Re: [ESPResSo-users] AdResS - state of the art

[Christoph Junghans]

2013/11/21 Jakub Krajniak <address@hidden>:
> On Thu, Nov 14, 2013 at 4:17 PM, Christoph Junghans <address@hidden>
> wrote:
>>
>> Hi Jakub,
>>
> (...)
>>
>> To get a fair answer, you should also ask the same question on
>> espresso-pp mailing list!
>>
>> When Simon and I wrote the AdResS code for ESPResSo, we had to fit it
>> into an existing MD package, while ESPResSo++  was designed with
>> having AdResS in mind. So, like Olaf said,  ESPResSo has a
>> proof-of-concept implementation, which is an extension to normal MD
>> while in ESPResSo++ AdResS is part of the MD core.
>>
>> The AdResS code in ESPResSo hasn't been used very extensively in
>> recent years due to the fact that the center of AdResS development,
>> MPI-P, is also developing ESPResSo++. And because AdResS is also part
>> of Gromacs 4.6, which will be most likely your best choice, if you
>> want to use AdResS to speed-up simulations. An estimate for the
>> speedup in ESPResSo/Gromacs can be found on the slides of my 2012 talk
>> at the ESPResSo cecam tutorial.
>>
>
> Thanks for making the situation much more clear what tools to use as a
> foundation!
> I rather want to experiment with the AdResS for some other setup so Espresso
> is
> more flexible than Gromacs.
>
> One more question, related to the Espresso itself. I saw somewhere the
> comment that
> Espresso is not design to run the all-atom md simulation. In what sense is
> that? Cause
> as I briefly go through the documentation, it seems that technically it is
> possible to create
> all-atom setup with defined desired interaction and run it.
Well, technically Espresso is capable of doing pretty much the same as
Gromacs and in some cases even more.
However, Espresso has no capability to use force fields like Amber,
Charmm or Gromos automatically , so it quickly gets very cumbersome to
setup big all-atom simulations.
We always wanted to implement something like this in tcl, but we never
came to it.
Espresso++ has at least a parser for Gromacs topologies, which make it
much easier when you did an all-atom simulation in Gromacs before.

Christoph
>
> Best regards,
> Jakub Krajniak
>
> --
> PhD student at the department of computer science
> KU Leuven



-- 
Christoph Junghans
Web: http://www.compphys.de