Quantum Mpemba Effects

Updated 9 April 2026

Quantum Mpemba Effects are anomalous phenomena where quantum systems further from equilibrium relax faster than those closer, challenging conventional thermalization ideas.
Experimental and numerical studies show that both open systems with Lindblad dynamics and isolated many-body systems exhibit up to 30% faster relaxation via two-step protocols.
Leveraging controlled symmetry breaking, QPME enhances quantum state preparation and simulation efficiency, offering practical insights for speeding up thermalization and algorithm design.

The quantum Mpemba effect, and its generalization as the quantum Pontus-Mpemba effect (QPME), describes the anomalous and counterintuitive phenomenon whereby quantum systems initially farther from equilibrium—measured with respect to thermal, symmetry, or resource-theoretic criteria—can relax or 'thermalize' to their equilibrium or symmetric state faster than states that are initially closer. In quantum settings, such effects manifest both in open systems governed by dissipative (Lindblad) dynamics and in isolated many-body systems evolving unitarily, and have deep connections to integrability, symmetry, coherence, resource monotones, and quantum simulation algorithms. The QPME further demonstrates that an initial, judicious symmetry-breaking 'detour' or pre-thermalization protocol can accelerate relaxation in quantum many-body dynamics and algorithms.

1. Definition and Conceptual Framework

The quantum Mpemba effect extends the classical observation that, under certain nonequilibrium conditions, a 'hotter' (more out-of-equilibrium) system can cool faster than a 'colder' one. In the quantum regime, this phenomenon encompasses several forms:

Open Quantum Systems: For a system with density matrix $\rho(t)$ governed by a Lindblad generator or Davies map, the effect is observed when two initial states $\rho_1(0)$ and $\rho_2(0)$ , with $\rho_1(0)$ being initially farther from the steady state $\rho_\mathrm{ss}$ under a suitable metric (trace distance, non-equilibrium free energy, etc.), exhibit a crossing in relaxation: for some $t^*>0$ , $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ and $D[\rho_1(t),\rho_\mathrm{ss}] < D[\rho_2(t),\rho_\mathrm{ss}]$ for $t > t^*$ .
Isolated Many-body Quantum Systems: In unitarily evolving systems, local relaxation (e.g., of symmetry or entanglement) is probed by tracking the evolution of reduced density matrices $\rho_A(t)$ towards their equilibrium (e.g., generalized Gibbs ensemble). A quantum Mpemba effect arises when an initial state with larger symmetry breaking or greater measure of 'distance' crosses below a more symmetric (closer) state during the evolution (Ares et al., 12 Feb 2025).
Quantum Pontus-Mpemba Effect (QPME): This generalizes the above by comparing two protocols starting from the same initial state: (i) direct evolution under a symmetric Hamiltonian $\rho_1(0)$ 0, and (ii) a two-step protocol with initial transient evolution under a symmetry-breaking Hamiltonian $\rho_1(0)$ 1 (which generically does not conserve the target symmetry), followed by evolution under $\rho_1(0)$ 2. The total relaxation/convergence time under the two-step protocol can be strictly smaller than the direct protocol, manifesting an accelerated approach to symmetry or the ground state (Yu et al., 2 Sep 2025).

2. QPME: Protocols and Analytical Structure

The canonical setting for QPME is a one-dimensional spin chain with global $\rho_1(0)$ 3 symmetry, for example total magnetization. Consider two evolution protocols from a symmetry-broken initial state $\rho_1(0)$ 4, typically a tilted ferromagnetic state parameterized by a small tilt $\rho_1(0)$ 5,

$\rho_1(0)$ 6

Protocol 1 (Direct Quench): Evolve under the $\rho_1(0)$ 7-symmetric Hamiltonian $\rho_1(0)$ 8 for time $\rho_1(0)$ 9 (or imaginary time $\rho_2(0)$ 0). For small $\rho_2(0)$ 1, equilibration under $\rho_2(0)$ 2 is slow due to 'Hilbert-space imprint': the initial state overlaps only weakly with the majority of the symmetry sectors, and relaxation rates scale as $\rho_2(0)$ 3.
Protocol 2 (Two-step or QPME Protocol): First, evolve under a $\rho_2(0)$ 4-breaking Hamiltonian $\rho_2(0)$ 5 (which differs from $\rho_2(0)$ 6 by, e.g., $\rho_2(0)$ 7 in the $\rho_2(0)$ 8 term) for a time $\rho_2(0)$ 9 (or $\rho_1(0)$ 0). This mixes charge sectors, effectively 'thermalizing' the initial state's overlaps, and is followed by evolution under $\rho_1(0)$ 1 for $\rho_1(0)$ 2 (or $\rho_1(0)$ 3). The total relaxation time $\rho_1(0)$ 4 can be much less than $\rho_1(0)$ 5, with $\rho_1(0)$ 6 for fixed small $\rho_1(0)$ 7.

Analytical perturbation theory supports the result: under $\rho_1(0)$ 8, dynamics out of a near-symmetric state proceed at rates $\rho_1(0)$ 9, while the initial $\rho_\mathrm{ss}$ 0 evolution for $\rho_\mathrm{ss}$ 1 spreads amplitude across charge sectors, restoring decoherence rates $\rho_\mathrm{ss}$ 2. For small $\rho_\mathrm{ss}$ 3, robust speed-ups of up to $\rho_\mathrm{ss}$ 4 are observed when $\rho_\mathrm{ss}$ 5 is of order the dephasing time and $\rho_\mathrm{ss}$ 6 (Yu et al., 2 Sep 2025).

3. Diagnostics: Symmetry Restoration and Numerical Scaling

The main diagnostic is the entanglement asymmetry $\rho_\mathrm{ss}$ 7: the difference between the von Neumann (or Rényi) entropy of the reduced density matrix $\rho_\mathrm{ss}$ 8 and its block-diagonal, symmetry-projected form $\rho_\mathrm{ss}$ 9 (projected onto subsystems' charge sectors). $t^*>0$ 0 measures the degree of local symmetry breaking at time $t^*>0$ 1, vanishing only when the subsystem is block-diagonal in $t^*>0$ 2.

Operational Definition: Define the relaxation time as the first solution to $t^*>0$ 3. In both real and imaginary time, $t^*>0$ 4 (or the expectation value $t^*>0$ 5) reaches its equilibrium value faster under the two-step QPME protocol than in the direct protocol.
Scaling and Robustness: Exact diagonalization (up to $t^*>0$ 6) and time-evolving block decimation (TEBD) numerics (up to $t^*>0$ 7) confirm that the speed-up $t^*>0$ 8 saturates at $t^*>0$ 9 for $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 0, indicating thermodynamic-limit stability.
Universality: The effect is pronounced for small tilts $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 1, but disappears for large $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 2, where the symmetry-breaking initial state already overlaps uniformly with symmetry sectors. It is suppressed or vanishes for Néel-type (antiferromagnetic) initial states lying in a single symmetry sector.

4. Relation to Symmetry Dynamics, Many-Body Localization, and Random Circuits

QPME is part of a broader set of quantum Mpemba phenomena:

Integrable and Chaotic Systems: In integrable spin chains with $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 3 symmetry, faster restoration of symmetry (decrease of entanglement asymmetry) is achieved from more symmetry-broken initial conditions, provided the state has charge fluctuations extending across multiple sectors (Rylands et al., 2023, Yu et al., 3 Jul 2025).
Many-Body Localization (MBL): In MBL systems, even without thermalization in the usual sense, universal Mpemba-type crossings of symmetry restoration are observed for all tilted initial product states. The restoration time scale is exponentially large in subsystem size, distinguishing the MBL case from chaotic thermalizing systems (Liu et al., 2024).
Random Circuits: Charge-preserving random circuit models demonstrate that operator spreading mechanisms underlie the effect: for certain initial conditions (e.g., tilted ferromagnets), stronger symmetry breaking produces more and faster-spreading nonconserved operator strings, resulting in faster symmetry restoration and a robust QME/QPME (Turkeshi et al., 2024).

5. Variants and Extensions: Open Systems, Noise, and Resource Theories

Open Quantum Systems: Traditional strong quantum Mpemba effects involve specially engineered initial states suppressing projections onto the slowest-decaying mode of the Lindbladian (Liouvillian) superoperator, leading to exponential acceleration of relaxation (Zhang et al., 2024, Moroder et al., 2024). The QPME shares conceptual structure but arises from protocol design rather than pure initial-state engineering.
Noise-Induced and Non-Markovian Effects: Random telegraph noise can both induce and eliminate quantum Mpemba effects by introducing new relaxation modes or reordering mode overlaps, according to the extended Liouvillian spectrum (Zhao et al., 16 Jul 2025). Non-Markovian quantum dynamics allows for extreme forms of Mpemba acceleration, where suitable initial system-environment correlations can lead to instantaneous or memory-limited relaxation (Strachan et al., 2024).
Resource Theories and Quantum Complexity: The QPME paradigm extends to resource-theoretic measures of quantum complexity such as coherence, imaginarity, and non-Gaussianity. In random circuit models, preheating or intentional resource-breaking detours accelerate resource dissipation compared to direct 'cooling', universally for coherence and imaginarity, with more resourceful initial states relaxing their monotones faster—a phenomenon now termed the Pontus-Mpemba effect (Aditya et al., 26 Sep 2025).

6. Practical Applications and Impact for Quantum Algorithms

The QPME is significant for quantum simulation and computation:

Quantum State Preparation: Two-step protocols exploiting symmetry-breaking detours can accelerate the equilibration of quantum simulators, allowing for faster preparation of thermal or symmetric states (Yu et al., 2 Sep 2025).
Algorithmic Speed-up: In tensor network approaches (TEBD) and projector Monte Carlo algorithms, the required imaginary-time projection length is reduced by $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 4, directly mitigating computational cost and, in Monte Carlo methods, partly alleviating the sign problem.
Optimal State Initialization: In dissipative quantum controllers (e.g., quantum Otto refrigerators, cooling protocols), deliberate symmetry-breaking can be employed at the initialization or early evolution stage, which is then erased by subsequent symmetric evolution, achieving both rapid approach to the target state and possible enhancement in operational metrics such as cooling power (Schnepper et al., 18 Nov 2025).
General Control Paradigm: More broadly, QPME motivates a paradigm in which temporary, controlled departures from symmetry or conservation laws are used to erase bottlenecks associated with high overlaps onto slow relaxation modes, before restoring the target symmetry for rapid completion of the protocol.

7. Limitations, Open Questions, and Outlook

Conditions for QPME: The acceleration is only observed when the initial state lacks significant overlap with slowly-mixing symmetry sectors in the symmetric Hamiltonian, and when the symmetry-breaking Hamiltonian during the 'detour' broadly spreads the state over symmetry sectors.
Absence of Enhancement: For large symmetry-breaking ( $D[\rho_1(t^*),\rho_\mathrm{ss}] = D[\rho_2(t^*),\rho_\mathrm{ss}]$ 5 large) or initial states already overlapping all symmetry sectors, the effect vanishes. Similarly, for antiferromagnetic or domain-wall-type initial states without such 'Hilbert-space imprint', no speed-up is found.
Relation to Classical and Strong Quantum Mpemba: QPME is fundamentally a protocol-based (rather than purely state-based) acceleration, generalizing and complementing state-engineering-based (strong) quantum Mpemba effects in open systems, and encompassing new regimes unreachable by classical mode engineering.
Extensions and Experimental Realizations: QPME and related phenomena are now observed and verified experimentally in ion-trap simulators and NMR systems, motivating further exploration in quantum hardware, higher dimensions, systems with non-Abelian symmetry, or mixed symmetry-breaking/bath-coupling regimes.
Theoretical Frontiers: The underlying operator-spreading mechanisms, connections to the eigenstructure of non-Hermitian Liouvillians, universality classes of speed-up (e.g., scaling behavior in system size and bath coupling), and their interplay with quantum control and resource theories remain at the forefront of current research (Yu et al., 2 Sep 2025, Aditya et al., 26 Sep 2025, Ares et al., 12 Feb 2025).

In summary, the quantum Pontus-Mpemba effect establishes that brief, controlled symmetry-breaking in the early dynamics of a quantum many-body system can 'erase' initial state restrictions slowing relaxation under a symmetric Hamiltonian, thereby accelerating the approach to equilibrium or the ground state in both unitary and dissipative protocols. This introduces a new dimension to non-equilibrium quantum control and algorithm design, unifying and extending quantum Mpemba physics across isolated, open, and resource-theoretic settings (Yu et al., 2 Sep 2025).