REVIEW
Observing the magic Mpemba effect in localized dynamics on a digital quantum computer
Quantum Physics · Submitted 6 Oct 2026 · 287 references
Pith's verdict: A clean, parameter-free case for a dephasing-driven magic Mpemba effect in localized models; the numerics and hardware support the crossing, the asymptotic ordering rests on an acknowledged but unproven self-averaging step, and the abstract overstates what the hardware observed.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
Dephasing alone, with no transport or thermalization, can make a state that starts with less quantum magic overtake a more magical one and settle at a higher value sooner — an effect this paper derives, simulates, and observes on a…
The reading
What carries the argument
Three objects carry the argument. The Hamiltonian is the phenomenological $\ell$-bit model of localization, $H = \sum_i h_i\tau^z_i + \sum_{i<j} J_{ij}\tau^z_i\tau^z_j + \sum_{i<j<k} K_{ijk}\tau^z_i\tau^z_j\tau^z_k + \cdots$, whose conserved $\tau^z_i$ freeze the configuration weights while interactions dephase them; the measure is the second stabilizer R\'enyi entropy $M_2 = -\log_2 I_4$ with $I_4 = 2^{-N}\sum_{P\in\mathcal{P}_N}|\langle\psi|P|\psi\rangle|^4$; and the carrying identity is the random-phase average of $I_4$ over independent uniform configuration phases, $I_4^{\mathrm{sat}}(\theta) = 4[(1+\cos^4\theta)/2]^N - 6[(1+\cos^2\theta)^2/4]^N + 3[(1+6\cos^2\theta+\cos^4\theta)/8]^N$, whose thermodynamic limit $1-\log_2(1+\cos 4\theta)$ slopes up where the initial value slopes down. On hardware the observation rides on two circuit idioms: a fixed-depth diagonal $\ell$-bit circuit that encodes simulated time in the gate angles so that long dephasing is reached without Trotterization, and a kicked-XXZ chain whose states are prepared by individually compiled circuits, with the noninteracting limit of the same $\ell$-bit circuit supplying the control that cancels the reversal.
Extended reading notes
Core claim
The central discovery, on the paper's own terms, is that dephasing alone suffices to invert the magic ordering. In the interacting $\ell$-bit description of many-body localization, every $\tau^z_i$ is conserved, so configuration probabilities are frozen while the interaction terms make each precession depend on the full configuration, dephasing the relative phases. Averaging the fourth Pauli intensity $I_4$ over independent uniform configuration phases yields saturated magic $M_2^{\mathrm{sat}}(\theta)/N = 1-\log_2(1+\cos 4\theta)$, which increases with the tilt $\theta$, whereas the initial magic $M_2(0,\theta)/N = -\log_2[1-\tfrac{1}{4}\sin^2(2\theta)]$ decreases with $\theta$ on $\pi/4<\theta<\pi/2$; a less-magical state therefore acquires more magic at saturation, with the endpoint order identical for tilted ferromagnetic, N\'eel, and domain-wall states. Because the full Mpemba criterion also demands that the farther state arrive sooner, the paper verifies numerically that $t_{\mathrm{ss}}(0.50\pi)<t_{\mathrm{ss}}(0.35\pi)$ across all three initial-state families and across four localized settings (interacting $\ell$-bit, random-disorder, quasiperiodic, and Stark spin chains). On hardware, six-qubit Bell measurements and ten-qubit kicked-XXZ runs reproduce the reversal, the Anderson limit of the same $\ell$-bit circuit removes it, and the microscopic chain shows pattern-robust reversal in the localized regime but a pattern-selective one under the same U(1) symmetry in the ergodic regime.
Load-bearing premise
The load-bearing premise is that the random-phase ensemble — independent uniform configuration phases over frozen conserved weights — describes the coarse-grained late-time magic of the actual Hamiltonian evolution; the paper itself states that this 'identification with a typical long-time-window magic density assumes self-averaging of the random-phase ensemble.' Because a finite closed chain is quasiperiodic and possesses no pointwise infinite-time limit, the reversal of the asymptotic ordering, and the saturation-time inversion that goes with it, rest on that unproved self-averaging step rather than on the exact ensemble derivation alone.
Editorial extensions
If this is right
- The magic Mpemba effect requires neither transport nor conventional thermalization: any localized system whose interactions dephase conserved configuration amplitudes can reverse the magic ordering, extending anomalous resource generation beyond the chaotic, charge-conserving circuits where it was first identified.
- The reversal is pattern-robust in the localized regime — tilted ferromagnetic, N\'eel, and domain-wall states share the endpoint ordering — while the same U(1)-symmetric ergodic control reverses only for two of the three families, so the conserved charge alone does not determine the effect.
- The noninteracting Anderson limit generates magic through local precession yet reverses no ordering, so interactions, specifically many-body configuration dephasing, are essential rather than localization on its own.
- The fixed-depth $\ell$-bit circuit, which folds simulated time into gate angles, reaches long dephasing windows without Trotterization, a practical route to slowly relaxing localized dynamics on noisy hardware.
- The hardware data resolve the magic-ordering crossing, while the saturation-time inversion (which state reaches its own higher value sooner) is established by exact numerics; determining it directly on the processor remains the stated next step.
Reading between the lines
- I infer the mechanism should transfer beyond localization: any many-body setting with conserved densities plus dephasing interactions — quantum scars, Hilbert-space-fragmenting chains, dipole-constrained models — should show the same frozen-weight reversal, settings the paper lists as natural without testing them.
- The random-phase saturation is an annealed average taken over $I_4$ before the logarithm; a quenched average, or a single shot's window average of $M_2$ on a finite closed chain, would give different finite-size numbers, so a sharp form of the claim is that the dephased ensemble has the reversed ordering, with the per-realization robustness a separate testable question.
- Because magic in the $\ell$-bit model is basis-dependent (dressed $\ell$-bit operators differ from physical spins), the quantitative reversal curve $M_2^{\mathrm{sat}}(\theta)/N$ should shift with the local Pauli basis, so measuring the same dynamics with physical-spin versus $\ell$-bit moments would expose that dependence.
Original abstract
The Mpemba effect challenges the intuition that states closer to equilibrium must relax faster. We ask whether an analogous magic-ordering reversal can occur in the generation of quantum magic (nonstabilizerness), a key resource for universal quantum computation: can a state with less initial magic overtake one with more and reach its asymptotic value sooner? Here, we uncover interaction-induced dephasing as a distinct mechanism for this magic Mpemba effect, requiring neither transport nor conventional thermalization. For tilted product states in an interacting $\ell$-bit model, a random-phase analysis predicts a reversal of the magic ordering: states with less initial magic attain higher saturated magic, independent of their spatial pattern. Exact simulations of the $\ell$-bit model and localized spin chains further show that these states also saturate earlier. We observe both features on IBM superconducting quantum processors. An effective $\ell$-bit implementation accesses long dephasing times at fixed circuit depth by encoding time in gate angles, while a microscopic spin model exhibits the reversal universally across initial patterns in the localized regime but only selectively in the ergodic regime. These results establish interaction-induced dephasing as a route to anomalous quantum-resource relaxation beyond transport-driven thermalizing dynamics.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
The Mpemba effect challenges the intuition that states closer to equilibrium must relax faster. We ask whether an analogous magic-ordering reversal can occur in the generation of quantum magic (nonstabilizerness), a key resource for universal quantum computation: can a state with less initial magic overtake one with more and reach its asymptotic value sooner? Here, we uncover interaction-induced dephasing as a distinct mechanism for this magic Mpemba effect, requiring neither transport nor conventional thermalization. For tilted product states in an interacting $\ell$-bit model, a random-phase analysis predicts a reversal of the magic ordering: states with less initial magic attain higher saturated magic, independent of their spatial pattern. Exact simulations of the $\ell$-bit model and localized spin chains further show that these states also saturate earlier. We observe both features on IBM superconducting quantum processors. An effective $\ell$-bit implementation accesses long dephasing times at fixed circuit depth by encoding time in gate angles, while a microscopic spin model exhibits the reversal universally across initial patterns in the localized regime but only selectively in the ergodic regime. These results establish interaction-induced dephasing as a route to anomalous quantum-resource relaxation beyond transport-driven thermalizing dynamics.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | The Mpemba effect challenges the intuition that states closer to equilibrium must relax faster. | background 1.00 | 0.05 | 0.03 | asserted 1.00 |
| s1 | We ask whether an analogous magic-ordering reversal can occur in the generation of quantum magic (nonstabilizerness), a key resource for universal quantum computation: can a state with less initial magic overtake one with more and reach its asymptotic value sooner? | gap 0.58 | 0.19 | 0.05 | hedged possibility 0.59 |
| s2 | Here, we uncover interaction-induced dephasing as a distinct mechanism for this magic Mpemba effect, requiring neither transport nor conventional thermalization. | result 0.58 | 0.93 | 0.43 | asserted 1.00 |
| s3 | For tilted product states in an interacting $\ell$-bit model, a random-phase analysis predicts a reversal of the magic ordering: states with less initial magic attain higher saturated magic, independent of their spatial pattern. | result 0.62 | 0.84 | 0.15 | asserted 0.50 |
| s4 | Exact simulations of the $\ell$-bit model and localized spin chains further show that these states also saturate earlier. | result 0.98 | 0.95 | 0.14 | asserted 1.00 |
| s5 | We observe both features on IBM superconducting quantum processors. | result 0.91 | 0.96 | 0.09 | asserted 1.00 |
| s6 | An effective $\ell$-bit implementation accesses long dephasing times at fixed circuit depth by encoding time in gate angles, while a microscopic spin model exhibits the reversal universally across initial patterns in the localized regime but only selectively in the ergodic regime. | result 0.64 | 0.91 | 0.12 | asserted 1.00 |
| s7 | These results establish interaction-induced dephasing as a route to anomalous quantum-resource relaxation beyond transport-driven thermalizing dynamics. | interpretation 1.00 | 0.87 | 0.87 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
Figures
Showing 2 of the paper's 5 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:2610.08769 by the authors.
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.