[gmx-developers] Missing things in copy_QMrec?

[Gerrit Groenhof]

Hi Ake,

That is weird. the options are not supposed to have an effect on the
routines in qm_gamess.c. The only effect is that MM particles might
appear that have no charge, which should have no effect on the
computation of the wavefunction.

The option is meant to let the QM program carryout the optimization of
the QM subsystem, keeping the MM atoms frozen.
But the convergence was poor, and I no longer believe this is the right
way to optimize a structure.

I am convinced now I shuld remove these options asap.

Gerrit

Ake Sandgren wrote:
> On Fri, 2009-12-11 at 08:33 +0100, Gerrit Groenhof wrote:
>   
>> Åke Sandgren wrote:
>>
>> Hi, I think these options (optimizing and optimizing the transition
>> state) are obsolete. I have trial-and-errored with that a long time ago,
>> but the results were never satisfactory.
>>
>> To avoid furhter confusion, and possibly undocumented crashes I am
>> thinking of removing them. I think that  bTS is always false in qmcopy.
>>     
>
> I just ran a test case with and without bOPT which gave quite different
> results (due to generated input params to gamess). Without bOPT it
> failed to converge in gamess, with bOPT it managed to converge.
>
> I'm not 100% sure that that was the real reason (since i'm doing too
> many things at the same time) but it's likely at least.
>
> Anyhow, I'm cleaning up all the patches all the patches i have for both
> gamess-uk and gromacs and will send them hopefully later today.
>
> Should i send the gromacs ones to gmx-dev or is there a better place.
> The gromacs ones are mostly the qm_gamess.c from Huub with my changes.
>
>