{"id":"07c1bd17-a8a0-47a2-bb31-c985ebb0b068","arxiv_id":"2412.10145","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"In the 2D transverse-field Ising model, domain wall interfaces show long-lived prethermal plateaus in the smooth-interface regime, linked to an interface roughening transition and captured by a 1D solid-on-solid model.","lead":"Large-scale tensor network simulations of the 2D quantum Ising model show that a flat magnetic domain wall can remain stable for very long times, creating prethermal plateaus that are not explained by standard fragmentation mechanisms. The paper attributes this to a roughening transition of the interface, captured quantitatively by a 1D effective model.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The roughening–relaxation link is inferred, not demonstrated: the single-domain-wall projection is validated only for g/J≤1, while the claimed transition and fast-decay regime lie at g/J≈1.38>1.","rationale":"The reader's weakest assumption—that the single-domain-wall projection is only validated for g/J≤1 and the full dynamics are not directly simulated in the rough regime—is also the most load-bearing concern here. I agree with that assessment. The paper's strengths are real: the TTN convergence checks, the quantitative match between the effective and full models for g/J≤1, and the analytical treatment of the classical-limit crossover all support the smooth-regime half of the central claim. However, the rough-regime half, namely that interfaces decay quickly above the roughening transition, is inferred rather than observed, and the transition itself is located only within the effective model. The proposed test is a direct, feasible check: extending the existing full-model simulations to fields just below and above gR would reveal whether the single-domain-wall subspace and the dynamical crossover actually align in the full 2D TFIM. Since the missing piece is a specific addressable validation rather than a demonstrated contradiction, the conditional verdict remains appropriate; our read does not change it.","tokens_in":18302,"tokens_out":6141,"duration_ms":68292,"concrete_test":"Run full TTN simulations on an 8×8 (and, if feasible, 12×12) lattice for g/J = 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 from the flat domain-wall initial state, measuring Dx/Lx(t), the imbalance I(t), and an unambiguous interface width defined by the crossing of the local magnetization profile. If Dx/Lx deviates substantially from 1 before g/J≈1.38, or if the plateau lifetime decays smoothly across gR without a qualitative change near gR, then the roughening transition is not established as controlling the full dynamics. If instead interface width grows and the imbalance decays quickly only for g/J≳1.38, the central claim is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim asserts that the roughening transition controls interface relaxation in the full 2D TFIM. For this to hold, the single-domain-wall subspace behind the effective model (Eq. 3) must remain the relevant dynamical manifold at and above the claimed transition gR/J≈1.38. The paper's only direct validation of that subspace is Dx/Lx≈1 for g/J≤1 (Fig. 3a and Fig. S7a), strictly below gR. For g/J>1, the authors explicitly state that bulk and interface contributions cannot be disentangled (Fig. S6b) and therefore do not show full-model data for the kink operator. Consequently, the transition itself is located only in the effective model (End Matter BKT fit with Nmax≤14, which is not fully converged near gR), and the full-model crossover from prethermal plateaus to fast decay is inferred from the phase diagram (Fig. 2d) rather than directly observed with an interface order parameter. Bubble and overhang degrees of freedom—precisely those excluded by the projection—are expected to become important near roughening, so they could shift or preempt the transition in the full model. This gap is load-bearing because 'rough interfaces decay quickly' is one half of the central dichotomy; without a direct full-model check, the roughening transition is not demonstrably the mechanism controlling relaxation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the real-time dynamics of flat domain-wall interfaces in the two-dimensional transverse-field Ising model using large-scale tree tensor network simulations. For transverse fields up to g/J ≈ 1 the authors observe long-lived prethermal plateaus in the magnetization imbalance and in a modified kink operator. To explain these plateaus, they introduce an effective one-dimensional solid-on-solid model (Eq. 3) obtained by projecting onto single-domain-wall states with one horizontal interface segment per column. Using VUMPS in the thermodynamic limit, they locate a BKT roughening transition in the effective model at gR/J ≈ 1.38 and give an analytic treatment of the classical limit where the smooth-interface crossover temperature vanishes logarithmically with system size. The effective model is shown to quantitatively match the full-model dynamics for the kink operator up to g/J = 1. The central claim is that the roughening transition controls the relaxation of interfaces: smooth interfaces exhibit prethermal stability, while rough interfaces decay quickly, constituting a new prethermalization mechanism distinct from fragmentation or proximity to integrability.","tokens_in":18562,"tokens_out":5741,"duration_ms":61421,"significance":"If established, the connection between the interface roughening transition and the relaxation dynamics of domain walls would be a genuinely new mechanism for prethermalization in a two-dimensional quantum many-body system, with direct implications for Rydberg-atom quantum simulators. The paper has several clear strengths: the full-model TTN simulations are checked against multiple bond dimensions up to χ = 362; the comparison between the full model and the effective model is quantitative and favorable in the accessible regime; the classical-limit analysis is solved essentially exactly via a transfer-matrix diagonalization, leading to the closed-form prediction (5); and the authors release data and code. The main weakness is that the load-bearing connection between roughening and fast decay is inferred from the effective model rather than directly demonstrated in the full 2D model: full-model interface observables are only validated for g/J ≤ 1, while the claimed transition lies at gR/J ≈ 1.38. The paper is therefore an interesting and well-executed study whose headline conclusion currently outruns the direct numerical evidence.","major_comments":[{"comment":"The central dichotomy — smooth interfaces show prethermal plateaus, rough interfaces decay quickly — is not directly established in the full 2D model. The modified kink operator KM is presented only for g/J ≤ 1 in Fig. 3b,c, and the SM explicitly states that separating bulk and interface contributions becomes infeasible for g/J > 1, with the unphysical growth in Fig. S6b flagged as an artifact of the method. The roughening transition is located in the effective model at gR/J ≈ 1.38 (End Matter), which lies outside the validated range of the full-model comparison. Consequently, the fast-decay regime is inferred from the effective-model phase diagram (Fig. 2d) rather than observed with an interface order parameter in the full model. This is load-bearing because 'above the roughening transition the domain wall decays quickly' is half of the paper's central claim. Please either provide a full-model observable that can be followed through gR (for example, a different bulk-subtraction scheme, or the late-time behavior of the bare kink operator with a quantitative error estimate) or explicitly present the roughening–relaxation link as a prediction of the effective model rather than a demonstrated property of the full 2D dynamics.","section":"Dynamical signature of roughening (Fig. 3, SM Fig. S6b)"},{"comment":"The single-domain-wall projection excludes bubbles and overhangs, and its direct validation rests on the horizontal domain-wall length Dx/Lx staying close to one for g/J ≤ 1 up to the simulated times. Since the claimed transition lies at g/J ≈ 1.38, the subspace is not verified in precisely the regime where the transition occurs in the effective model. If bubble-pair or overhang degrees of freedom proliferate before gR in the full model, the effective model could either misplace the transition or miss its preemption. A quantitative estimate of the weight outside the single-domain-wall subspace as a function of g and t — for example, the fraction of columns with more than one horizontal interface segment, or the density of vertical domain-wall pairs — would substantiate the projection at and above the transition. As it stands, the projection is a controlled approximation only in the smooth regime, which is also the regime where the full-model comparison is available.","section":"Effective model and its validation (Eq. 3, Fig. 3a, SM Fig. S7a)"},{"comment":"The location of the roughening transition relies on VUMPS data with Nmax ≤ 14, and the manuscript itself states that for Nmax = 14 the points around the transition are not yet fully converged and show numerical artifacts for g > gR. The BKT correlation-length fit to Eq. (6) is performed on this Nmax = 14 data up to g/J = 1.385, i.e., close to the fitted gR/J ≈ 1.38. It would materially strengthen the determination of gR to include an extrapolation in 1/Nmax or data at Nmax = 16 for a few points near gR, so that the fitted critical point is not controlled by unconverged data. This matters because Fig. 2d uses gR as the boundary separating smooth and rough interface regimes in the full model.","section":"End Matter (BKT fit, Fig. 4b,c)"}],"minor_comments":[{"comment":"The name 'Berezinski-Kosterlitz-Thouless' should be corrected to 'Berezinskii-Kosterlitz-Thouless' for consistency with the later usage 'Berezinskii–Kosterlitz–Thouless'.","section":"Introduction"},{"comment":"In the sentence 'Next, we turn towards the question wether its signatures survive even at non-zero temperatures', 'wether' should be 'whether'.","section":"Interface dynamics"},{"comment":"The phrase 'timescales which scale exponentially with J/g 1' is confusing because the footnote marker '1' follows the expression as if it were part of the formula; consider moving the footnote marker or rewriting the sentence.","section":"Interface dynamics (footnote 1)"},{"comment":"The SM derivation writes the vertical contribution as proportional to ∑|N_i − N_{i−1}|, while Eq. (3) in the main text uses |N_i − N_{i+1}|; aligning the index convention or explicitly stating the open-boundary convention would avoid confusion.","section":"Effective model"},{"comment":"In the sentence 'We want to stress that the observed rising of the kink operator an artifact of the method and not a physical effect', the verb 'is' is missing before 'an artifact'.","section":"SM, Kink operator"},{"comment":"The sentence 'More generally, the impact of curvature on the phenomenology of the roughening dynamics remains to be explored' is a suitable outlook, but the earlier statement that 'the results presented here rely on a Z2 symmetry-broken phase' would benefit from a brief explanation of how the Z2 structure enters the effective model, which is not explicitly Z2-symmetric.","section":"Discussion"}],"recommendation":"major_revision","confidential_remarks":"This is a solid numerical study with a clear and well-executed comparison between full-model and effective-model dynamics, and the data/code release is commendable. The concern is not with the existence of the prethermal plateaus, which are directly observed, but with the headline causal claim that the roughening transition controls the relaxation dynamics. The paper's own text concedes that the full-model interface observable cannot be computed for g/J > 1, which is exactly the regime where the claimed transition sits. In my view this is fixable in revision: the authors could either add a full-model diagnostic that remains meaningful through gR or consistently reframe the roughening–relaxation link as a predictive consequence of the effective model. I would not reject the paper, but the current abstract and discussion overstate the directness of the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe prethermal plateaus are real; the roughening link is a reasonable but unproven inference. The headline result—long-lived prethermal plateaus for flat domain walls in the 2D TFIM at g/J up to about 1—is solidly established by TTN simulations with bond-dimension checks, and the proposed single-domain-wall effective model reproduces the full-model dynamics quantitatively in that regime. That is a genuine step beyond the perturbative PXP analysis of Balducci et al. The classical-limit transfer-matrix solution, with TR vanishing as 1/log Lx, is neat and exactly checked.\n\nWhere I part ways with the authors is the slope from 'the effective model has a BKT roughening transition at gR/J ≈ 1.38' to 'the roughening transition controls interface relaxation.' The transition is located only in the effective model via VUMPS with Nmax ≤ 14, and the authors admit the data are not fully converged near the transition. Full-model dynamics are matched only up to g/J ≤ 1, strictly below gR. For g/J > 1 they explicitly cannot separate bulk from interface contributions, so the 'rough interfaces decay quickly' half of the dichotomy is inferred from the phase diagram, not observed. The single-domain-wall projection is validated only by Dx/Lx ≈ 1 for g ≤ 1; bubbles and overhangs are precisely the degrees of freedom expected to proliferate near roughening, so the projection could break first. That gap is load-bearing: without a direct full-model check in the rough regime, the roughening transition is a plausible explanation for the dynamical crossover, not a demonstrated one.\n\nNone of this is fatal. The prethermal plateaus stand on their own, and the effective model is a useful tool even if the transition location shifts or the mechanism is modified by bubble physics. The paper is honest about its limitations—it explicitly flags the bulk/interface separation failure and the Nmax artifacts, which I respect.\n\nMy recommendation: send it to review. The referee should push on the inference: either soften the claim to 'consistent with a roughening-controlled crossover' or find a way to probe full-model dynamics in the rough regime. Also worth asking: α = 1 is tuned for agreement; a scan over α would show how robust the matching is.\n\nWho reads it: people doing quantum simulation of Rydberg arrays, interface dynamics, and prethermalization. It deserves referee time.","headline":"Solid numerics establish prethermal plateaus in the 2D TFIM, but the claim that the roughening transition controls relaxation is inferred from an effective model, not directly observed in the full model.","tokens_in":19180,"tokens_out":3362,"would_cite":true,"duration_ms":30647,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B20","82B26","82B80"],"pacs":["05.30.-d","75.10.Jm","68.35.Ct"],"model":"deepseek-v4-flash","headline":"The roughening transition controls how long flat quantum interfaces stay stable in the two-dimensional quantum Ising model, with smooth interfaces showing long-lived prethermal plateaus and rough ones decaying quickly.","keywords":["quantum Ising model","interface roughening transition","domain wall dynamics","prethermalization","tree tensor networks","solid-on-solid model","BKT transition","Rydberg quantum simulators"],"falsifier":"Simulate or measure the full two-dimensional dynamics at $g/J = 0.75$ to $1.0$ beyond $tJ = 100$, for example in a Rydberg atom array or with a larger tensor-network simulation, and record both the horizontal domain-wall length $D_x/L_x$ and the modified kink operator. If $D_x/L_x$ drops appreciably below 1, or if the kink plateau decays within a time that does not grow rapidly as the field approaches the inferred roughening transition, then the single-domain-wall projection and the roughening-controlled prethermalization claim are falsified.","tokens_in":18061,"feed_emoji":"⚛️","tokens_out":7877,"duration_ms":86065,"temperature":0.7,"pith_summary":"This paper tries to establish that the equilibrium roughening transition of interfaces has a sharp dynamical signature in the two-dimensional transverse-field Ising model. Starting from a straight domain wall, smooth interfaces below the critical transverse field exhibit long-lived prethermal plateaus, whereas above the roughening transition the domain wall decays quickly. The authors derive a one-dimensional solid-on-solid effective model for a single domain-wall profile per column and show that its dynamics match the full two-dimensional tensor-network simulation quantitatively up to $g/J \\approx 1$. If correct, this identifies a distinct, non-perturbative mechanism for prethermalization, separate from proximity to integrability, Hilbert-space fragmentation, or quantum scars, and one that is directly testable in Rydberg atom arrays.","feed_headline":"Roughening transition sets quantum domain-wall lifetimes","feed_subtitle":"In the 2D quantum Ising model, flat interfaces below the critical field stay stable in long-lived prethermal plateaus.","key_machinery":"The argument is carried by the effective Hamiltonian $H_{\\mathrm{eff}} = 2J \\sum_i |N_i - N_{i+1}| - g \\sum_i (E_i + E_i^\\dagger)$, where the height operators $N_i$ measure the perpendicular displacement of the interface in column $i$ and $E_i^\\dagger, E_i$ flip spins next to the wall, obeying $[E_i, N_j] = E_i \\delta_{i,j}$. This projection onto single-valued interface configurations, with no bubbles or overhangs, maps interface dynamics onto a one-dimensional quantum-rotor/solid-on-solid problem. The order parameter is the kink operator $K_\\alpha(l) = \\cos\\big(\\alpha(N_1 - N_l)\\big)$, which equals 1 for a flat smooth interface and tends to 0 for a rough one; it is evaluated both in ground states, via uniform matrix-product-state methods, and in classical thermal states, via an exact transfer-matrix solution. The same operator, with bulk contributions divided out, provides the quantitative comparison between the full 2D tree-tensor-network dynamics and the effective model.","core_discovery":"The central discovery is that the relaxation of a flat interface in the 2D quantum Ising model is governed by the same physics as the roughening transition. Projecting the dynamics onto states with exactly one horizontal interface segment per column yields an effective one-dimensional solid-on-solid model whose ground state has a smooth-to-rough Berezinskii-Kosterlitz-Thouless transition at $g_R/J \\approx 1.38$; within the smooth regime the effective model thermalizes while the initially flat profile remains stable, which in the full model appears as prethermal plateaus in the imbalance and the modified kink operator. Large-scale tree tensor network simulations of the full 2D dynamics agree quantitatively with matrix-product-state simulations of the effective model for $g/J \\leq 1$, including late-time plateau values on $8 \\times 8$ and $16 \\times 16$ lattices. In the classical limit of the effective model the smooth-interface phase is only a finite-size effect, with the crossover temperature vanishing as $1/\\log L_x$, but the smooth regime survives to very large system sizes, which the authors argue is relevant for current experiments.","pith_inferences":["Editorial inference: if the mechanism is robust, the smooth-interface regime of a two-dimensional Ising-type simulator acts as a tunable metastable memory whose lifetime is set by the distance to the roughening transition, suggesting that the roughening transition could be used to control decoherence in quantum simulation platforms.","Editorial inference: the same single-interface projection should apply to non-straight profiles such as zigzag walls, curved interfaces, and false-vacuum bubbles, so roughening may set the timescale for their straightening or decay; the paper lists these as future directions but does not make quantitative predictions for them.","Editorial inference: the paper validates the effective model only for $g/J \\leq 1$ and up to the simulated times; a direct full two-dimensional simulation of the rough regime, $g/J > 1$, is not performed because separating bulk from interface contributions becomes infeasible, so the rapid decay above the transition remains an extrapolation from the effective model rather than a directly simulated ","Editorial inference: the logarithmic dependence of $T_R$ on system size means that distinguishing a genuine roughening transition from a finite-size crossover in the full model will require careful scaling analysis; measuring the kink correlator as a function of system size and transverse field in a Rydberg array would be a direct test of the BKT character."],"forward_implications":["For $g/J \\lesssim 1$ the initially flat domain wall remains essentially frozen up to at least $tJ \\approx 100$ on an $8 \\times 8$ lattice and $tJ \\approx 30$ on a $16 \\times 16$ lattice; since level-spacing statistics are GOE-like, the eventual fate is thermalization, so the plateaus are prethermal rather than equilibrium order.","The mechanism does not rely on emergent conservation laws or Hilbert-space fragmentation, because the horizontal domain-wall length $D_x$ changes from its initial value during the dynamics; it therefore adds a distinct route to prethermalization.","The effective one-dimensional model is quantitatively predictive for the full two-dimensional interface dynamics in the tested regime, and the imbalance and kink operator are directly measurable through snapshots in Rydberg atom quantum simulators.","The inferred roughening transition at $g_R/J \\approx 1.38$ marks the upper bound where the single-domain-wall description works; above it the interface is expected to decay quickly, although the full two-dimensional model is not directly simulated there.","The smooth-interface regime persists in finite systems up to very large sizes because the classical crossover temperature scales as $T_R \\sim 1/\\log L_x$, so current experimental system sizes should display the prethermal plateaus even though the thermodynamic classical limit roughens at any positive temperature."],"supporting_citations":[{"why":"Establishes that quantum fluctuations can drive a T=0 roughening transition in two-dimensional systems, the phenomenon whose dynamical signature the paper seeks.","marker":"[9]"},{"why":"Identifies the roughening transition as BKT-like in solid-on-solid models and supplies the exactly solvable reference model that the effective Hamiltonian generalizes.","marker":"[4, 5]"},{"why":"Provides the tree tensor network and time-dependent variational principle methods used for the full two-dimensional real-time evolution.","marker":"[12–14]"},{"why":"Demonstrates interface localization and melting in the perturbative PXP limit of the 2D quantum Ising model, defining the fragmentation-based mechanism that the new prethermalization mechanism is contrasted with.","marker":"[33, 34]"},{"why":"Shows programmable Rydberg simulators with transverse fields and domain-wall coarsening, the experimental platform the paper says can test its predictions.","marker":"[31]"},{"why":"Supplies the ferromagnetic-phase critical line and transition temperature used to place the roughening transition inside the phase diagram.","marker":"[32]"},{"why":"Provides the variational uniform matrix product state algorithm used to extract the ground-state kink operator and the BKT correlation-length divergence in the effective model.","marker":"[37–39]"},{"why":"Supplies the quantum Monte Carlo loop algorithm used to assign effective temperatures to the initial domain-wall states in the inferred phase diagram.","marker":"[40, 41]"}],"fun_headline_variants":["Quantum roughening controls domain-wall lifetimes","Smooth interfaces survive roughening transition","Roughening transition explains prethermal domain walls","Domain-wall decay rate set by quantum roughening","Roughening dictates quantum interface stability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the interface stays single-valued, with exactly one horizontal domain-wall segment per column and no bubbles or overhangs; this is validated only indirectly by $D_x/L_x$ remaining near 1 for $g/J \\leq 1$ up to the simulated times, and if multi-segment configurations become important at longer times or larger fields, the effective model and the inferred link to the roughening transition would break down.","fun_headline_variants_meta":{"raw":{"variants":["Quantum roughening controls domain-wall lifetimes","Smooth interfaces survive roughening transition","Roughening transition explains prethermal domain walls","Domain-wall decay rate set by quantum roughening","Roughening dictates quantum interface stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000732,"raw_usage":{"total_tokens":3236,"prompt_tokens":864,"completion_tokens":2372,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":480,"completion_tokens_details":{"reasoning_tokens":2309}},"tokens_in":480,"tokens_out":2372,"duration_ms":18525,"temperature":1.0,"reasoning_tokens":2309,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T16:18:01.437495+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or measure the full two-dimensional dynamics at $g/J = 0.75$ to $1.0$ beyond $tJ = 100$, for example in a Rydberg atom array or with a larger tensor-network simulation, and record both the horizontal domain-wall length $D_x/L_x$ and the modified kink operator. If $D_x/L_x$ drops appreciably below 1, or if the kink plateau decays within a time that does not grow rapidly as the field approaches the inferred roughening transition, then the single-domain-wall projection and the roughening-controlled prethermalization claim are falsified.","supporting_citations":[{"cited_title":"Fradkin, Roughening transition in quantum interfaces, Phys","cited_arxiv_id":null,"evidence_quote":"Establishes that quantum fluctuations can drive a T=0 roughening transition in two-dimensional systems, the phenomenon whose dynamical signature the paper seeks."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the ferromagnetic-phase critical line and transition temperature used to place the roughening transition inside the phase diagram."}],"review_version":1}