NoiseModel#

class NoiseModel(basis_gates=None)[source]#

Bases: object

Noise model class for Aer simulators.

This class is used to represent noise model for the QasmSimulator. It can be used to construct custom noise models for simulator, or to automatically generate a basic device noise model for an IBMQ backend. See the noise module documentation for additional information.

Example: Basic device noise model

An approximate NoiseModel can be generated automatically from the properties of real device backends from the IBMQ provider using the from_backend() method.

from qiskit import IBMQ, Aer
from qiskit_aer.noise import NoiseModel

provider = IBMQ.load_account()
backend = provider.get_backend('ibmq_vigo')
noise_model = NoiseModel.from_backend(backend)
print(noise_model)

Example: Custom noise model

Custom noise models can be used by adding QuantumError to circuit gate, reset or measure instructions, and ReadoutError to measure instructions.

import qiskit_aer.noise as noise

# Error probabilities
prob_1 = 0.001  # 1-qubit gate
prob_2 = 0.01   # 2-qubit gate

# Depolarizing quantum errors
error_1 = noise.depolarizing_error(prob_1, 1)
error_2 = noise.depolarizing_error(prob_2, 2)

# Add errors to noise model
noise_model = noise.NoiseModel()
noise_model.add_all_qubit_quantum_error(error_1, ['rz', 'sx', 'x'])
noise_model.add_all_qubit_quantum_error(error_2, ['cx'])
print(noise_model)

Initialize an empty noise model.

Parameters:

basis_gates (list[str] or None) – Specify an initial basis_gates for the noise model. If None a default value of [‘id’, ‘rz’, ‘sx’, ‘cx’] is used (Default: None).

Additional Information: Errors added to the noise model will have their instruction appended to the noise model basis_gates if the instruction is in the QasmSimulator basis_gates. If the instruction is not in the QasmSimulator basis_gates it is assumed to be a label for a standard gate, and that gate should be added to the NoiseModel basis_gates either using the init method, or the add_basis_gates() method.

Attributes

basis_gates#

Return basis_gates for compiling to the noise model.

noise_instructions#

Return the set of noisy instructions for this noise model.

noise_qubits#

Return the set of noisy qubits for this noise model.

Methods

add_all_qubit_quantum_error(error, instructions, warnings=True)[source]#

Add a quantum error to the noise model that applies to all qubits.

Parameters:
  • error (QuantumError) – the quantum error object.

  • or (instructions (str or list[str]) – Instruction or list[Instruction]): the instructions error applies to.

  • warnings (bool) – Display warning if appending to an instruction that already has an error (Default: True).

Raises:

NoiseError – if the input parameters are invalid.

Additional Information:

If the error object is ideal it will not be added to the model.

add_all_qubit_readout_error(error, warnings=True)[source]#

Add a single-qubit readout error that applies measure on all qubits.

Parameters:
  • error (ReadoutError) – the quantum error object.

  • warnings (bool) – Display warning if appending to an instruction that already has an error (Default: True)

Raises:

NoiseError – if the input parameters are invalid.

Additional Information:

If the error object is ideal it will not be added to the model.

add_basis_gates(instructions)[source]#

Add additional gates to the noise model basis_gates.

This should be used to add any gates that are identified by a custom gate label in the noise model.

Parameters:

or (instructions (list[str]) – list[Instruction]): the instructions error applies to.

add_quantum_error(error, instructions, qubits, warnings=True)[source]#

Add a quantum error to the noise model.

Parameters:
  • error (QuantumError) – the quantum error object.

  • or (instructions (str or list[str]) – Instruction or list[Instruction]): the instructions error applies to.

  • qubits (Sequence[int]) – qubits instruction error applies to.

  • warnings (bool) – Display warning if appending to an instruction that already has an error (Default: True).

Raises:

NoiseError – if the input parameters are invalid.

Additional Information:

If the error object is ideal it will not be added to the model.

add_readout_error(error, qubits, warnings=True)[source]#

Add a readout error to the noise model.

Parameters:
  • error (ReadoutError) – the quantum error object.

  • qubits (list[int] or tuple[int]) – qubits instruction error applies to.

  • warnings (bool) – Display warning if appending to an instruction that already has an error [Default: True]

Raises:

NoiseError – if the input parameters are invalid.

Additional Information:

If the error object is ideal it will not be added to the model.

classmethod from_backend(backend, gate_error=True, readout_error=True, thermal_relaxation=True, temperature=0, gate_lengths=None, gate_length_units='ns', warnings=None)[source]#

Return a noise model derived from a devices backend properties.

This function generates a noise model based on:

The error (noise) parameters are tuned for each individual qubit based on the \(T_1\), \(T_2\), frequency and readout error parameters for each qubit, and the gate error and gate time parameters for each gate obtained from the device backend properties.

Note that if un-physical parameters are supplied, they are internally truncated to the theoretical bound values. For example, if \(T_2 > 2 T_1\), \(T_2\) parameter will be truncated to \(2 T_1\).

Additional Information

The noise model includes the following errors:

  • If readout_error=True include single qubit readout errors on measurements.

  • If gate_error=True and thermal_relaxation=True include:

  • If gate_error=True is True and thermal_relaxation=False:

  • If gate_error=False and thermal_relaxation=True include single-qubit thermal_relaxation_errors() on each qubits participating in a multi-qubit gate.

For best practice in simulating a backend make sure that the circuit is compiled using the set of basis gates in the noise module by setting basis_gates=noise_model.basis_gates and using the device coupling map with coupling_map=backend.configuration().coupling_map

Specifying custom gate times

The gate_lengths kwarg can be used to specify custom gate times to add gate errors using the \(T_1\) and \(T_2\) values from the backend properties. This should be passed as a list of tuples gate_lengths=[(name, value), ...] where name is the gate name string, and value is the gate time in nanoseconds.

If a custom gate is specified that already exists in the backend properties, the gate_lengths value will override the gate time value from the backend properties. If non-default values are used gate_lengths should be a list

Parameters:
  • backend (Backend) – backend. For BackendV2, warnings options are ignored, and their default values are used.

  • gate_error (bool) – Include depolarizing gate errors (Default: True).

  • readout_error (Bool) – Include readout errors in model (Default: True).

  • thermal_relaxation (Bool) – Include thermal relaxation errors (Default: True).

  • temperature (double) – qubit temperature in milli-Kelvin (mK) for thermal relaxation errors (Default: 0).

  • gate_lengths (list) – Custom gate times for thermal relaxation errors. Used to extend or override the gate times in the backend properties (Default: None))

  • gate_length_units (str) – Time units for gate length values in gate_lengths. Can be ‘ns’, ‘ms’, ‘us’, or ‘s’ (Default: ‘ns’).

  • warnings (bool) – DEPRECATED, Display warnings (Default: None).

Returns:

An approximate noise model for the device backend.

Return type:

NoiseModel

Raises:

NoiseError – If the input backend is not valid.

classmethod from_backend_properties(backend_properties: BackendProperties, gate_error: bool = True, readout_error: bool = True, thermal_relaxation: bool = True, temperature: float = 0, gate_lengths: list | None = None, gate_length_units: str = 'ns', dt: float | None = None)[source]#

Return a noise model derived from a backend properties.

This method basically generates a noise model in the same way as from_backend(). One small difference is that the dt option is required to be set manually if you want to add thermal relaxation noises to delay instructions with durations in dt time unit. Because it is not supplied by a BackendProperties object unlike a Backend object. Note that the resulting noise model is the same as described in from_backend() so please refer to it for the details.

Parameters:
  • backend_properties (BackendProperties) – The property of backend.

  • gate_error (Bool) – Include depolarizing gate errors (Default: True).

  • readout_error (Bool) – Include readout errors in model (Default: True).

  • thermal_relaxation (Bool) – Include thermal relaxation errors (Default: True). If no t1 and t2 values are provided (i.e. None) in target for a qubit, an identity QuantumError` (i.e. effectively no thermal relaxation error) will be added to the qubit even if this flag is set to True. If no ``frequency is not defined (i.e. None) in target for a qubit, no excitation is considered in the thermal relaxation error on the qubit even with non-zero temperature.

  • temperature (double) – qubit temperature in milli-Kelvin (mK) for thermal relaxation errors (Default: 0).

  • gate_lengths (Optional[list]) – Custom gate times for thermal relaxation errors. Used to extend or override the gate times in the backend properties (Default: None))

  • gate_length_units (str) – Time units for gate length values in gate_lengths. Can be ‘ns’, ‘ms’, ‘us’, or ‘s’ (Default: ‘ns’).

  • dt (Optional[float]) – Backend sample time (resolution) in seconds (Default: None). Required to convert time unit of durations to seconds if including thermal relaxation errors on delay instructions.

Returns:

An approximate noise model for the device backend.

Return type:

NoiseModel

Raises:

NoiseError – If the input backend properties are not valid.

static from_dict(noise_dict)[source]#

Load NoiseModel from a dictionary.

Parameters:

noise_dict (dict) – A serialized noise model.

Returns:

the noise model.

Return type:

NoiseModel

Raises:

NoiseError – if dict cannot be converted to NoiseModel.

is_ideal()[source]#

Return True if the noise model has no noise terms.

reset()[source]#

Reset the noise model.

to_dict(serializable=False)[source]#

Return the noise model as a dictionary.

Parameters:

serializable (bool) – if True, return a dict containing only types that can be serializable by the stdlib json module.

Returns:

a dictionary for a noise model.

Return type:

dict