[gmx-users] lincs with mttk

[Michael Shirts]

Good question.  Short answer, no -- LINCS doesn't play well with a
velocity verlet based pressure control algorithm.

Long answer: MTTK has ended up not being robust because you need to
solve a self consistent set of equations every timestep using the
pressure estimator, which is extremely noisy, so it crashes randomly
for no good reason.

The good news is that I've done some fairly extensive testing, and
leapfrog MD + Parrinello-Rahman + Nose-Hoover or vrescale is very
close to the correct distribution -- there's really not that much
difference for any practical purpose (I have a preprint with more
details).  The volume fluctuations are almost exactly right. So I
think that for anything that doesn't require getting free energies to
kJ/mol accuracy you are OK with that combination (just don't use
Berendsen anything).

Longer term (5.0), MTTK + shake will be removed -- its just too
numerically unstable, and the self-consistent iteration part means
it's very hard to parallelize.  MTTK will be left in, but only without
constraints.  Hopefully, for 5.0 we will also have RESPA-like
functionality, so that the bondeds can be performed at a higher
frequency than the nonbondeds, which will be another useful way to get
longer times steps. We also have planned to implement a Monte Carlo
barostat, which will give exactly the correct NPT distribution for any
integrator.

Best,
Michael

On Fri, Jul 27, 2012 at 2:01 PM, Katie Maerzke <maerzkek at gmail.com> wrote:
> Hi all -
>
> Is there any plan to get LINCS working with the MTTK barostat in the near future?
>
> Thanks
> Katie
>
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Only plain text messages are allowed!
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists