# [lammps-users] Dihedral Problem

Dear all, I am having an interesting dihedral problem which is giving me a lot of trouble. Let me explain.

About some time ago, I equilibrated a system to run nve simulations. The last step for that equilibration is given below. Please note that pressure and dihedral values. I used setting lj/coul 0 0 1. Here I was using an old version of Jan 9, 2009.
---------------- Step 140160 ----- CPU = 49938.7675 (sec) ----------------
TotEng = 80205.2346 KinEng = 27563.5497 Temp = 300.5981
PotEng = 52641.6849 E_bond = 16991.9813 E_angle = 15002.7855
E_dihed = 16910.3186 E_impro = 147.9991 E_vdwl = -2960.6385
E_coul = 44059.3724 E_long = -44949.7777 Press = -33.6011

Now, I used the restart file created by above simulation and tried to run using latest (3 Nov version). To my surprise, I find that dihedral values are changed for the same timestep and geometry. (which changes overall pressure also). All others are exactly similar. Please see below.

---------------- Step 140160 ----- CPU = 0.0000 (sec) ----------------
TotEng = 80205.2346 KinEng = 27563.5497 Temp = 300.5981
PotEng = 52641.6849 E_bond = 16991.9813 E_angle = 15002.7855
E_dihed = 32131.3013 E_impro = 147.9991 E_vdwl = -2960.6385
E_coul = 44059.3724 E_long = -44949.7777 Press = 768.3282

More surprisingly, using the latest version, when I ran it on a different machine, I get different values for the dihedral. Please see below.
---------------- Step 140160 ----- CPU = 0.0000 (sec) ----------------
TotEng = 80205.2346 KinEng = 27563.5497 Temp = 300.5981
PotEng = 52641.6849 E_bond = 16991.9813 E_angle = 15002.7855
E_dihed = 24349.9629 E_impro = 147.9991 E_vdwl = -2960.6385
E_coul = 44059.3724 E_long = -44949.7777 Press = 451.3653

I am not sure what might be causing this problem. Restartfile reading? When I converted the rst file to datafile and ran the simulation, it resulted in the same answers in both cases. I also looked at the dihedral coefficients (in the datafile, just to be sure), they are same.

When I tried to run the same file older version (lammps 22 May 2008), it gave
---------------- Step 140160 ----- CPU = 0.0000 (sec) ----------------
TotEng = 72765.5902 KinEng = 27563.5497 Temp = 300.5981
PotEng = 45202.0406 E_bond = 16991.9813 E_angle = 15002.7855
E_dihed = 16910.3186 E_impro = 147.9991 E_vdwl = -2960.6386
E_coul = 44059.3724 E_long = -44949.7777 Press = -33.6011

If you are saying you can read the same data file (converted from restart file)
with the Jan 09 version versus the current Nov 09 version and
get radically different dihedral energies, using the same input
script, then that sounds like a problem. Please send me the
data file and input script.

Steve