Dear users,
I ran a simulation using lambda_test.txt . I am trying to restart the simulation from 2950000 time steps using lambda_rst.txt. I am using the lammps 17Nov/16 version.
The thermo diagnostics for restart are not exactly matching the 2950000 time step data in the log file. I have kept the number of processors same and used the default newton setting in both the restart and original simulation.
Restart thermo
Step Temp E_pair E_mol TotEng Press Volume Density
2950000 300.44816 -37677.886 44295.361 20505.773 544.40548 224319.2 1.7848003
Original run thermo
Step Temp E_pair E_mol TotEng Press Volume Density
2950000 300.44816 -37679.196 44295.361 20504.462 548.64091 224319.2 1.7848003
As you can see, the diagnostics are not exactly the same. Is this acceptable ?
Please let me know if other files are needed.
Thanks
Soumyadipta
lambda_rst.txt (2.24 KB)
lambda_test.txt (1.87 KB)
Hi Soumyadipta,
I would think that, since restart files are binary, the numbers should be completely identical if the simulation setup is completely identical. Perhaps some of the commands issued in the original script do not store their settings in the binary restart file?
Magically, only the pair energies seem different. Can you check the logs to see if cutoffs, shifts etc. are the same between the two runs?
As to whether or not the difference is acceptable: I would not feel comfortable assessing that until the reason for the difference is known.
Dear users,
I ran a simulation using lambda_test.txt . I am trying to restart the
simulation from 2950000 time steps using lambda_rst.txt. I am using the
lammps 17Nov/16 version.
The thermo diagnostics for restart are not exactly matching the 2950000
time step data in the log file. I have kept the number of processors same
and used the default newton setting in both the restart and original
simulation.
Restart thermo
Step Temp E_pair E_mol TotEng Press Volume Density
2950000 300.44816 -37677.886 44295.361 20505.773 544.40548
224319.2 1.7848003
Original run thermo
Step Temp E_pair E_mol TotEng Press Volume Density
2950000 300.44816 -37679.196 44295.361 20504.462 548.64091
224319.2 1.7848003
As you can see, the diagnostics are not exactly the same. Is this
acceptable ?
it depends. from the information you provide, it is not possible to
decode, which part of the potential energy is actually different.
please have a closer look at the output for PPPM. are the FFT grids
different?
for PPPM, the FFT grid is determined only at the beginning of a run. if
your system volume changes during the run, the grid may change and thus
cause a different in E_long.
axel.
Thanks for the help. Yes, the grids are changing which causes the slight difference in the e_long.
Soumyadipta