Quantum Pontus–Mpemba Acceleration

• Quantum Pontus–Mpemba Effect is a protocol-driven acceleration of quantum relaxation achieved via a two-step evolution that suppresses slow decay modes.
• It employs an intermediate stage—such as preheating or a dissipation quench—to redistribute spectral weight and filter out slow relaxation channels.
• This effect has broad applications in closed systems, open quantum systems, and quantum resource theories to optimize relaxation speeds and state convergence.

The Quantum Pontus–Mpemba Effect (QPME) denotes a protocol-induced acceleration of quantum relaxation in which a two-step evolution reaches a target state faster than a corresponding direct one-step evolution. In the formulations that explicitly use the term, the defining feature is not a comparison of different initial states under the same dynamics, as in the ordinary quantum Mpemba effect (QME), but a comparison of different dynamical routes, often starting from the same initial state and differing by an intermediate preparation, preheating, auxiliary bath, symmetry-breaking detour, or dissipation quench (Nava et al., 20 May 2025). Across the recent literature, QPME has been studied in closed random circuits and spin chains, Markovian open systems, quasiperiodic dephasing models, cavity quantum electrodynamics, and quantum resource theories, with the common mechanism being a reshaping of the system’s overlap with slow relaxation modes rather than a change in the final target itself (Yu et al., 2 Sep 2025).

1. Definition and relation to the quantum Mpemba effect

In the standard quantum Mpemba effect, two different initial states evolve under the same final dynamics, and the initially more distant state can later become closer to the target than the initially nearer one. In the resource-theoretic formulation, this is expressed as

$$M(\rho_1(0)) > M(\rho_2(0)) \quad\text{yet}\quad M(\rho_1(t)) < M(\rho_2(t)) \;\;\text{for some } t>\tau.$$

The signature is therefore a crossing of relaxation orderings under one and the same evolution law (Aditya et al., 26 Sep 2025).

By contrast, the Pontus–Mpemba formulation is intrinsically two-step. In its general Markovian statement, two copies start from the same state ${\bf S}$, one relaxes directly to the target ${\bf F}$, and the other is first driven toward an auxiliary state before being switched to the target dynamics. The effect occurs when the total preparation-plus-relaxation time satisfies

$t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$

This criterion is stricter than the ordinary Mpemba comparison because it includes the time spent creating the advantageous intermediate state (Nava et al., 20 May 2025).

A closely related formulation appears in quantum resource dynamics. There the reduced state after preheating is

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$

and the QPME condition is

$$M(\tilde{\rho}(T)) > M(\rho(T)) \quad\text{and}\quad M(\tilde{\rho}(T+\tau)) < M(\rho(T+\tau)).$$

Here the intermediate stage is interpreted as resourceful preheating, and the effect is diagnosed through later faster decay under the same free dynamics (Aditya et al., 26 Sep 2025).

Several papers also stress that the ordinary QME and QPME are conceptually distinct. One concise summary is: QME compares different initial states under the same symmetric Hamiltonian, whereas QPME compares the same initial state evolved either directly under the symmetric Hamiltonian or through a temporary asymmetric stage before returning to the symmetric evolution (Yu et al., 2 Sep 2025). This path-based definition is also used in dissipative quasiperiodic chains and in cavity QED, where the key control knob is the protocol rather than the initial preparation (Song et al., 17 Feb 2026).

2. Spectral mechanism and classification

A recurrent mechanism behind QPME is the redistribution of weight among Liouvillian or effective relaxation modes. In open-system notation, the evolution can be written as

$$\rho(t)=\rho_{\mathrm{ss}}+\sum_{n=2}^{D^2} \mathrm{Tr}\left[l_n\rho(0)\right] r_n e^{\lambda_n t},$$

with the slowest nonzero mode associated with the eigenvalue of smallest magnitude real part. The literature repeatedly emphasizes that anomalous acceleration is governed not only by the decay spectrum but by the overlap of the evolving state with the slow modes (Song et al., 17 Feb 2026).

This principle is explicit in the strong quantum Mpemba effect, where the initial state is prepared so that

$$c_1 = \mathrm{Tr}[L_1 |\mathrm{sME}\rangle\langle \mathrm{sME}|] = 0.$$

The slowest decaying mode is then absent, and the late-time relaxation is governed by the next mode, yielding an exponential speed-up (Zhang et al., 2024). Although that work focuses on the strong QME rather than the Pontus protocol, it provides a direct spectral interpretation that several QPME studies echo: the effective shortcut works because the preparation stage reduces or removes the weight on the slow channel.

The general Pontus–Mpemba theory further classifies all such effects in Markovian classical or quantum systems coupled to multiple reservoirs into three classes (Nava et al., 20 May 2025). These are summarized below.

Class	Defining auxiliary-path behavior	Interpretation
Weak type-I PME	Auxiliary trajectory first approaches the target and the intermediate state is already closer than the direct trajectory at the switching time	No essential crossing is required
Weak type-II PME	Auxiliary trajectory approaches the target, but the intermediate state is not uniformly superior to the direct trajectory	Finite-time crossing can occur
Strong PME	Auxiliary trajectory initially moves away from the target	The most counterintuitive case

The strong class is the closest analogue of the Aristotle-inspired Pontus picture: the system can first move “the wrong way” relative to the target and still win overall because the switch places it on a faster relaxation route (Nava et al., 20 May 2025). A plausible implication is that QPME is best understood as a dynamical path optimization problem rather than merely a state-ordering anomaly.

Several works make the same point in model-specific language. In dissipative quasiperiodic chains, the two-step finite-temperature preparation stage does not change the dephasing Liouvillian spectrum itself; instead it changes the coefficients $\mathrm{Tr}[l_n\rho]$, especially suppressing the overlap with the slowest mode (Song et al., 17 Feb 2026). In temporary bond-dissipation quenches, the short quench modifies the slow-mode amplitude

$$\Delta\mu_1 \approx \tau \sum_j e^{\lambda_j t_1}\alpha_j\, \mathrm{Tr}\!\left[l_1^\dagger(\mathcal{L}_1-\mathcal{L}_0)[r_j]\right],$$

so the protocol acts as a Liouvillian-mode filter that can either suppress slow modes and realize QME or enhance them and realize anti-QME (Liu et al., 6 Nov 2025).

3. Closed-system and unitary realizations

A prominent closed-system realization appears in real- and imaginary-time dynamics with respect to ${\bf S}$0-symmetry. There the protocol starts from the same tilted initial state ${\bf S}$1, evolves first under a symmetry-breaking Hamiltonian ${\bf S}$2, and then switches to the symmetric Hamiltonian ${\bf S}$3. The central result is that this two-step protocol can accelerate both thermalization in real time and convergence to the ground state in imaginary time relative to direct evolution under ${\bf S}$4 alone (Yu et al., 2 Sep 2025).

In that work the model is the disordered one-dimensional spin chain

${\bf S}$5

with ${\bf S}$6 giving the ${\bf S}$7-symmetric Hamiltonian and ${\bf S}$8 the asymmetric detour. The main real-time diagnostic is the entanglement asymmetry

${\bf S}$9

while imaginary-time acceleration is tracked through the energy expectation value. The effect is reported for smallly tilted ferromagnetic states, whereas larger tilts and tilted antiferromagnetic states suppress it (Yu et al., 2 Sep 2025).

The proposed mechanism is Hilbert subspace imprint (HSI). For small tilt, direct symmetric evolution behaves as if the state were constrained by weak asymmetry and therefore relaxes anomalously slowly. The initial asymmetric stage broadens the state across Hilbert space and destroys the HSI bottleneck, so subsequent symmetric evolution becomes more thermal-like and faster (Yu et al., 2 Sep 2025). In imaginary time the same broadening enlarges the charge distribution and helps the state reach the ground-state sector more rapidly after the switch.

Another closed many-body perspective is provided by dynamical phase transitions (DPTs). In one-dimensional interacting lattice fermions corresponding to a dissipative variant of the Gross–Neveu model, the protocol is engineered so that relaxation is deliberately routed through a DPT with a long-lived metastable regime ${\bf F}$0. The central claim is that this metastable structure can be exploited to optimize the approach to a predesignated target state (Nava et al., 11 Sep 2025). The paper distinguishes genuine DPT pathways from smooth relaxation crossovers and diagnoses the transition using the order-parameter harmonics

${\bf F}$1

An important nuance is emphasized there: the ordinary QME can be misleading as an optimization principle because a DPT-induced metastable interval may also slow total relaxation. That is why the paper argues that the Pontus–Mpemba protocol is the more relevant notion when the full dynamical route, including metastability, determines performance (Nava et al., 11 Sep 2025).

4. Open-system and dissipative implementations

Open-system QPME has been formulated in several distinct architectures. In dissipative quasiperiodic chains, the system obeys a Lindblad master equation with local dephasing,

${\bf F}$2

and all initial states relax to the maximally mixed infinite-temperature steady state

${\bf F}$3

The protocol first couples the system to a finite-temperature bosonic bath that drives it toward

${\bf F}$4

then switches back to the original dephasing stage (Song et al., 17 Feb 2026). The paper reports a strictly shorter overall relaxation time than direct dephasing for both localized and extended eigenstates, and also in the presence of power-law long-range hopping. The stated mechanism is a redistribution of spectral weight that suppresses the overlap with the slowest mode, not any modification of the decay spectrum itself (Song et al., 17 Feb 2026).

A second route uses a temporary bond-dissipation quench in Markovian open systems. The added jump operator is

${\bf F}$5

By turning this channel on only during a finite time window, the protocol can realize either QME or anti-QME in systems with dephasing or boundary loss. The key advance claimed there is that the effect becomes protocol-driven rather than state-engineered, so it need not rely on specially chosen initial states (Liu et al., 6 Nov 2025).

A minimal and experimentally accessible open-system realization is proposed in cavity quantum electrodynamics using the dissipative Jaynes–Cummings model with photon loss. In the single-excitation sector the Hamiltonian is

${\bf F}$6

with Lindblad evolution

${\bf F}$7

The same initial state ${\bf F}$8 is used in both protocols. Under constant large ${\bf F}$9, the system decays slowly through a Purcell-like channel; under the two-step protocol, a low-loss stage of duration

$t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$0

approximately transfers the excitation to $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$1, after which a quench to large $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$2 produces rapid photon leakage to $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$3 (Longhi, 7 May 2026). The slow and fast decay scales in the strong-dissipation limit are

$t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$4

and the paper identifies an exceptional point at

$t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$5

The effect is therefore traced to the interplay between coherent atom–photon exchange and dissipation, which prepares a state strongly aligned with the fast mode (Longhi, 7 May 2026).

5. Quantum resource theories and quantum complexity

A particularly broad generalization of QPME is the extension from thermodynamic observables to quantum-complexity resources. In random brickwork circuits on one-dimensional chains of qudits, the subsystem state is

$t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$6

with $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$7 a small subsystem and $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$8 its complement. The global evolution is unitary and resource-preserving, but local resource monotones on $t_{\rm SI} + t_{\rm IF} < t_{\rm SF}.$9 can decay because for $$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$0, the subsystem behaves as if coupled to a large reservoir (Aditya et al., 26 Sep 2025).

The study considers four quantum resource theories: coherence, imaginarity, non-Gaussianity, and magic. The corresponding monotones are the relative entropy of coherence,

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$1

the relative entropy of imaginarity,

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$2

the relative entropy of Gaussianity / non-Gaussianity,

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$3

and the mana,

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$4

The paper’s main numerical conclusion is sharply differentiated. Under direct free evolution from tilted product states, coherence and imaginarity show a clear QME, while non-Gaussianity and magic do not. Under preheating, however, all four resources exhibit the QPME (Aditya et al., 26 Sep 2025). This is described as especially striking for non-Gaussianity and magic because the ordinary one-stage QME is absent in the original setup but the two-stage Pontus effect persists.

The physical intuition offered is that preheating can reshape the state so that, when free evolution begins, the local resource content couples differently to the slow modes of the dynamics. The paper explicitly notes a Markovian open-system example for coherence in which the more resourceful state has smaller overlap with the slowest mode $$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$5, thereby accelerating relaxation (Aditya et al., 26 Sep 2025). This suggests a unifying interpretation: resource monotones can play the role of nonequilibrium observables, and QPME can operate as a general mechanism of accelerated local resource depletion in theories relevant to quantum information and quantum computation.

6. Diagnostics, terminology, and current scope

The literature uses several diagnostics to identify QPME or closely related effects. In symmetry-based closed-system studies, the most common is entanglement asymmetry

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$6

which quantifies symmetry breaking by measuring the relative entropy between a subsystem density matrix and its symmetry-projected version (Xu et al., 11 Aug 2025). In open systems, trace distance to the steady state,

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$7

is widely used, including in dissipative quasiperiodic chains and cavity QED (Song et al., 17 Feb 2026). Some works additionally use Hilbert–Schmidt distance, relative entropy, forward and backward fidelities, or problem-specific order-parameter distances (Longhi, 7 May 2026).

An important practical issue is that detecting Mpemba-type behavior usually seems to require full state tomography. One recent proposal argues that this can be avoided by identifying “good observables” $$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$8 satisfying

$$\tilde{\rho}_t\equiv \begin{cases} \mathrm{Tr}_{B}\!\left[\tilde{U}_t |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_t^\dagger \right] & t\le T\ \mathrm{Tr}_{B}\!\left[U_{t-T} \tilde{U}_T |\Psi_{1/2}\rangle\langle \Psi_{1/2}| \tilde{U}_T^\dagger U_{t-T}^\dagger\right] & t> T \end{cases}$$9

so that the observable carries the slowest-decaying mode’s signature (Bagui et al., 2 Dec 2025). In the presence of a known unique steady state and known preparation protocol, measuring such observables can suffice to detect QME without reconstructing the full density matrix. A plausible implication is that similar criteria may become important for scalable experimental studies of QPME.

The terminology itself remains nonuniform. Several papers explicitly define QPME as a Pontus-type, two-step protocol (Yu et al., 2 Sep 2025, Song et al., 17 Feb 2026, Longhi, 7 May 2026). Others use the broader language of quantum Mpemba effect and do not introduce a separate “Pontus” concept (Li et al., 16 Apr 2026, Yu et al., 3 Jul 2025, Saliba et al., 15 Dec 2025). The general theory of Pontus-Mpemba effects provides the strict protocol-level definition and the three-class taxonomy for Markovian classical and quantum systems (Nava et al., 20 May 2025). This suggests that current usage spans at least two conventions: one in which “QPME” is the natural quantum version of the Pontus protocol, and another in which only “QME” is treated as standard nomenclature.

Within those terminological differences, the substantive message is consistent. Whether realized by a symmetry-breaking detour, a preheating stage, a temporary dissipative quench, a finite-temperature intermediate bath, or a low-loss cavity interval, QPME is characterized by path-dependent accelerated relaxation from the same initial state. The best-established mechanism is the engineered suppression of overlap with slow decay modes, while the best-established limitation is that the advantage depends on the entire dynamical route, including any metastable or preparatory stage, rather than on initial distance to the target alone (Nava et al., 11 Sep 2025).