Running QM DMD Jobs

Setting Up Calculations

To setup a QM/DMD simulation with phd3, all that is needed is a .pdb file. Calling the script setupphd.py will copy over a default phdinput.json file. You can then edit this file to make sure that you have all of the proper parameters for your calculation (see here for a detailed description of options).

Calling setupphd.py again will have it test your parameters and attempt to pass the DMD structure to the QM portion. It is highly advisable to inspect both the dmd_setup and qm_setup directories to ensure that the inConstr, coord, and control file reflect the parameters you want to use for the QM/DMD simulation. Note that if everything is successful, a new file start.pdb will be created with a reformted version of your initial pdb. This will be the structure that phd3 will use for the simulation (and has appropriately edited the phdinput.json to reflect naming and numbering changes in the pdb).

Running Jobs

Running a qm/dmd job is very simple. It only requires the OpenBabel babel command to be in the PATH environmental variable. Just call the runphd.py script with the appropriate arguments.

runphd.py -n 16 -t 24 -s /scratch/

This will run the QM/DMD calculation with 16 processors for 24 hours and use the /scratch as a scratch directory for the heavy computations (DMD and QM portions). In general, the -n command specifies the number of processors, -t the amount of time to run for before stopping, and -s the scratch directory. Note that if -t is set, and Resubmit is true (in phdinput.json), then when the time limit is reached, it will call submitphd.py with the corresponding arguments so as to continue the calculation. This should be useful since often times a 40 iteration QM/DMD will take roughly 40 CPU days to complete.

If you have setup the submit.j2 file and are working on a computing cluster, you can take advantage of the auto-queuing feature by running submitphd.py:

submitphd.py -n 16 -t 24 -sub

This will use the submit.j2 file as a template submit script to successfully string together multiple jobs to run all the QM/DMD iterations. Note that when the job finishes, it can easily be restarted (or resubmitted) with either the runphd.py or submitphd.py commands. No clean up of the directories should be required. Furthermore, in order to decrease storage space, phd3 will automatically tar the movie and mo files after the iteration is done. This reduces memory by 50% or larger depending on the system and compression efficiency.