Training Statistics¶

qucumber.utils.training_statistics.KL(nn_state, target_psi, space, bases=None, **kwargs)[source]¶

A function for calculating the total KL divergence.

$KL(P_{target} \vert P_{RBM}) = \sum_{x \in \mathcal{H}} P_{target}(x)\log(\frac{P_{RBM}(x)}{P_{target}(x)})$

Parameters

nn_state (qucumber.nn_states.WaveFunctionBase) – The neural network state (i.e. complex wavefunction or positive wavefunction).
target_psi (torch.Tensor or dict(str, torch.Tensor)) – The true wavefunction of the system. Can be a dictionary with each value being the wavefunction represented in a different basis, and the key identifying the basis.
space (torch.Tensor) – The basis elements of the Hilbert space of the system $\mathcal{H}$ . The ordering of the basis elements must match with the ordering of the coefficients given in target_psi.
bases (np.array(dtype=str)) – An array of unique bases. If given, the KL divergence will be computed for each basis and the average will be returned.
**kwargs – Extra keyword arguments that may be passed. Will be ignored.

Returns

The KL divergence.

Return type

float

qucumber.utils.training_statistics.NLL(nn_state, samples, space, bases=None, **kwargs)[source]¶

A function for calculating the negative log-likelihood (NLL).

Parameters

nn_state (qucumber.nn_states.WaveFunctionBase) – The neural network state (i.e. complex wavefunction or positive wavefunction).
samples (torch.Tensor) – Samples to compute the NLL on.
space (torch.Tensor) – The basis elements of the Hilbert space of the system $\mathcal{H}$ .
bases (np.array(dtype=str)) – An array of bases where measurements were taken.
**kwargs – Extra keyword arguments that may be passed. Will be ignored.

Returns

The Negative Log-Likelihood.

Return type

float

qucumber.utils.training_statistics.fidelity(nn_state, target_psi, space, **kwargs)[source]¶

Calculates the square of the overlap (fidelity) between the reconstructed wavefunction and the true wavefunction (both in the computational basis).

$F = \vert \langle \psi_{RBM} \vert \psi_{target} \rangle \vert ^2$

Parameters

nn_state (qucumber.nn_states.WaveFunctionBase) – The neural network state (i.e. complex wavefunction or positive wavefunction).
target_psi (torch.Tensor) – The true wavefunction of the system.
space (torch.Tensor) – The basis elements of the Hilbert space of the system $\mathcal{H}$ . The ordering of the basis elements must match with the ordering of the coefficients given in target_psi.
**kwargs – Extra keyword arguments that may be passed. Will be ignored.

Returns

The fidelity.

Return type

float

qucumber.utils.training_statistics.rotate_psi(nn_state, basis, space, unitaries, psi=None)[source]¶

A function that rotates the reconstructed wavefunction to a different basis.

Parameters

nn_state (qucumber.nn_states.WaveFunctionBase) – The neural network state (i.e. complex wavefunction or positive wavefunction).
basis (str) – The basis to rotate the wavefunction to.
space (torch.Tensor) – The basis elements of the Hilbert space of the system $\mathcal{H}$ .
unitaries (dict(str, torch.Tensor)) – A dictionary of (2x2) unitary operators.
psi (torch.Tensor) – A wavefunction that the user can input to override the neural network state’s wavefunction.

Returns

A wavefunction in a new basis.

Return type

torch.Tensor