[gmx-developers] Warnings

[hessb at mpip-mainz.mpg.de]

> hessb at mpip-mainz.mpg.de wrote:
>>> Hi,
>>>
>>> I recently saw some questionable grompp parameters and was wondering
>>> what
>>> you think about adding some more Warning messages to grompp.
>>>
>>> At the moment I could think of the following cases:
>>> - comm_grps not none or System. I can't think about a reason were it
>>> makes
>>> sense and certainly in most cases you don't want to do that. Also the
>>> code
>>> in readdir.c doesn't work correctly in that case. It doesn't compute
>>> the
>>> degrees of freedoms for these groups and then you get an error message
>>> saying vcm temp is inifinite.
>>> - DispCorr = no when using constant temperature. If I understand it
>>> correct
>>> this should always be done in that case.
>>>
>>> Regarind the thermostat settings I'm not sure whether one should give
>>> warnings, messages on a info level or a link to the thermostat wiki
>>> page
>>> to
>>> encourage people to check these parameters.
>>> The thinks which are often unclear I think are:
>>> - tc_grps: I think often system is used but the wiki page recommends
>>> Protein
>>> Non-Protein  (for the Hot-Solute Cold-Solvent problem)
>>> - Berendsen: It is not well known that it is not NVT/NPT
>>
>> Very many people use Berendsen thermo/barostats and for reasonably
>> large systems this would not affect nearly all properties that
>> one is interested in. So I don't know if we should warn the user here.
>> We could add a note that one can use the new v-rescale thermostat
>> instead of Berendsen to obtain a proper ensemble.
>>
>>> - tau_t is selected not careful
>>
>> We could generate a warning when tau_t is <= 10*delta_t.
>
> And the same for tau_p/beta..
>
> Wouldn't it be more useful to supply a few mdp files e.g. on the wiki?
> Maybe one for OPLS and one for GROMOS?
> As always, the problem is that people would use them without checking
> the contents, as people now take the mdp file from the demo and use that
> for production.

That might be a good idea.
Maybe we want to have two mdp files for OPLS.
One for optimal performance using plain cut-off's.
And one for very accurate integration with vdw=shift
and PME-switch and nstlist=-1.

Berk.