pygsti.extras.devices.experimentaldevice.ExperimentalDevice

pygsti.extras.devices.experimentaldevice.ExperimentalDevice#

class ExperimentalDevice(qubits, graph, gate_mapping=None)#

Bases: object

Specification of an experimental device.

Initialize an IBMQ device from qubits and connectivity info.

Parameters:

  • qubits (list) – Qubit labels

  • graph (QubitGraph) – QubitGraph depicting device connectivity.

  • gate_mapping (dict, optional) – Mapping between pyGSTi gate names (keys) and IBM native gates (values). If None, simply use {‘Gcnot’: ‘cx’} to recover legacy behavior.

Methods

__init__(qubits, graph[, gate_mapping])

Initialize an IBMQ device from qubits and connectivity info.

create_error_rates_model([caldata, ...])

Create an error rates model (OplessModel) from calibration data.

create_local_depolarizing_model([caldata, ...])

Create a LocalNoiseModel with depolarizing noise based on calibration data.

create_processor_spec([gate_names, ...])

Create a QubitProcessorSpec from user-specified gates and device connectivity.

from_legacy_device(devname)

Create a ExperimentalDevice from a legacy pyGSTi pygsti.extras.devices module.

from_qiskit_backend(backend[, gate_mapping])

Construct a ExperimentalDevice from Qiskit provider backend information.
create_error_rates_model(caldata=None, calformat='ibmq-v2019', model_type='TwirledLayers', idle_name=None)#

Create an error rates model (OplessModel) from calibration data.

Parameters:

  • caldata (dict) – Calibration data. Currently, this can be retrieved via backend.properties().to_dict().

  • calformat (One of ['ibmq-v2018', 'ibmq-v2019', 'rigetti', 'native']) – Calibration data format, defaults to ibmq-v2019. TODO: It seems this has been changed, what version are we actually on?

  • model_type (One of ['TwirledLayers', 'TwirledGates', 'AnyErrorCausesFailure', 'AnyErrorCausesRandomOutput']) – Type of OplessModel to create

  • idle_name (str) – Name for the idle gate

Return type:

OplessModel

create_local_depolarizing_model(caldata=None, calformat='ibmq-v2019', qubits=None)#

Create a LocalNoiseModel with depolarizing noise based on calibration data.

Note: this model is * NOT * suitable for optimization: it is not aware that it is a local depolarization with non-independent error rates model.

Parameters:

  • caldata (dict) – Calibration data. Currently, this can be retrieved via backend.properties().to_dict().

  • calformat (One of ['ibmq-v2018', 'ibmq-v2019', 'rigetti', 'native']) – Calibration data format, defaults to ibmq-v2019. TODO: It seems this has been changed, what version are we actually on?

  • qubits (list) – Qubit labels to include in the model

Return type:

OplessModel

create_processor_spec(gate_names=None, qubit_subset=None, subset_only=False, remove_edges=None)#

Create a QubitProcessorSpec from user-specified gates and device connectivity.

Parameters:

  • gate_names (list of str) – List of one-qubit and two-qubit gate names. If None, use the keys of self.gate_mapping.

  • qubit_subset (list) – A subset of qubits to include in the processor spec. If None, use self.qubits.

  • subset_only (bool) – Whether or not to include all the device qubits in the processor spec (False, default) or just qubit_subset (True).

  • remove_edges (list) – A list of edges to drop from the connectivity graph.

Return type:

The created QubitProcessorSpec

classmethod from_legacy_device(devname)#

Create a ExperimentalDevice from a legacy pyGSTi pygsti.extras.devices module.

Parameters:

devname (str) – Name of the pygsti.extras.devices module to use

Return type:

Initialized ExperimentalDevice

classmethod from_qiskit_backend(backend, gate_mapping=None)#

Construct a ExperimentalDevice from Qiskit provider backend information.

Provider backends can be obtained via:

IBMQ.load_account() provider = IBMQ.get_provider() # with potential optional kwargs backend = provider.get_backend(<device name>)

Parameters:

  • backend (IBMQBackend) – Backend obtained from IBMQ

  • gate_mapping (dict, optional) – Mapping between pyGSTi gate names (keys) and IBM native gates (values). If None, simply use {‘Gcnot’: ‘cx’} to recover legacy behavior.

Return type:

Initialized ExperimentalDevice