2025/01/24 by Callum W. Duncan, Duncan, Callum W
Computer Science · Physics and Astronomy · #FOS: Physical sciences #Mechanical and Optical Resonators #Neural Networks and Applications #Quantum Physics (quant-ph) #Statistical Mechanics and Entropy
paper · pdf · doi:10.48550/arxiv.2501.14881
openalex publication_date 2025/01/24 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Counterdiabatic driving, which enforces adiabatic evolution in arbitrary timescales, can be realised by engineering a Floquet Hamiltonian which oscillates between the Hamiltonian and its derivative requiring no additional control terms. However, the coefficients of the Floquet Hamiltoinan require knowledge of the counterdiabatic terms, which can be difficult to derive outside of a limited set of examples. We introduce a new hybrid technique for the control of quantum systems, Counterdiabatic-influenced Floquet-engineering or CAFFEINE for short. CAFFEINE parameterises the Floquet Hamiltonian for counterdiabatic driving and utilises numerical quantum optimal control in order to obtain the desired target state. This removes the need to both obtain and implement counterdiabatic terms, however, it does require the ability to quickly oscillate each term in the Hamiltonian. If this oscillation is possible, then CAFFEINE provides a framework to implement quantum annealing protocols and general quantum state preparation. We illustrate this through optimisation of two numerical examples of preparing a Bell state with two qubits and performing annealing protocols for the one-dimensional Ising model. Beyond this, we also illustrate CAFFEINE's capabilities to learn the counterdiabatic terms, which can potentially be used as a probe of quantum chaos and the geometry of quantum dynamics.