qml-jax

Tests with PennyLane and JAX.

Description of files

• jaxopt_run.py: simple sine-function learning of a VQC with pennylane, jax.jit and jaxopt.GradientDescent. The .run() method is called on the optimizer.
• jaxopt_optimization_loop.py: same as jaxopt.py, but breaks down the optmization loop, rather than calling the run() method on the optimizer object.
• optax_full_set.py: same as jaxopt_optimization_loop.py, but uses optax.adam as optimizer and updates with the full dataset.
• optax_batch.py: same as optax_full_set.py but stochastic: uses a data iterator to batch X and y.
• optax_classification.py: classification of the iris dataset with optax. Optimization part is the same as optax_batch.py, but the data now is preprocessed and converted from NumPy to JAX.
• optax_kta.py: optimizes the kernel alignment of a quantum kernel wrt the (binary) iris dataset.