AW: PBC BUG in coorfile + pair interaction

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Mon Jan 25 2016 - 02:33:18 CST

Any comments on this?

 

Norman Geist

 

Von: Norman Geist [mailto:norman.geist_at_uni-greifswald.de]
Gesendet: Donnerstag, 21. Januar 2016 11:33
An: 'namd-l_at_ks.uiuc.edu' <namd-l_at_ks.uiuc.edu>; 'Norman Geist'
<norman.geist_at_uni-greifswald.de>
Betreff: AW: namd-l: PBC BUG in coorfile + pair interaction

 

It could also be related to the fact, that VMDs NAMD Energy Plugin generates
a NAMD input that doesn't contain CellBasisVector statements. So NAMD
performs its domain decomposition before having the right box information.
Anyway, when actually having the box after using "coorfile read", NAMD
shouldn't fail to include all interactions within cutoff, or?

 

Norman Geist

 

Von: owner-namd-l_at_ks.uiuc.edu <mailto:owner-namd-l_at_ks.uiuc.edu>
[mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von Norman Geist
Gesendet: Donnerstag, 21. Januar 2016 10:51
An: <mailto:namd-l_at_ks.uiuc.edu> namd-l_at_ks.uiuc.edu
Betreff: namd-l: PBC BUG in coorfile + pair interaction

 

Hey there,

 

seems I've encountered a bug in how namd handles the missing pairlists when
using the coorfile command to read in frames from a dcd trajectory
(basically by using VMDs NAMD Energy plugin). It looks like NAMD doesn't
consider the periodic boundary condition to find out which atoms are in
interaction distance. This can be seen by trying to compute pair interaction
energies for two selection that do only meet across the periodic boundaries.
One will get only zeros for all energies, which will only change by choosing
high cutoff values like 100A.

 

Best wishes

 

Norman Geist

 

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:46 CST