[gmx-users] energy drift in nve with conservative parameters

[Claudio J. Margulis]

Hi, I am trying to simulate a protein in flexible spc water and I 
believe I have very conservative parameters. I however see a drift in 
the total energy, and although it's not too large it will become large 
as I need the simulation to be several ns long. If you plot the total 
energy you see that the system is consistently heating up. I have 
equilibrated the system for more than 100 ps doing npt and before that 
using position restraints,
so everything should be fine ...

time step is 1 fs ,  1nm for rvdw, fourierspacing is 0.12 for pme , i 
update the neighbor list every step nstlist =1, etc ...
the mdp file is attached.
Do you think it's the PME? if so,  what should i use for fourier spacing 
and tolerance to make it not drift? I am running single precision though.
Thanks!!!
Claudio.

here is the output from g_energy statistics over 380 ps

Energy                      Average       RMSD     Fluct.      Drift Tot-Drift
-------------------------------------------------------------------------------
Bond                        15717.5    254.829    252.943  -0.282489 -107.205
Angle                         10366    194.383    193.899   0.125137  47.4897
Proper Dih.                 122.033    12.9951    12.9539 0.00944353   3.58383
Ryckaert-Bell.              1424.82    52.2058    52.0629 -0.0352289 -13.3694
LJ-14                       1983.58    41.7575    41.6305 -0.0297046  -11.2729
Coulomb-14                  11277.9    55.3076    53.6154   0.123927  47.0305
LJ (SR)                     23216.1    415.996    415.932 -0.0663132  -25.1659
Disper. corr.              -776.671          0          0          0    0
Coulomb (SR)                -161371    627.813    626.229   0.406744   154.36
Coulomb (LR)                 -36995    51.9416    50.1115  -0.124746   -47.3413
Potential                   -135035    251.587    251.203   0.126766    48.1079
Kinetic En.                 37008.7    245.956    245.244   0.170733    64.7933
Total Energy               -98026.1    35.0014    12.7625   0.297498    112.901


-- 
-----------------------------------------------
Claudio Javier Margulis, Ph.D.                       
Chemistry Department                               
Columbia University                                
Web http://www.chem.columbia.edu/~claudiom/        
Work phone (212) 854-8470                          
-----------------------------------------------

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: nve.mdp
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20030304/79d2ca0e/attachment.ksh>