
>> Otherwise, if you include chloromethane only, there should not be any
>> problem, at least in my test run.
>>
>> Best,
>> Qiantao
>>
>> On Wed, Jul 25, 2012 at 2:31 PM, <DJH71.pitt.edu> wrote:
>>
>>> Hi all,
>>>
>>> I was having problems running PM6 on a system that contained chlorine
>>> atoms. I decided to experiment with a much simpler system: methane v.
>>> chloromethane. My original system was solvated in acetonitrile and
>>> water,
>>> so for consistency I solvated the methane and chloromethane as well.
>>> After
>>> taking the same steps to equilibrate the solvated methane and
>>> chloromethane, when methane was treated quantum mechanically with PM6,
>>> the
>>> simulations ran without problems. When chloromethane was treated with
>>> QM,
>>> the simulations immediately gave SCF convergence errors. I suspect it's
>>> an
>>> issue with the d-orbital parameters, but that is speculation. A more
>>> conclusive test might be to try fluoromethane as well.
>>>
>>> I've attached all input, output, prmtop and restart files for the two
>>> step
>>> equilibration as well as the input and output files for the QM
>>> dynamics.
>>> The files are in a tar-ed folder 'chloromethane.tar.gz'.
>>>
>>> It's entirely possible that this error is due to a fault on my part,
>>> but I
>>> thought it was worth pointing out for investigation
>>>
>>> Dan Hatfield
>>> _______________________________________________
>>> 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jul 31 2012 - 10:30:03 PDT
