How to read massive LAMMPS trajectory file using ASE? I tried
iread and read('lmtrj', index=":", parallel=True)
But could not succeed. If some one share few tricks that will be a great help.
Best,
Abhi
It helps if you provide full input and output to reproduce the problem. Please include error message with entire stack trace. Seeing as the input is huge, it would be wise to reproduce the same problem on a small file first.
I used a single trajectory file containing 108K snapshots each frame has 77K atoms (File size ~ 0.5 TB). Using
traj = read( `'lmtrj', index=":", parallel=True)
Even though I used node having 1 TB RAM. I got an error “out of memory”.
For sake of simplicity I chopped a lmtrj file (LAMMPS dump file) into a 10K snapshot and use the same command, It took quite some time. I am not sure, only invoking parallel=True flag, ASE ensures the trajectory reading has to execute in parallel, or are there any other export commands specific to MPI need to pass?