is the aMD implementation compatible with AMBER force fields

From: Thomas Evangelidis (tevang3_at_gmail.com)
Date: Wed Jun 18 2014 - 06:20:06 CDT

Greetings,

My query is a possible problem of the aMD implementation in NAMD 2.10
(CVS-2013-11-09) in the case of explicit solvent simulation with the AMBER
ff. In the 2011 paper about the aMD implementation, CHARMM ff was used.
However, in order to use an AMBER ff in NAMD I have to switch on the
LJcorrection. In that case the vdW and Potential energies printed in the
"ACCELERATED MD:" line are different from those in the "ENERGY:" line. In
contrast, when I switch off the LJcorrection flag all the energies are
identical. Have a look at some sample lines:

with LJcorrection on:

ACCELERATED MD: STEP 87543000 dV 72.2669 dVAVG 82.8525 BOND 5288.23 ANGLE
18672.4 DIHED 16735.6 IMPRP 0 ELECT -204303 VDW 15665.3 POTENTIAL -147940
ETITLE: TS BOND ANGLE DIHED
IMPRP ELECT VDW BOUNDARY MISC
  KINETIC TOTAL TEMP POTENTIAL
TOTAL3 TEMPAVG PRESSURE GPRESSURE VOLUME
PRESSAVG GPRESSAVG
ENERGY: 87543000 5288.2277 18672.4165 16735.6300
0.0000 -204302.7087 13520.8743 0.0000 1.6135
47684.6230 -102399.3237 311.5958 -150083.9467
-102080.5748 311.3023 -259.4794 -246.9799
738674.3246 -29.1724 -29.0266

with LJcorrection off:

ACCELERATED MD: STEP 87543000 dV 76.7961 dVAVG 94.015 BOND 5120.57 ANGLE
18918.2 DIHED 16759.4 IMPRP 0 ELECT -204135 VDW 15280.7 POTENTIAL -148058
ETITLE: TS BOND ANGLE DIHED
IMPRP ELECT VDW BOUNDARY MISC
  KINETIC TOTAL TEMP POTENTIAL
TOTAL3 TEMPAVG PRESSURE GPRESSURE VOLUME
  PRESSAVG GPRESSAVG
ENERGY: 87543000 5120.5713 18918.1941 16759.3518
0.0000 -204135.4881 15280.7398 0.0000 -1.3280
47291.1523 -100766.8069 309.0247 -148057.9592
-100448.3863 309.6632 -155.2256 -154.3696
726366.2601 -22.7045 -22.6922

Apparently the analytical tail correction is not considered in dV
calculation, which results to a moderate overestimation of the potential
energy and consequently to a lower boost, which is not bad. Do you reckon I
can still use these dV values for reweighting? I don't see any problem
using them, just wanted to have the expert opinion of a more experienced
user/developer.

thanks in advance,
Thomas

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:20:52 CST