ReverseEstimatorGradient

class ReverseEstimatorGradient(derivative_type=DerivativeType.REAL)

Bases: BaseEstimatorGradient

Estimator gradients with the classically efficient reverse mode.

Note

This gradient implementation is based on statevector manipulations and scales exponentially with the number of qubits. However, for small system sizes it can be very fast compared to circuit-based gradients.

This class implements the calculation of the expectation gradient as described in [1]. By keeping track of two statevectors and iteratively sweeping through each parameterized gate, this method scales only linearly with the number of parameters.

References:

[1]: Jones, T. and Gacon, J. “Efficient calculation of gradients in classical simulations

of variational quantum algorithms” (2020). arXiv:2009.02823.

Parameters:

derivative_type (DerivativeType) – Defines whether the real, imaginary or real plus imaginary part of the gradient is returned.

Attributes

SUPPORTED_GATES = ['rx', 'ry', 'rz', 'cp', 'crx', 'cry', 'crz']

derivative_type

Return the derivative type (real, imaginary or complex).

Returns:

The derivative type.

precision

Return the precision used by the run method of the Estimator primitive. If None, the default precision of the primitive is used.

Returns:

The default precision.

Methods

run(circuits, observables, parameter_values, parameters=None, *, precision=None)

Run the job of the estimator gradient on the given circuits.

Parameters:

• circuits (Sequence[QuantumCircuit]) – The list of quantum circuits to compute the gradients.

• observables (Sequence[BaseOperator]) – The list of observables.

• parameter_values (Sequence[Sequence[float]]) – The list of parameter values to be bound to the circuit.

• parameters (Sequence[Sequence[Parameter] | None] | None) – The sequence of parameters to calculate only the gradients of the specified parameters. Each sequence of parameters corresponds to a circuit in circuits. Defaults to None, which means that the gradients of all parameters in each circuit are calculated. None in the sequence means that the gradients of all parameters in the corresponding circuit are calculated.

• precision (float | Sequence[float] | None) – Precision to be used by the underlying Estimator. If a single float is provided, this number will be used for all circuits. If a sequence of floats is provided, they will be used on a per-circuit basis. If not set, the gradient’s default precision will be used for all circuits, and if that is None (not set) then the underlying primitive’s (default) precision will be used for all circuits.

Returns:

The job object of the gradients of the expectation values. The i-th result corresponds to circuits[i] evaluated with parameters bound as parameter_values[i]. The j-th element of the i-th result corresponds to the gradient of the i-th circuit with respect to the j-th parameter.

Raises:

ValueError – Invalid arguments are given.

Return type:

AlgorithmJob