Learning to detect optical nonclassicality

This repository includes the datasets used for training and testing as well as the saved parameters. datasets: ds12_fixed: Single-mode dataset, simulated with a perfect photon-number-resolving (PNR) detector with a cutoff at $n=29$ photons ds14_PNR: Single-mode dataset, simulated with the experimentally reconstructed POVM of a detector with finite photon-number resolution. Coherent states were measured experimentally. ds15_8Bin: Single-mode dataset, simulated with the experimentally reconstructed POVM of a time-bin multiplexing scheme with $N=8$ bins in total. Coherent states were measured experimentally. ds16_6modes: 6-mode dataset, simulated with the experimentally reconstructed POVM of a detector with finite photon-number resolution. code used for data simulation Each folder contains a file state_configs.txt that lists the amplitudes of the simulated states. saved_params: contains the parameters after training on the different datasets the subfolders 1el and 2el represent the training with one and two encoding layers the subfolders 1kshots, 10kshots etc. represent the number of samples per state the subfolder dense_decoder represents the training with a model whose algebraic decoder is replaced with a dense feed-forward neural network code: PolynomialRegression EncoderOutputsAndModelsPrediction: tuples of encoder outputs and predicted labels that are used as input for the polynomial regression ResultsPolyRegression: results of the polynomial regression code used to perform the polynomial regression tests: test functions code used for set-up and training of the algebraic classifier .yaml file that allows to create the environment for the code