[gmx-developers] plans for GROMACS 5.0 beta December 1

[Mark Abraham]

Hi Florian,

I do remember the error estimate discussion we had then - I'd been
hoping we'd have made more progress on data structure refactoring by
now, but that is life. My email thread from then doesn't actually link
any code, but if there's a g_pme_error update that works on 4.6-era
code, then it should be trivial (for me, at least) to do a cut and
paste job on it now. Is there somewhere I can see it, please?

It would be doubly nice if we could use the logic for real-time
tuning, but I guess that remains to be seen (and likely only in the
post-5.0 time frame).

Mark

On Sat, Sep 28, 2013 at 12:14 PM, Florian Dommert
<dommert at icp.uni-stuttgart.de> wrote:
>
>
>> -----Original Message-----
>> From: gmx-developers-bounces at gromacs.org [mailto:gmx-developers-
>> bounces at gromacs.org] On Behalf Of David van der Spoel
>> Sent: Saturday, September 28, 2013 12:05 PM
>> To: gmx-developers at gromacs.org
>> Subject: Re: [gmx-developers] plans for GROMACS 5.0 beta December 1
>>
>> On 2013-09-28 11:10, Florian Dommert wrote:
>> > Hi,
>> >
>> >   just a question regarding update of tools. I sent you quite a time
> ago, an
>> > improved version of g_pme_error, which is much fast than the current
> one, is
>> > able to tune the splitting parameter and grid spacing, and is completely
>> > based on an analytical calculation. This means that the MC approach to
>> > estimate the self-error is not required anymore. The results are
> contained
>> > in my PhD thesis which is available from:
>> >
> http://elib.uni-stuttgart.de/opus/volltexte/2013/8179/pdf/Dissertation_Flori
>> > an_Dommert.pdf
>> >
>> > The corresponding part starts on p.50ff
>> >
>> > The error estimate is working for the SPME and PPPM-AD. The code is
> neither
>> > present in the master branch nor in the 4.6 branch, but I realized that
>> > gmx_pme_error is already slightly converted to C++ code, which means
> only
>> > the type casting has been changed.
>> > Are there any plans to include this new version?
>> >
>> The only way to get anything in is to port it to the master branch and
>> upload it to gerrit.gromacs.org
>
> The only reason why I am asking is, that in June there was already some
> correspondence with Mark and Rossen. It sent them the prepared code for 4.6,
> but it did not get into the branch. Now I left the university and quite busy
> with my work in industry. Actually I was just curious, because it is not my
> loss, I just wanted to remind you that there is an updated version and it
> would be very nice for Gromacs to have it, because it would be then the only
> all-atom MD package, which can do such a tuning.
>
> /Flo
>>
>> > /Flo
>> >
>> >> -----Original Message-----
>> >> From: gmx-developers-bounces at gromacs.org [mailto:gmx-developers-
>> >> bounces at gromacs.org] On Behalf Of David van der Spoel
>> >> Sent: Friday, September 27, 2013 9:13 PM
>> >> To: gmx-developers at gromacs.org
>> >> Subject: Re: [gmx-developers] plans for GROMACS 5.0 beta December 1
>> >>
>> >> On 2013-09-27 20:46, Justin Lemkul wrote:
>> >>>
>> >>>
>> >>> On 9/27/13 2:21 PM, David van der Spoel wrote:
>> >>>> On 2013-09-27 19:45, Justin Lemkul wrote:
>> >>>>>
>> >>>>>
>> >>>>> On 9/27/13 1:43 PM, David van der Spoel wrote:
>> >>>>>> I have a ton of code, most of it won't make it. But here's what
> will:
>> >>>>>>
>> >>>>>> - WAXS/SAXS refinement in mdrun
>> >>>>>> - New analysis tool for RNA
>> >>>>>>
>> >>>>>> Hopefully:
>> >>>>>> - QM/MM framework
>> >>>>>> - Gaussian output to GAFF topology including RESP (needs OpenBabel
>> >>>>>> linking though)
>> >>>>>>
>> >>>>>> If people want any new polarization stuff, let's discuss it on the
>> >>>>>> list. If
>> >>>>>> there is interest I can add my coulomb integrals code (Gaussian,
>> >>>>>> Slater, need
>> >>>>>> CLN and GMP libraries) even though the routines that actually use
> it
>> >>>>>> will not
>> >>>>>> make it before Dec 1st.
>> >>>>>>
>> >>>>>
>> >>>>> I'd be glad to have the conversation about polarization.  I
> understand
>> >>>>> you may have already had some discussions with Alex about it?
> Either
>> >>>>> way, if we can pool our efforts (via this list or via direct email),
>> >>>>> that would be great.
>> >>>>>
>> >>>>> -Justin
>> >>>>>
>> >>>> As far as I understand you were going to implement the Charmm
>> >>>> polarization
>> >>>> features into gromacs right? I have indeed chatted with Alex a few
>> >>>> times about
>> >>>> this.
>> >>>>
>> >>>
>> >>> Yes, that is the plan on our end.  I'm trying to keep up with the
> mdrun
>> >>> cleanup so that whatever I do integrates (no pun intended) with the
> new
>> >>> framework cleanly.
>> >> Good!
>> >>
>> >>>
>> >>>> Tsjerk also mentioned being interested, but I don't know what he's
>> >>>> planning.
>> >>>>
>> >>>> We have implemented interactions between distributed charges
>> >>>> (Gaussian, Slater)
>> >>>> - this can be ported straightaway to the master branch.
>> >>>>
>> >>>> Polarization in the charge on a spring or drude or shell model has
>> >>>> been in
>> >>>> gromacs since 1995, but there are still issues with slow convergence
>> >>>> of shell
>> >>>> optimization. New solutions are appreciated - but not by just upping
>> > the
>> >>>> tolerance. Thole polarization for bonded terms is also implemented
> but
>> >>>> needs
>> >>>> checking (start with the manual!). Feel free to contact me on or off
>> >>>> list.
>> >>>>
>> >>>
>> >>> Yes, I've done some simple tests with the existing algorithm.  We're
>> >>> interested in incorporating some different methods, like
>> >>> http://dx.doi.org/10.1063/1.1589749, which has been implemented in
>> NAMD
>> >>> - http://dx.doi.org/10.1021/ct600180x.
>> >>
>> >> Both these things should work out of the box!
>> >>
>> >>>
>> >>> -Justin
>> >>>
>> >>
>> >>
>> >> --
>> >> 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
>> >> --
>> >> gmx-developers mailing list
>> >> gmx-developers at gromacs.org
>> >> http://lists.gromacs.org/mailman/listinfo/gmx-developers
>> >> Please don't post (un)subscribe requests to the list. Use the
>> >> www interface or send it to gmx-developers-request at gromacs.org.
>> >
>>
>>
>> --
>> 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
>> --
>> gmx-developers mailing list
>> gmx-developers at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-developers
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-developers-request at gromacs.org.
>
> --
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.