[gmx-users] Re: Membrane Runs Crashing

[Mirco Wahab]

Addendum:

On 19.06.2013 23:16, Mirco Wahab wrote:
> ...
> MARTINI W is a strongly associating fluid and will freeze
> trough your box in no time even when setting negative pressure
> below 300K. In newer Gromacs versions (from 4.6) there seems
> to be additional shift (?) in the potential, so the water phase
> transition took another hit and went up to over 320K (with
> verlet) and to about 315K (with shift).
> ...

I can't let this without further comment. After reading my own post
(and staring to thinking about it), I *checked back the results* of
"freezing MARTINI water" that allegedly changed it's freezing point
along different Gromacs versions.

These "freezings" really did occur - but were exclusively related
to equilibration problems, where "untercooled" (unequilibrated)
membrane patches caused violent density fluctuations in the coarse
grained water (W) on expanding on the way to their target temperature.

These high local densitiy fluctuations probably caused the Martini W
to freeze instantaneously at contact with the membrane and "blocked"
the further membrane expansion. This could be averted by choosing
higher temperatures (+20K).

**********************************************************************
  Checked again with well-equilibrated membranes in water, the Martini
  water (W) *does not freeze* above 300K with Gromacs 4.6.x.  The mem-
  brane properties  are in accord with expectations from the model do-
  cumentation.  Even with verlet integration and GPU usage.  The only
  significant difference is the speed, which is now *much* higher.
**********************************************************************

Sorry about my rant from the former posting, my information was
not correct/complete when I wrote the posting.

M.