mindquantum.algorithm.nisq.ansatz_variance

mindquantum.algorithm.nisq.ansatz_variance(ansatz: Circuit, ham: Hamiltonian, focus: str, var_range: Tuple[float, float] = (0, np.pi * 2), other_var: np.array = None, atol: float = 0.1, init_batch: int = 20, sim: Union[Simulator, str] = 'mqvector')[source]

Calculate the variance of the gradient of certain parameters of parameterized quantum circuit.

Parameters

ansatz (Circuit) – The given parameterized quantum circuit.
ham (Hamiltonian) – The objective observable.
focus (str) – Which parameters you want to check.
var_range (Tuple[float, float]) – The random range for focusing parameter. Default: (0, 2*np.pi).
other_var (numpy.array) – The value of other parameters. If None, it will be random number at every sampling. Default: None.
atol (float) – Tolerance for variance when samping. Default: 0.1.
init_batch (int) – Initial sampling size. Default: 20.
sim (Union[Simulator, str]) – Simulator you want to use. Default: mqvector.

Examples

>>> import numpy as np
>>> from mindquantum.algorithm.nisq import ansatz_variance
>>> from mindquantum.algorithm.nisq.chem import HardwareEfficientAnsatz
>>> from mindquantum.core.gates import RY, RZ, Z
>>> from mindquantum.core.operators import Hamiltonian, QubitOperator
>>> np.random.seed(42)
>>> ham = Hamiltonian(QubitOperator('Z0 Z1'))
>>> q_list = [4, 6, 8, 10]
>>> vars = []
>>> for q in q_list:
...     circ = HardwareEfficientAnsatz(q, [RY, RZ], Z, depth=50).circuit
...     vars.append(ansatz_variance(circ, ham, circ.params_name[0]))
...     print(f'qubit: {q},\t var: {vars[-1]}')
qubit: 4,        var: 0.03366677155540075
qubit: 6,        var: 0.007958129595835611
qubit: 8,        var: 0.0014247908876269244
qubit: 10,       var: 0.0006696567877430079