Code to implement numerical results in "Robust Data Encodings for Quantum Classifiers" by Ryan LaRose & Brian Coyle arXiv:
Running these functions will generate the figures seen in the above paper.
Requirements:
python3 visualise_datasets.py
- Generates Figure 4 for each 2d Dataset studied.
python3 visualise_decision_boundaries.py
- Generates Figure 6 for each encoding strategy.
python3 plots.py
- Generates Figure 8 and Figure 13 for Pauli and Measurement Noise in the single qubit classifier.
python3 encoding_learning_algorithm.py
- Generates Figure 10 in the single qubit classifier, for each dataset and encoding strategy.
python3 encoding_learning_algorithm.py
- Generates Figure 10 in the single qubit classifier, for each dataset and encoding strategy.
python3 decision_boundary_vertical.py setting 'compare' = True
- Generates Figure 11 in the single qubit classifier, for the 'vertical' dataset. Compares all encoding parameters in the noisy and noiseless case.
python3 decision_boundary_vertical.py setting:'analytic'=True, 'ideal'=True, 'noise'=True
- Generates Figure 7 in the single qubit classifier, for the 'vertical' dataset using Denseangle Encoding. Tests analytic condition deciding misclassification. Set 'encoding' = 'wavefunction_param' for results shown in Figure 14 (Appendix)
python3 fidelity_analysis_two_qubit.py
- Generates Fig 12 for the two qubit classifier on the Iris dataset, for each type of noise. NOTE: This approach does not use the Rigetti simulator, instead simulates the density matrices directly.