boulderopalscaleup.routines.OneQubitRecalibration
class boulderopalscaleup.routines.OneQubitRecalibration(*, transmon, gate, repetitions=None, force_rerun=False, run_rb=True, run_stark_phase_cal=True, run_voltage_bias_fine_tune=False)Parameters for running the one-qubit recalibration routine.
This is a lighter, periodic variant of OneQubitCalibration that
assumes every qubit has already been calibrated at least once. It therefore
persists the existing gate duration (no duration optimization), runs the
fine-frequency Ramsey first (falling back to a coarse Ramsey only when the
fine step fails), and starts the fine-amplitude and Stark-phase calibrations
at their maximum repetitions, decreasing them only when they cannot converge.
A successful run ends with one-qubit randomized benchmarking.
The drive amplitude is likewise persisted, with one exception: if the fine-amplitude loop cannot converge, the routine escalates once to a Power Rabi at the persisted duration to re-fit the amplitude, then re-runs the fine-amplitude loop and re-measures the Stark phase (which scales as the square of the drive amplitude and is therefore invalidated).
Variables
- transmon (str) – The reference for the transmon to target.
- gate ( "sx" or "x") – The gate to calibrate.
- repetitions (list [ int ] or None , optional) – List of repetition counts for the DRAG leakage calibration. If not provided, a set of repetitions will be chosen based on the gate.
- force_rerun (bool , optional) – Whether to run the routine regardless of the transmon’s current calibration status. Defaults to False.
- run_rb (bool , optional) – Whether to run a one-qubit randomized benchmarking experiment at the end of the routine. Defaults to True.
- run_stark_phase_cal (bool , optional) – Whether to run the Stark phase calibration step. Automatically skipped when the drive waveform duration is 32 ns or above. Defaults to True.
- run_voltage_bias_fine_tune (bool , optional) – Whether to re-tune the flux bias to the sweet spot before calibrating. Defaults to False, since the bias is usually maintained separately; when disabled the routine starts at the readout classifier.