Circuit features

---

dynamarq provides an API to efficiently compute a set of circuit features tailored to dynamic circuits. These features are detailed in our arxiv paper.

Note that the values of the features may depend on the gate error rates which must be retrieved from the hardware provider. Here is an example on how to calculate circuit feature values using gate fidelity data from IBM Quantum platform.

>>> import dynamarq
>>> print(dynamarq.QiskitMetrics.get_metric_names(None))
    ...
>>> from qiskit_ibm_runtime import QiskitRuntimeService
>>> service = QiskitRuntimeService() # Make sure you have your IBM Quantum token
>>> backend = service.backend('ibm_pittsburgh')
>>> testbench = dynamarq.get_testbench()
>>> for benchmark in testbench :
        qiskit_metrics = dynamarq.QiskitMetrics(benchmark, backend, stretch_dd=False)
        metric_values = qiskit_metrics.get_metrics()
        print(benchmark.name(), metric_values)

We can compute features for Quantinuum’s Helios hardware/emulator using dynamarq.QuantinuumMetrics(benchmark) object. It provides a similar get_metrics() function.

The set of circuit features and hardware fidelity scores can be used for fitting a statistical model on a collection of benchmarks to evaluate the features that have the greatest impact on the fidelity, and also for predicting the fidelity of new circuits without running them on hardware.