Code written by Duncan Barlow at Universite de Bordeaux. Some code taken from other sources but cited.

Code for optimizing laser illumination configurations for multi-beam facilities.

Quick start

Change directory into "python_scripts".

To generate training data in file "Data/Data_input" with "10" examples:

 python training_data_generation.py ../Data/Data_input 10 run_type=full

The "input deck" for changing parameters is found in: "python_scripts/training_data_generation.py" within functions: "define_system_params", "define_dataset_params" and "define_scan_parameters".

To run the optimisation suite use:

 python optimize.py ../Data/Data_output 100 2 10 1 10 1 10 0 12345 ../Data/Data_input

A brief guide to the meaning is given in the table below but the feature is still underdevelopment so read the source code for a more accurate understanding. # dir iex init_type bayes_opt grad_descent random_sampler random_seed dir #python optimize.py ../Data/Data_output 100 0-2 10 0-1 10 0-1 10 0 12345 ../Data/Data_input

Additional Install

 conda create -n <write_environment_name_here> "scipy>=1.9.1" jupyterlab netcdf4 numpy
 conda activate <write_environment_name_here>
 conda install -c conda-forge healpy bayesian-optimization

To run data generation you will need:

Ifriit (University of Rochester, inverse ray tracing module) will need to be installed. Requests to acol@lle.rochester.edu. You will need the python modules: healpy and netcdf4 These can be installed via conda using:

 conda config --add channels conda-forge
 conda install netcdf4 healpy

To run the optimizers you will need:

You will need python modules: bayesian-optimization

 conda install -c conda-forge bayesian-optimization

License

This project is licensed under the MIT License - see the LICENSE.md file for details