Stack of lammps simulatins to run, which can be updated dynamically.
- The command list is cleaned upon reset of the stack. DON'T erease lines of this file, just add them in the end.
- You can see what's going on in the file logSimuStack
- The command list is automatically updated with the most recent restart file
- The restart input file is updated with the pending operations
~$ make
~$ ./simuStack &
Edit the file simuList and add lines with this pattern:
#cd path/to/simulation/directory
mpirun -n 8 lmp_openmpi < yourInputFile.in > yourOutputFile.out
Recommended: create a temporary file with the command and then
cat cmdTempFile >> path/to/simuList
If your simulation was stopped at step=140000, the restart file was written at step=130000 and your current fix started in step=120000 and had to do 50000 steps, the new list would be:
#cd path/to/simulation/directory
mpirun -n 8 lmp_openmpi < yourInputFile.restart.130000.in > yourOutputFile.restart.130000.out
and the new input file will set the run of the pending fix to 40000, and add all the rest of the fixes.
#THERMO_INFO
thermo
thermo
thermo
#END_THERMO_INFO
#OP
fix F1 etc
dump ID etc
restart 10000 etc
[...]
#OP
fix F2 etc
[...]
If two different code blocks (run command) use the same fix, let the stack know this by adding an empty comment line where a fix should be placed. This allows the proper computation of run command's steps left.
#OP
#
When a stopped simulation is found, this program searches for a file called input.restart.in that is a template with the common information for all simulation fixes. Then it adds to this template the pending operations and stores the new file in input.restart.STEP.in WITHOUT modifying the template. It is necessary to leave a template file for this program to work.
Please read and follow the tips in the Issues section.