Tuning of Flux-Tunable Transmon Qubits using Bayesian Optimisation

Efficient and fast qubit tune-up is essential for scaling up quantum computers to the large number of qubits required for a quantum advantage. As the number of qubits in our devices scale, the time taken to tune our devices manually becomes impractical. Therefore, automatic tuning is required. Transmon devices have shown great promise as a way to physically implement qubits with a promising trade-off between coherence time and gate speed. Our experiment demonstrates the use of Bayesian optimisation to automatically tune flux-tunable Transmon devices. This is accomplished by optimising over the parameter output space defined by our tuning experiments, in an attempt to minimise the number of error syndromes experienced by our qubits. Our results show that probabilistic decision-making frameworks used for device-tuning and characterisation can offer a demonstrable advantage over previous methods in speed, performance and generalisability. This paves the way for more flexible decision-making methods that will enable large-scale tuning of multi-qubit devices.<!-- notionvc: 406da35f-23ba-40ad-aae3-8c1ecada34e7 -->