Nearly optimal quasienergy estimation and eigenstate preparation of time-periodic Hamiltonians by Sambe space formalism (2401.02700v1)

Abstract: Time-periodic (Floquet) systems are one of the most interesting nonequilibrium systems. As the computation of energy eigenvalues and eigenstates of time-independent Hamiltonians is a central problem in both classical and quantum computation, quasienergy and Floquet eigenstates are the important targets. However, their computation has difficulty of time dependence; the problem can be mapped to a time-independent eigenvalue problem by the Sambe space formalism, but it instead requires additional infinite dimensional space and seems to yield higher computational cost than the time-independent cases. It is still unclear whether they can be computed with guaranteed accuracy as efficiently as the time-independent cases. We address this issue by rigorously deriving the cutoff of the Sambe space to achieve the desired accuracy and organizing quantum algorithms for computing quasienergy and Floquet eigenstates based on the cutoff. The quantum algorithms return quasienergy and Floquet eigenstates with guaranteed accuracy like Quantum Phase Estimation (QPE), which is the optimal algorithm for outputting energy eigenvalues and eigenstates of time-independent Hamiltonians. While the time periodicity provides the additional dimension for the Sambe space and ramifies the eigenstates, the query complexity of the algorithms achieves the near-optimal scaling in allwable errors. In addition, as a by-product of these algorithms, we also organize a quantum algorithm for Floquet eigenstate preparation, in which a preferred gapped Floquet eigenstate can be deterministically implemented with nearly optimal query complexity in the gap. These results show that, despite the difficulty of time-dependence, quasienergy and Floquet eigenstates can be computed almost as efficiently as time-independent cases, shedding light on the accurate and fast simulation of nonequilibrium systems on quantum computers.