How can visualise the old error map of Ibm backend if one save the calibration data of the backend at that time?

Question

I have the calibration data of an Ibm backend and now I want to plot the error map of the device at that specific time. How can I do this?

Answer 1

You can load the calibration data into the corresponding fake backend. The following function does that

from qiskit.providers.fake_provider.fake_backend import FakeBackendV2
 
def load_calibrations(self, calibrations_csv):
    if self._props_dict is None:
        self._set_props_dict_from_json()
 
    prop_names = ['T1', 'T2', 'frequency', 'anharmonicity', 'readout_error', 'prob_meas0_prep1', 'prob_meas1_prep0', 'readout_length']
    gate_names = ['id', 'sx', 'x']
    import csv
    with open(calibrations_csv) as csvfile:
        reader = csv.reader(csvfile)
        for idx, row in enumerate(reader):
            if idx == 0:
                continue
            index = int(row[0])
            for node in self._props_dict['qubits'][index]:
                for prop_index, prop_name in enumerate(prop_names):
                    if node['name'] == prop_name:
                        node['value'] = float(row[prop_index + 1])
 
            # Single-qubit gate errors:
            for node in self._props_dict['gates']:
                if node['qubits'] == [index]:
                    for gate_index, gate_name in enumerate(gate_names):
                        if node['gate'] == gate_name:
                            for param in node['parameters']:
                                if param['name'] == 'gate_error':
                                    param['value'] = float(row[gate_index + 9])
 
            # CNOT error:
            for s in row[12].split(';'):
                key, val = s.strip().split(':')
                idxs = key.split('_')
                for node in self._props_dict['gates']:
                    if node['qubits'] == [int(idxs[0]), int(idxs[1])] and node['gate'] == 'cx':
                        for param in node['parameters']:
                                if param['name'] == 'gate_error':
                                    param['value'] = float(val)
 
            # CNOT gate time:
            for s in row[13].split(';'):
                key, val = s.strip().split(':')
                idxs = key.split('_')
                for node in self._props_dict['gates']:
                    if node['qubits'] == [int(idxs[0]), int(idxs[1])] and node['gate'] == 'cx':
                        for param in node['parameters']:
                                if param['name'] == 'gate_length':
                                    param['value'] = float(val)

# Monkey patch FakeBackendV2 class 
setattr(FakeBackendV2, 'load_calibrations', load_calibrations)

_{^{* Code is taken from this blog post}}

After calling this function you can use plot_error_map with this fake backend instance. Assuming IBM backend is ibm_perth, your code should be:

from qiskit.providers.fake_provider import FakePerth
from qiskit.visualization import plot_error_map

cal_file_path = '/path/to/calibrations.csv'

backend = FakePerth()
backend.load_calibrations(cal_file_path)
plot_error_map(backend)

Answer 2

One way to achieve this (for a depolarisation noise model) is to define a graph, where:

nodes => qubits

edges => probability of a noise Pauli operator.

This example was created by converting the qobj JSON data into the Python networkx format: