[gmx-users] Velocity distribution (a bug?)

[Mikhail Stukan]

Dear gmx-users,

I have performed further investigation on the subject  of the velocity distribution

http://lists.gromacs.org/pipermail/gmx-users/2011-April/060316.html

and what I found looks like a sort of a bug to me.  Namely, the velocity distribution profiles are different if I simulate a thermalized wall via dimmers with one on the particles being fixed:
====================================
[ moleculetype ]
; molname       nrexcl
  ROCK            1

[ atoms ]
;id     type    resnr   residu  atom    cgnr    charge
1      PRC       1     ROCK    PRC     1       0
2      FRC       1     ROCK    FRC     2       0

[bonds]
; i j funct length        force.c.
  1 2 1       0.0          1000

====================================

energygrps               = FRC PRC

tcoupl                   = v-rescale

tc-grps                  = PRC FRC

tau_t                    = 0.5 0

ref_t                    = 300 0

freezegrps               = FRC

freezedim                = Y Y Y

energygrp_excl           = FRC FRC FRC PRC PRC PRC
===================================

or  using position_restraints  approach:

====================================
[ moleculetype ]
; molname       nrexcl
  ROCK            1

[ atoms ]
;id     type    resnr   residu  atom    cgnr    charge
1      PRC       1     ROCK    PRC     1       0

[ position_restraints ]
; iatom type   fx      fy      fz
1      1       10000     10000     10000
==================================

energygrps               = PRC
tcoupl                   = v-rescale
tc-grps                  = PRC
tau_t                    = 0.5
ref_t                    = 300
energygrp_excl           = PRC PRC

==============================

For both simulations I built the velocity distribution function using g_traj.

In both  cases the average kinetic temperature (I have calculated it from the averaged kinetic energy based on the velocity distributions) corresponds well to the one assigned to  the thermostat (300 K).

But, a) the velocity distributions for these two cases  differ from each other; b) neither  of them is Maxwellian.

In my simulations I use 4.5.3 version of GROMACS.

I am really lost and would very much appreciate any comments.

Many thanks in advance.
Mikhail.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110413/d9513c0a/attachment.html>