qcmet.devices.NoisySimulator#

class qcmet.devices.NoisySimulator(num_qubits=5, overrotation_amount=0.031415926535897934, detuning_amount=0.02617993877991494, error_1q=0.005, error_2q=0.05, t1=50000.0, t2=70000.0, **kwargs)[source]#

Noisy AerSimulator using custom noise model.

__init__(num_qubits=5, overrotation_amount=0.031415926535897934, detuning_amount=0.02617993877991494, error_1q=0.005, error_2q=0.05, t1=50000.0, t2=70000.0, **kwargs)[source]#

Initialize a noisy Aer-based simulator with a configurable noise model.

This constructor sets up a Qiskit Aer simulator that emulates realistic device imperfections. The noise model includes coherent over-rotation and detuning on single-qubit gates, depolarizing channels for single- and two-qubit gates, and optional thermal relaxation (T1/T2). The constructed NoiseModel and its basis gates are passed to the base AerSimulator.

Parameters:

  • num_qubits (int) – Number of qubits used for assigning per-qubit thermal relaxation channels and for metadata. Default: 5.

  • overrotation_amount (float) – X-axis over-rotation applied after each single-qubit sx (π/2) gate, in radians. Default: π/100.

  • detuning_amount (float) – Z rotation applied after each single-qubit sx gate to model detuning-induced phase error, in radians. Default: π/120.

  • error_1q (float) – Single-qubit depolarizing probability applied to sx gates. Must be in [0, 1]. Default: 0.005.

  • error_2q (float) – Two-qubit depolarizing probability applied to cx gates. Must be in [0, 1]. Default: 0.05.

  • t1 (float | None) – Energy relaxation time T1 in nanoseconds. If None or 0, thermal relaxation is disabled. Default: 50e3 (50 μs).

  • t2 (float | None) – Dephasing time T2 in nanoseconds. If provided larger than 2*T1, it is clipped to satisfy the physical constraint T2 ≤ 2*T1. If None or 0, thermal relaxation is disabled. Default: 70e3 (70 μs).

  • **kwargs – Additional keyword arguments passed through to the AerSimulator constructor.

Behavior:
  • Builds a NoiseModel with:

    • coherent errors after sx gates (X over-rotation and Z detuning),

    • depolarizing noise on sx (1-qubit) and cx (2-qubit) gates,

    • optional T1/T2 thermal relaxation channels on id, sx, and cx.

  • Clips t2 to 2*t1 if larger, to maintain physical consistency.

  • Passes the noise model and its basis gates to the base AerSimulator.

  • Stores num_qubits in self.properties['num_qubits'].

num_qubits#

Number of qubits.

Type:

int

overrotation_amount#

X over-rotation angle (rad) applied after sx.

Type:

float

detuning_amount#

Z detuning angle (rad) applied after sx.

Type:

float

error_1q#

Depolarizing probability for single-qubit gates.

Type:

float

error_2q#

Depolarizing probability for two-qubit gates.

Type:

float

t1#

T1 coherence time (ns).

Type:

float | None

t2#

T2 coherence time (ns), clipped to 2*T1 if needed.

Type:

float | None

thermal_relaxation#

Whether thermal relaxation channels are enabled.

Type:

bool

properties#

Includes 'num_qubits' entry set to num_qubits.

Type:

dict

Example

Create a 7-qubit noisy simulator with custom depolarizing strengths:

>>> sim = NoisySimulator(num_qubits=7, error_1q=5e-3, error_2q=5e-2)

Methods

__init__([num_qubits, overrotation_amount, ...])

Initialize a noisy Aer-based simulator with a configurable noise model.

get_properties()

Return device metadata including noise model.

noise_model()

Define custom noise model.

reverse_bitstrings(counts_dict)

Reverse bitstrings in a measurement count dictionary.

run(circuits[, num_shots, max_circs_per_job])

Execute one or more quantum circuits on the device.
noise_model()[source]#

Define custom noise model.

T1 and T2 times are selected from a random normal distribution with a mean of 50μs and 70μs respectively, and a standard deviation of 1μs for both. To ensure repeatability, a random seed is set such that the T1 times selected for each qubit remain the same.

In order to apply amplitude and phase damping noise when executing a circuit, the gate times must also be known. The following gate times are used: time for idle gate (I) is 50ns, time for Rx(π/2) gate is 50ns, for CX gate is 300ns, and time for measurement is 1000ns. There is no noise applied on the Rz(theta) as it is modelled to be a virtual gate that is applied by adding a phase to the following gates.

For all of the Rx(π/2) gates applied, after the ideal Rx(π/2) gate the following noise contributions are added: - an over-rotation around the x-axis of π/100 to simulate coherent

calibration errors,

  • a rotation about the z-axis of π/120 to simulate the coherent phase error

    occurring due to the applied pulse being detuned from the qubit frequency

  • a depolarizing noise channel to approximate effectively averaged noise in

    a large quantum circuit. The depolarizing parameter used for this gate is gamma_D = 0.0005.

For all of the 2-qubit CX gates applied, after the ideal CX gate the following noise contributions are added: - an exp^(-i*ZX*theta_zx/2) operation and an exp^(-i*ZZ*theta_zz/2) operation

on the 2-qubit subspace the CX gate acts on.

  • The parameters theta_zx and theta_zz are both set to pi/100. The zx- and zz-

    rotation axes are chosen to reproduce some of the dominant sources of coherent error when applying a cross-resonance gate in superconducting qubits

  • a depolarizing noise channel, with depolarizing parameter gamma_D = 0.005.

    It is larger the value used for single qubit gates, as two-qubit gates typically have larger average errors.

Returns:

custom noise_model

Return type:

NoiseModel