[gmx-users] eneconv / g_energy troubles

[Pim Schravendijk]

Dear list,

I have aa problem analyzing an energy file that has been glued together 
from three ener.edr files.

The run was a 90 ns run. I write away to my energy file every 500 steps, 
for a 2 fs timestep that means every picosecond.

First I'll show the correct output for a 90 ns run:

>>
g_energy -f -e 90001 ener.edr

Reading frame  90000 time 90000.008           

Statistics over 45000001 steps [ 0.0000 thru 90000.0078 ps ], 1 data sets

Energy                      Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------------------
Temperature                 282.475    3.11509    3.11506 5.11065e-07 0.0459959
Heat Capacity Cv:       12.474 J/mol K (factor = 0.000121613)
<<

Due to general computer crashes, I also have a run which had to be
restarted twice, I use tbpconv for this purpose and at the end used
eneconv to create a glued file. Analysis with g_energy resulted in:

>>
g_energy -f fixed.edr -e 90001

Reading frame  90000 time 90000.008           

Statistics over 89500001 steps [ 0.0000 thru 90000.0078 ps ], 1 data sets

Energy                      Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------------------
Temperature                 142.034    70.8184    70.8184 3.62051e-07 0.0325846
<<

For some reason it thinks it has almost twice as many timesteps, and the
average is therefore more or less half the expected value (should have
been the same temperature!). I tried eneconv with manually given start and
end values as well, but that didn't make any difference. Analysis of the
ener.edr parts did give the correct results (but without giving rmsd
values). Also, as far as I can see by taking a quick look with gmxdump,
the fixed.edr file didn't seem to be corrupted.

I hope someone has any clues on how to make this work again so I will get 
my statistics on my whole run right.

Greetings, Pim

--
Pim Schravendijk - PhD Student
Max Planck Institute for Polymer Research
http://www.mpip-mainz.mpg.de/~schraven/