[gmx-developers] g_dipole molecule index

[hessb at mpip-mainz.mpg.de]

> On Fri, Mar 6, 2009 at 3:00 AM,  <hessb at mpip-mainz.mpg.de> wrote:
>>> Roland Schulz wrote:
>>>> Hi,
>>>>
>>>> is there a reason why g_dipole uses a molecule index file and not an
>>>> atom index file? I think the later would be more general and would
>>>> also allow e.g. to to the analysis based on residues not molecules.
>>>>
>>>> Roland
>>>>
>>> Yes, because one is usually interested in the molecular dipole, and
>>> e.g.
>>> for polarizable molecules can make a dipole distribution. However,
>>> since
>>> the notion of molecules has disappeared from the topology the code
>>> probably needs to be updated.
>>>
>>> --
>>> David van der Spoel, Ph.D., Professor of Biology
>>
>> Hi,
>>
>> The notion of molecules has not changed at all.
>>
>> Also, g_dipoles reads atom indices, not molecule indices.
>> Roland, why do you think it reads molecule indices?
>>
>> I have changed this a while ago and I always use g_dipoles
>> without and index file and selected SOL only.
>
> Yes I didn't write it well. The index contains atom numbers but these
> are then translated into molecule index. Thus one can not select parts
> of a molecule. I would like to do the correlation of residue dipoles.
> Since g_current only does autocorrelation I think Flo's tip doesn't
> help.
>
> Roland

We could add an option to split the group into residues iso molecules.
But this seems a bit exotic to me, since residues are always
connected and there might not be a clear charge separation.

Berk