Machine learning optimal control protocols for strongly-coupled open quantum systems

Quantum technologies promise to leverage quantum effects such as coherence and superpositions to achieve advantages over existing classical technologies. Yet, quantum systems are often strongly coupled to their environment, beset with complex dynamics and dissipation that degrade such effects, and hinder control and manipulation. This project addresses this challenge, seeking to optimise control protocols that mitigate against and operate in spite of significant environmental influence.

The primary aim of the project is to develop finite-time control protocols for quantum systems in strongly-coupled environments that reduce dissipation and maintain quantum coherence.

The approach will combine tensor networks -- an advanced numerical method for simulating many-body and open quantum systems -- with machine learning algorithms such as Bayesian Neural Networks. This will enable the optimisation of control protocols operating on practical timescales within highly-accurate -- yet numerically tractable -- models of the system.

Beyond the direct application of designing stabilising control protocols in quantum technologies, this will also enhance our understanding of the thermodynamics of strongly-coupled quantum systems.