[gmx-developers] qmm branch

[David van der Spoel]

On 2012-12-14 11:47, Kubar, Tomas (IPC) wrote:
> I see there is a lot going on now.
>
> I have already written DFTB in Gromacs, so it is fully integrated in there (rather than a separate package) and I use the PME routines of Gromacs.
>
> I would be happy to contribute. What I do not know is if it is necessary to re-write my code in C++, or if it is fair enough to modify the interface (input/output data structures) somehow.
>
We will completely remove the old interface, however a lot of code can 
be moved to the new C++ classes. Stay tuned for a week or so to see the 
new data structures.

What we'll implement is exporting either MM point charges or MM electric 
fields to the QM system (is that what you mean with PME incorporation?), 
and vice versa, that is represent the QM system as point charges or as 
electric fields to the MM system.

Everything is not thought through completely though, so you input is 
very valuable.


> Thanks, tomas
> ________________________________________
> From: David van der Spoel [spoel at xray.bmc.uu.se]
> Sent: Friday, December 14, 2012 11:23 AM
> Cc: Discussion list for GROMACS development; Kubar, Tomas (IPC)
> Subject: Re: [gmx-developers] qmm branch
>
> On 2012-12-14 11:11, Kubar, Tomas (IPC) wrote:
>> Hallo Gerrit and David
>>
>> I am basically done with DFTB for QM/MM, including PME; I just need to make it work if one has multiple threads.
>>
>> So, I have just one question: Is anything changing with the QM/MM interface right now?
>>
> Yes it is. Any new code can only go into the master branch, and in the
> master branch a transfer to C++ is obligatory. We (read: Lee-Ping) are
> now working hard to build a framework in C++ for this - from scratch. A
> basic layout will be available in some days. The main idea is to split
> off quantum package specific code into separate classes. You are of
> course welcome to contribute!
>
> The development is being done in a separate branch for now, called qmmm.
>
> What is not clear to me is whether the DFTB code you are working on is
> included in gromacs or by using a separate package? Which one in that case?
>
>> Best, tomas
>>
>> ________________________________________
>> From: David van der Spoel [spoel at xray.bmc.uu.se]
>> Sent: Friday, December 14, 2012 9:30 AM
>> To: Groenhof Gerrit
>> Cc: Kubar, Tomas (IPC); Discussion list for GROMACS development
>> Subject: Re: [gmx-developers] qmm branch
>>
>> On 2012-12-14 09:09, Groenhof Gerrit wrote:
>>> Something else came to my mind. With Tomas Kubar (cc'ed), we worked last summer on the DFTB interface with gromacs. We worked on including the PME term into the QM/mM. To keep this, we should probably work towards including it now.
>>>
>> (moving this back to the developer list).
>>
>> Yes we need to take PME into account in some way. Right now we are
>> working on porting the existing code to a C++ framework. The main design
>> decision we made so far is to split oniom from qmmm (but implement both
>> of course) in order to make it simpler.
>>
>> Cheers,
>> --
>> David van der Spoel, Ph.D., Professor of Biology
>> Dept. of Cell & Molec. Biol., Uppsala University.
>> Box 596, 75124 Uppsala, Sweden. Phone:  +46184714205.
>> spoel at xray.bmc.uu.se    http://folding.bmc.uu.se
>>
>
>
> --
> David van der Spoel, Ph.D., Professor of Biology
> Dept. of Cell & Molec. Biol., Uppsala University.
> Box 596, 75124 Uppsala, Sweden. Phone:  +46184714205.
> spoel at xray.bmc.uu.se    http://folding.bmc.uu.se
>


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se