Deep Neural Nework Synthetic Diagnostic Testing Code

This repository contains the code used to create and test the joint predictor DNN - \beta-VAE architecture as a synthetic diagnostic of experimental data, predicting proton energy spectra from laser inputs and backreflection diagnostic moments.

Repository content

Below are descriptions of each file or folder in this repository

TrainTestSplit.pkl

This file contains the randomly split data (80% for training and 20% for testing) in a dictionary structure.

synthdiag/

This folder contains NN classes and functions used throughout, with a further 'Readme.md' inside

Ensemble/

Contains a trained version of the SDE model, which is opened by 'SDE Performance plots & PFI.ipynb'

SDE Performance plots & PFI.ipynb

An example running of the SDE, with example flux-sorted, CE & Flux prediction, and PFI plots

PFI results.pkl

This file contains the results of the PFI test. Stored as the test takes a while to run.