This package features routines for the training and development of ML based inference for BES. The main objectives of the package is multiple methods to infer beam emission from plasma density and vice-versa infer plasma density based on beam emission.
Create packages that are essential to a BES inference package.:
emission --> package with scripts and codes that use deployed NN to infer emission from density. Further dev might include device-specific custom scripts: w7x, jet, east, ...
density --> package with scripts and codes that use deployed NN to infer density from emission. Further dev might include device-specific custom scripts: w7x, jet, east, ...
networks --> add the scripts and code necessary to generate and optimise NN of various flavours: elm, pinn, perception, ... and others
utility --> should be a package that holds various utility functions that prep and curate data, download it
The expectation is that packages such as networks, density and emission should have well-defined universal data input and output formats --> allowing for any possible independent deployment of each. Utility aims to provide the tools to transform input data into the required format and other features.
Create packages that are essential to a BES inference package.:
The expectation is that packages such as networks, density and emission should have well-defined universal data input and output formats --> allowing for any possible independent deployment of each. Utility aims to provide the tools to transform input data into the required format and other features.