Bayesian methods for optimising qubit quality factors

The development of error resistant and scalable quantum computers relies on having robust control over qubits with optimal metrics. In many qubit implementations, including semiconductor systems, these qubit metrics can be tuned by altering system control parameters. Here, we present an algorithm which enables the automatic optimisation of qubit quality factors. The algorithm performs appropriate measurements, and automatically analyses the results. It then intelligently selects the next set of device parameters to query in order to efficiently converge upon the optimum. We experimentally demonstrate the versatility of our method by optimising the properties of two different qubit parameterisations in two distinct semiconductor devices. We use Bayesian optimisation, Bayesian inference and optimal experimental design, motivated by information theory, to efficiently characterise and optimise qubit performance. This contribution represents a step towards the complete automation of the realisation and control of large scale quantum information devices.