如何更改qiskit噪声模型的噪声属性？

Question

我的目标是创建一个噪声模型，其误差与假后端的误差成比例。例如，后端的每个错误强度都是 FakeManila 的 10%。

我从假后端中提取了噪声模型

from qiskit.providers.fake_provider import FakeManila
from qiskit.providers.aer.noise import NoiseModel
manila_backend = FakeManila()
nm_1 = NoiseModel.from_backend(manila_backend)

对于特定的门和特定的量子位（此处为“x”和 2），我可以使用

提取局部量子误差

error1 = nm_1._local_quantum_errors['x'][(2,)]
extracted_dict = error1.to_dict()

然后我可以操纵字典中的概率并将更改应用于噪声模型

new_error = error1.from_dict(extracted_dict)
nm_1._local_quantum_errors['x'][(2,)] = new_error

这适用于所有单个量子位以及该特定后端的所有量子位。然而，对于“cx”门，等效过程（甚至不改变任何概率）：

error2 = nm_1._local_quantum_errors['cx'][(3, 4)]
extracted_dict = error2.to_dict()
new_error = error2.from_dict(extracted_dict)
nm_1._local_quantum_errors['cx'][(3, 4)] = new_error

产生噪声错误：“指令未识别的错误门”。我怎样才能避免这个问题？

Answer 1

不要尝试直接操作“cx”门的现有错误对象，而是创建具有修改概率的新错误对象，然后将它们添加回噪声模型。

from qiskit.providers.fake_provider import FakeManila
from qiskit.providers.aer.noise import NoiseModel, depolarizing_error

manila_backend = FakeManila()
nm_1 = NoiseModel.from_backend(manila_backend)

scaling_factor = 0.1  # ex, scaling  down errors to 10%

for gate in ['x', 'y', 'z', 'h', 's', 't']:
    if (gate,) in nm_1._local_quantum_errors:
        error = nm_1._local_quantum_errors[gate][(0,)]  # lets assume its single qubit gate
        new_error = depolarizing_error(error.p * scaling_factor, 1)  
        nm_1._local_quantum_errors[gate][(0,)] = new_error

if (3, 4) in nm_1._local_quantum_errors['cx']:
    cx_error = nm_1._local_quantum_errors['cx'][(3, 4)]
    new_cx_error = depolarizing_error(cx_error.p * scaling_factor, 2)  
    nm_1._local_quantum_errors['cx'][(3, 4)] = new_cx_error

# check nm_1