Re: velocity error, still

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Thu Jan 12 2006 - 10:41:51 CST

Hi Bo,

rigidbonds being turned on has nothing to do with whether or not you're
doing an MD simulation or equilibration; this is not a rigid
distinction. "Equilibration" simply refers to a short period of MD
(usually on the order of a few hundred ps) that you run before you begin
taking results from your system, designed to work out any initial
instabilities present in the structure and give you an equilibrium
starting structure. For a cohesive unit like a protein, for example,
equilibration should proceed until a property like MSD or energy
stabilizes (this is NOT accurate for an unbonded structure like a water
box, where the system is "equilibrated" once its MSD follows the
diffusion law). Full MD is generally run with the same or very similar
settings, but starting with the structure after equilibration so that
you do not get any artifacts from the structure's early conformational
change in your results. I would highly recommend a book like Leach's
/Molecular Modelling: Principles and Applications/ and the recent paper
on NAMD in JCC
(http://www.ks.uiuc.edu/Publications/Papers/paper.cgi?tbcode=PHIL2005)
to you in order to more fully understand what is going on here; the
tutorials are excellent technical documents and primers on MD, but by
their nature cannot cover some of the more advanced topics (especially
theoretical topics) associated with running these simulations.

On a more technical note, turning on rigidbonds activates a constraining
algorithm that stops the motion of hydrogens from being calculated fully
independent of their parent atoms. Rigidbonds allows the use of a 2 fs
timestep instead of 1 fs because normally hydrogen-heavy atom bonds
oscillate on time scales ~10 fs, and thus will become unstable if too
long a timestep is used. This is likely the cause of instability in your
simulations before, which is why I previously recommended that you use a
1 fs timestep, but turning on rigidbonds works as well. Please see the
multiple timestep parameters section of the namd manual
(http://www.ks.uiuc.edu/Research/namd/2.6b1/ug/node25.html) for a more
detailed discussion on appropriate timestep settings.

Best,
Peter

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:43:14 CST