[gmx-developers] Qhop-implementation: integration of motion not correct

[hess at sbc.su.se]

Really plain cut-off's???
That itself is already a sin for polar systems such as water.
But since such a setup doesn't use properly buffered cut-off's,
you'll get nasty artifacts, especially if the center of geometry
of the charge groups is far off from the center of the dipole.

I actually don't see what is simpler about using plain cut-off's.
Changing cut-off to PME is all you would need to do.
But even then there might still be some buffering issue.
Using pme-switch with a larger buffer would solve this.
But reaction-field-zero with a large buffer would also do.

These issues are the reason why I would like to have buffered
cut-off by default in Gromacs, but unfortunately that will cost
us some performance. I'm working on that though.

Berk

> I'm simulatimg with cut-offs to keep things simple.
>
> I tried calculating the diffusion coefficient with g_msd -rmcomm too,
> but that didn't make much of a difference.
>
> It's true that thw center of geometry is shifted away from the dipole
> center, but I don't see how this would affect the simulations other than
> that the "effective" cut-off would be increased slightly. The extra
> particles have q=0 and invmass=0. I don't expect them to be subject to
> any force under these conditions, thus not propagating any force to the
> water atoms. Or am I wrong?
>
> Thanks for helping out,
>
> Erik
>
> hess at sbc.su.se skrev 2011-01-19 09.15:
>> Have you looked at the trajectory?
>> The whole system is not moving in one direction?
>>
>> And how do you treat your electrostatics?
>> With extra sites on the molecule the center of geometry
>> will not be at the same place as in SPC and much further
>> away from the center of the dipole.
>>
>> Berk
>>
>>> I re-based it on spc myself. But the difference between spc and tip3p
>>> is
>>> much smaller than what I observe. D = 15e-5 cm^2/s.
>>>
>>> Erik
>>>
>>> Gerrit Groenhof skrev 2011-01-19 08.36:
>>>> But the masses are the same as for normal water: The watermolecule (2
>>>> protons and one oxygen) is the framework on which the virtuals are
>>>> constructed (unless something has changed here as well).
>>>>
>>>> However, I remember the first water/hydronium hybdrid was based on
>>>> tip3p, which has a faster diffission. Can this be the source?
>>>>
>>>> Gerrit
>>>>
>>>>
>>>> On 18 Jan 2011, at 22:04, hess at sbc.su.se wrote:
>>>>
>>>>> Well, all properties that are right don't depend on the masses.
>>>>> The diffusion does depend on the masses.
>>>>> That should be a good hint.
>>>>>
>>>>> Berk
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> I am implementing the Qhop protocaol in gromacs. I use a special
>>>>>> topology for the water molecules with three extra virtual particles
>>>>>> representing the hydronium hydrogens. If a molecule is in the water
>>>>>> state the nonbonded parameters for the virtual particles are turned
>>>>>> off
>>>>>> (q=0, LJ-parameters are always zero). Thus, you'd expect them to
>>>>>> behave
>>>>>> just like ordinary spc, which has the same parameters for the water
>>>>>> part
>>>>>> of the molecule. However, it does not. I compared simulations with
>>>>>> my
>>>>>> special water in bulk phase (all molecules in the water state) and
>>>>>> simulations with ordinary spc. Energies, pressures and such thigs
>>>>>> are
>>>>>> almost identical between the simulations, but the diffusion
>>>>>> coefficient
>>>>>> is roughly tree times higher with my special water than in a box of
>>>>>> spc.
>>>>>> Note that there are no extra calls to do_force() or update
>>>>>> functions.
>>>>>> Do
>>>>>> any of you have any idea of why this happens?
>>>>>>
>>>>>> I'm using leap-frog with Nosé-Hoover (or no) for T-coupling and
>>>>>> Parinello-Rahman for P-coupling.
>>>>>>
>>>>>> Sincerely,
>>>>>>
>>>>>> --
>>>>>> -----------------------------------------------
>>>>>> Erik Marklund, PhD student
>>>>>> Dept. of Cell and Molecular Biology, Uppsala University.
>>>>>> Husargatan 3, Box 596,    75124 Uppsala, Sweden
>>>>>> phone:    +46 18 471 4537        fax: +46 18 511 755
>>>>>> erikm 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.
>>>
>>> --
>>> -----------------------------------------------
>>> Erik Marklund, PhD student
>>> Dept. of Cell and Molecular Biology, Uppsala University.
>>> Husargatan 3, Box 596,    75124 Uppsala, Sweden
>>> phone:    +46 18 471 4537        fax: +46 18 511 755
>>> erikm 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.
>>>
>
>
> --
> -----------------------------------------------
> Erik Marklund, PhD student
> Dept. of Cell and Molecular Biology, Uppsala University.
> Husargatan 3, Box 596,    75124 Uppsala, Sweden
> phone:    +46 18 471 4537        fax: +46 18 511 755
> erikm 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.
>