fairfolio
Saved Projects

Adaptive Calibration Scheduling for Quantum Processors via Conservative Bandits

CSEF · 2026 Computational Science (Senior Division)

Hongyi Li

Overview

Calibration is essential for the operation of quantum processors. However, each calibration process requires time and effort. Thus, the goal of this research is to rethink the calibration process as a real-time decision problem under cost constraints. Instead of fixed schedules, each decision takes both the current and the future risks into consideration. In order to make sure the algorithm does not act aggressively, I also designed a conservative policy that only allows the algorithm to act when the statistics show that the performance degrades below the baseline. The data for drifting used in this research is from AWS Braket, which is a physical superconducting quantum processor. According to the observations, there are a few distinct behaviors across channels: an increase of mean readout (1 to 0, 0 to 1) error, a burst increase in coherent gate anomaly, and shifts in two-qubit odd-parity measurements. The calibration strategy based on these drift parameters is compared with the corresponding budget adjustments using Monte Carlo simulations. Benchmark results show that within the low-budget range (e.g., mean cumulative payoff of 64.28 at budget B=10.0, 95% confidence interval [63.15, 65.41], the conservative bandit dominantly outperforms the periodic baseline. To quantify when the adaptive table maximizes value, I plot the boundary line B^(c) between the conservative table and periodic partial recalibration as a function of the recalibration cost multiplier c (here, we can interpret c as effective opportunity cost (e.g., peak-demand or reserved-time operation)). In the cost-sensitive region, the strict advantage region dramatically expands (e.g., B^_strict increases from 11 at c=1 to 53 at c=5 ). This shows the conservative policy's advantage as the opportunity cost increases. However, there is also a limitation when applying the heterogeneity stress test, in which the algorithm performs worse than the full and partial calibration. Overall, these results show that it is beneficial to treat calibration operations as making real-time sequential decisions, as it reduces cost and increases reliability.

Competition history

  • CSEF 2026 Computational Science (Senior Division) · Entry S-07-26

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: California Science & Engineering Fair public projects

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google