BayesProcess is a python package for Physics informed Bayesian network inference using neural network surrogate model for matching process / variable / performance in solar cells.
To install, just clone the following repository:
pip install -r requirements.txt
https://github.com/PV-Lab/BayesProcess.git
run surrogate_model.py , with the given datasets to create the neural network surrogate for numerical PDE solver.
run Bayes.py with the saved surrogate model. This performs Bayesian network inference to map the process variable (Temperature) to material descriptors.
The package contains the following module and scripts:
| Module | Description |
|---|---|
JV_surrogate.py |
Script for training neural network JV surrogate model |
Bayes.py |
Script for Bayesian inference using MCMC |
requirements.txt |
required packages |
"Danny" Zekun Ren and Felipe Oviedo