This repository contains all the Python files that uses different integrators to solve the time-dependent Schrodinger equation (TD-SE)
and the quantum-classical Liouville's equations (QCLE). We use the tsh_revision branch of the Libra code.
The calculations here are run for a set of standard nonadiabatic model problems where different types of integrators with different timesteps
are used to solve the TD-SE. We first setup all these integrators in form of multiple recipes where detailed information are brought in the adi_integrators/recipes.py file.
The calculations are run systematically for a set of parameters in the adi_integrators/run_all_integrators_adiabatic_dynamics.py file as follows:
model: This parameter takes the values of 0, 1, 2, and 3. Each of the models are defined in the model_params in the set_recipe_v2 function.
icond: The initial condition for position and momentum.
reps: This parameter defines the representation where integrators are defined. Different integrators cases are defined in this function.
tsh_ehr: This parameter is what we use for computing the forces derived from the adiabatic surfaces.
sh_opt: This parameter defines the surface hopping option such as FSSH, GFSH, etc. Since we do the dynamics only for the adiabatic case here we set it to 0.
deco_opt: This parameter defines the decoherence option such as ID-A, mSDM, etc. Since we do the dynamics in the adiabatic case we set it to 0.
deco_time_opt: This parameter defines how to compute the decoherence time.
hop_acc: This parameter is used for hop acceptance algorithm which is used when we perform surface hopping.
nac_update: This parameter is used for how to compute the nonadiabatic couplings where multiple options are available such as explicit method, NPI, and HST.
ssy: This parameter is used for applying nuclear phase correction by Shenvi, Subotnik, and Yang.
The calculations are run simultaneously for a set of recipes using:
# Make the recipes as below so that we can feed through Python arg parser
recipes = list(product(models, iconds, reps, tsh_ehr, sh_opt, deco_opt, deco_time_opt, hop_acc, nac_update, ssy))
#======= With only 1 trajectory
for dt in [200.0, 100.0, 80.0, 50.0, 40.0, 20.0, 10.0, 8.0, 5.0, 4.0, 2.0, 1.0, 0.5, 0.1, 0.05, 0.01, 0.005, 0.001]:
nsteps = int(25000/dt)
submit_jobs('submit.slm', 'run_namd_2states_models.py', recipes, dt=dt, nsteps=nsteps, ntraj=1)
In the above part, we pass the name of the submit file and the Python script used to run the calculations. The calculations in the submit file are
run through argparser for efficient submission and without regenerating new Python files. The set of parameters are passed thorugh the argparser and in the run_namd_2states_models.py
these parameters are parsed and passed to the main function tsh_dynamics.generic_recipe.