[gmx-users] evaporating system

[Anthony Costa]

hi all,

1) i have an isolated NVT system of a few thousands atoms which
evaporates at some reasonably constant rate via the temperature
coupling. the system is a droplet and is in vacuum (no pbcs). problem
is: at some point the evaporated molecules are very far separated in
space from the bulk droplet and i end up with some severe intertia
tensor errors:

Inertia tensor (3x3):
   Inertia tensor[    0]={ 4.60394e+08, -6.62964e+06, -3.39990e+07}
   Inertia tensor[    1]={-6.62964e+06,  1.33291e+08,  7.34221e+06}
   Inertia tensor[    2]={-3.39990e+07,  7.34221e+06,  2.22036e+07}

etc.

in other codes (i'm fairly new to gromacs), a simple "outflow" (used
to doing a lot of CFD) boundary condition was possible whereby at some
distance from the center of the box, molecules were simply removed
from the system. does this exist in gromacs, or is there some other
way i might get at this issue?

2) i would also like to perform this simulation in the NVE ensemble by
initially generating velocities after the NVT simulation (i usually
warm up the droplet first via NVT). however, generating velocities
performs VERY poorly, and the simulation temperature usually blows up
within a dozen or so picoseconds, which i don't understand. the basic
structure of the simulation is identical to the well performing NVT
simulation save adding the necessary LINCS iterations. any ideas?

thanks very much,
anthony

--
Anthony B. Costa
Purdue University
Department of Chemistry
560 Oval Drive, #365
West Lafayette, IN 47907