[gmx-users] tpbconv inexact restart due to nstxout!=nstenergy?

[chris.neale at utoronto.ca]

Hello,
I have a 10ns trajectory and also a 5ns trajectory restarted from the 5ns point
in the original trajectory. However, the runs diverge almost immediately (as
judged by g_dist selecting one Calpha and one water molecule).

The ensemble was NPT with PME electrostatics
nstlist             =  10
vdwtype             =  switch
rvdw_switch         =  0.9
rvdw                =  1.0
rlist               =  1.1

My commands were:
tpbconv -s orig.tpr -f orig.trr -e orig.edr -o restart.tpr -time 5000
mdrun -s restart.tpr -o restart.trr -g output.restart_mdrun  

1) Could the problem be caused by unequal save periods for energy vs. velocity
or coords? Here is what my original run used:
nstxout             =  250000
nstvout             =  250000
nstfout             =  250000
nstenergy           =  250

2) Should I have done something special with the NPT ensemble?

3) Is this caused by nstlist not equal to 1? That would be strange since nobody
updates every cycle and yet we all want proper restarts. Also my restart was a
multiple of nstlist.

Thanks for your help.
Chris.