pressure profile calculations

From: Grace Brannigan (grace_at_vitae.cmm.upenn.edu)
Date: Wed Feb 11 2009 - 16:58:06 CST

Hi everybody,

We have been running some pressure profile calculations using NAMD and would
like to suggest a potential improvement. Such calculations are often done
after the production run, and it seems that during analysis the code (in the
absence of velocity information) simply assigns velocities according to a
Maxwell-Boltzmann distribution. Any anisotropy in the assignment due to
chance will serve to increase the total surface tension (defined as Pzz -
(Pxx + Pyy)/2) - so the ideal gas contribution (as opposed to the Virial
contribution) to the surface tension will not cancel out. From my
perspective this doesn't make sense. We eventually settled on running the
analysis at close to zero temperature, since we are only interested in the
surface tension and not the individual components of the pressure tensor.
But for those who are interested in the components of the pressure tensor in
post-run analysis, using the density and the ideal gas law to calculate the
kinetic/ideal gas component, rather than generating random velocities, would
significantly cut down on the noise.

Thanks for listening,

Grace

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:50:29 CST