ABF 'Xi' log output producing strange results

From: Hugh Martin (hughtendo_at_gmail.com)
Date: Thu Aug 27 2009 - 10:30:22 CDT


I wonder if anyone else has experienced this issue: the Xi parameter for a
zCoord-1atom reaction coordinate (the Cartesian z-axis distance between atom
abf1 and abf2) is outputting incorrect values in my simulations. For
instance in one of my simulations, abf1 and abf2 are chosen as two oxygen
atoms, each belonging to a separate water molecule, these are precisely
4.4345 Angstroms apart in the z-direction at the start of my simulation,
however, in the log file the Xi output reads "TCL: ABF> Xi at timestep 0 :
104.585426303", this large value of Xi is not the result of a measurement
through the periodic image, as the waterbox is over 130 Angstroms in height,
making 104.6 Angstroms impossible in either direction.

This is a 330k atom system and seems to produce very random values for Xi
depending on the pair of atoms I choose for abf1 and abf2, sometimes being a
negative or positive value independent of whether abf1 or abf2 is higher in
the z-direction. This occurs in both NAMD2.6 and NAMD2.71b.

Any help on the matter will be greatly appreciated.

Many thanks,


This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:13 CST