Unraveling the Quantum Mpemba Effect on Markovian Open Quantum Systems
Rodrigo F. Saliba, Raphael C. Drumond
TL;DR
The paper investigates the quantum Mpemba effect (QME) in Markovian open quantum systems, seeking mechanisms that allow an out-of-equilibrium state to thermalize faster than a closer-to-equilibrium counterpart. It develops multiple viewpoints: a decoherence-free-subspace (DFS) mechanism that creates fast and slow decay channels, an extreme QME where relaxation accelerates with system size, and a trajectory-based analysis of Davies-map dynamics to illuminate how jump processes and coherences shape the effect. It also introduces a microscopic bosonic Gaussian-bath model to show how bath correlations and reduced coherences can accelerate thermalization, supporting the proposed mechanisms. The results highlight how control over dissipation channels, system size, and bath structure can engineer faster thermalization in quantum technologies, with potential impacts on quantum optics and ultracold-gas platforms.
Abstract
In recent years, the quantum Mpemba effect (QME), which occurs when an out-of-equilibrium system reaches equilibrium faster than another that is closer to equilibrium, has attracted significant attention from the scientific community as an intriguing and counterintuitive phenomenon. It generalizes its classical counterpart by extending the concept beyond temperature equilibration. This paper approaches the QME in Markovian open quantum systems from different perspectives. First, we propose a physical mechanism based on decoherence-free subspaces. Second, we show that an exponential enhancement of the decay rate toward equilibrium, scaling with system size, can be obtained, leading to an extreme version of the phenomenon in Markovian open quantum systems. Third, we study the strong Mpemba effect through the unravelings of Davies maps, revealing subtleties in the choice of figures of merit used to identify the QME. Finally, we propose a microscopic model to gain deeper insight into bath dynamics in this context.
