qp.labs.phox.mmd_loss

mmd_loss(params, circuit_config, mmd_config, target_data, key=None)

Estimate MMD loss using configuration dataclasses.

Parameters:

• params (ArrayLike) – Trainable circuit parameters.

• circuit_config (CircuitConfig) – Circuit configuration used to build the expval function.

• mmd_config (MMDConfig) – Hyperparameters for the MMD computation.

• target_data (ArrayLike) – Binary target samples with shape (m, n_qubits).

• key (ArrayLike | None) – Optional runtime PRNG key override for the training loop.

Returns:

Scalar average across sigma values by default, or list of per-sigma estimates when return_per_bandwidth=True.

Return type:

jnp.ndarray | list[jnp.ndarray]

Raises:

ValueError – If effective n_samples <= 1.

code/api/api/pennylane.labs.phox.mmd_loss

 

Download Python script

 

Download Notebook

 

View on GitHub