Surface Hopping
Welcome to Surface Hopping, a program to perform Tully's Fewest-Switches Surface Hopping (FSSH) to simulate nonadiabatic dynamics for one-dimensional model systems.
Currently, standard decoherence-corrected FSSH is implemented with momentum adjustments, but more features will be added in future.
What you need to run the program
- python3
- numpy
- scipy
- autograd (automatic differentiation. Install using
pip install autograd)
If your system is running default python2, you can install python3 alongside it and make use of virtual enrionment. This can be easily done by Anaconda distribution, for example.
Defining the model potential
The user has to define the model potential in diabatic picture in input_pes.py in the main module. Each element of the diabatic potential energy matrix, both diagonal and off-diagonal, need to be defined individually. At the end, one needs to collect all the diabats (diagonal elements) and all the couplings (off-diagonal elements) in two separate lists. This step is important as this defines the vibronic model for the program. For a two-state example, check the input_pes.py in the main module.
Running the program
Once the potneital has been defined, the Surface Hopping program is invoked from the top level directory by simply executing:
python surface_hopping.py
The input for SH dynamics must be defined in surface_hopping.py. The keywords are self-explanatory. This run will create three output files:
output_sh_dyn
md_data
populations
The main output file (output_sh_dyn) keeps track of hopping and one can see this by doing grep Hopping output_sh_dyn (case sensitive). The md_data file contains all the relevant information on dynamics as a function of time. The populations file writes all the electronic state populations.
Running the tests
The unit tests can be run by invoking the following command in the top-level directory:
python -m unittest discover -v