Fast-tracking and disentangling of qubit noise fluctuations using minimal-data averaging and hierarchical discrete fluctuation auto-segmentation

TL;DR

This paper introduces a novel framework for high-resolution, minimal-data noise characterization and auto-segmentation of overlapping qubit fluctuations, enabling better understanding and mitigation of noise in quantum computing.

Contribution

It presents a new method combining minimal measurements with hierarchical auto-segmentation to disentangle and track multiple overlapping qubit noise sources over time.

Findings

Successfully tracks qubit frequency fluctuations over hours.

Identifies fluctuation origins as charge parity and two-level systems.

Provides data to improve qubit calibration and error correction.

Abstract

Qubit noise and fluctuations of the noise over time are key factors limiting the performance of quantum computers. Characterising them with high temporal resolution is challenging due to multiple overlapping stochastic processes such as discrete jumps and continuous drifts. Hence, experiments typically probe individual sources of fluctuations rather than concurrent fluctuations caused by multiple sources. To overcome this limitation we develop a framework comprising a noise characterisation method with minimal measurements allowing high temporal resolution, combined with a hierarchical discrete fluctuation auto-segmentation tool to disentangle the overlapping fluctuations without human intervention, enabling their characterisation and tracking over long times. We show that on transmon qubits the method can track and disentangle qubit frequency fluctuations with temporal resolution of a few tens of milliseconds over hours. This enables us to identify the origins of the fluctuations as overlapping charge parity and two-level-systems switching. Beyond insights into the fluctuation origins, our method also provides information that can be used to improve qubit calibration, error mitigation and error correction.