{"id":"809a9c9c-4cce-4269-aaf9-ef6e06b003f3","arxiv_id":"2509.06679","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"PT-symmetric two-level systems have two topological phases, and switching between them in time makes wave intensity peak at the switch.","lead":"This paper shows that in any system with parity-time symmetry and exactly two coupled modes, a sudden switch between two topological phases in time makes wave intensity pile up at the switch moment. It provides a general theoretical framework for temporal topological localization, without requiring periodic driving or a spatial model.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'series of quenches' reduction in SI §1 does not justify the claim for continuous time-dependent H(t); the proof covers only an instantaneous quench between static imaginary-gapped Hamiltonians, and continuous passage through the exceptional point at the phase boundary is uncontrolled.","rationale":"The reader's weakest assumption—that arbitrary time-dependent H(t) can be treated as a series of static quenches—is exactly the load-bearing point. The proof in SI §1 establishes a robust result for a single instantaneous quench between two imaginary-gapped static Hamiltonians, with the post-quench decrease holding only for a finite time t_r. The leap to continuous time dependence is not justified, especially because the interface necessarily traverses the exceptional point where the gap closes and the two-level system becomes defective. This does not invalidate the quench result or the numerical demonstrations, but it undercuts the abstract's universality claim for 'every system' with arbitrary time variation. The four-level counterexample in SI §4 is a honest and useful boundary on the generalization to higher-level AI-class models, but it does not patch the continuous-time gap in the two-level argument. The proposed numerical sweep with variable ramp time directly tests whether the series-of-quenches reduction survives finite-rate switching; if it fails, the paper's theorem should be restricted to piecewise-constant Hamiltonians or supplemented with a rigorous adiabatic/continuity argument. Since this is the same concern the reader flagged, my recommendation is unchanged: CONDITIONAL, pending either a proof for continuous H(t) or a revised claim restricted to quenches.","tokens_in":20138,"tokens_out":8290,"duration_ms":96348,"concrete_test":"Numerically integrate i dψ/dt = H(t)ψ for H(t) = c σ_x + i γ(t) σ_z with c=1, γ(t) = γ0 + (γ1−γ0) * s((t−t0)/τ), where s is a smooth step (e.g., tanh), γ0=+2, γ1=−2, t0=0, and initial state is the instantaneous E+ eigenvector at t=−5. Compute I(t)=|ψ(t)|² for ramp times τ ∈ {0.01, 0.1, 0.5, 2.0}. If for τ ≥ 0.1 the intensity peak is suppressed, delayed, or dI/dt does not change sign at t0, the series-of-quenches reduction fails. Also run a same-phase fast variation (γ(t) oscillating between 2 and 3 before t0) to test whether monotonic increase before the interface survives arbitrary time dependence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that localization appears for arbitrary time-dependent PT-symmetric two-level Hamiltonians—rests on the SI §1 assertion: 'We may always consider such an interface as a series of quenches between static Hamiltonians.' This is not established for continuous H(t). The quench result requires both pre- and post-quench Hamiltonians to be imaginary-gapped (|γ|>c). But a continuous temporal interface between the two topological phases must pass through the exceptional point |γ|=c, where the Hamiltonian is defective and the two eigenvectors coalesce; the quench decomposition is singular there. Moreover, after the interface the monotonic decrease is only guaranteed for δt < t_r = (1/(2|E'|)) ln(|b_-|/|b_+|), a bound that depends on the overlap ratio at the quench instant; for a finite-rate ramp this ratio is not controlled by the static-quench analysis. The proof also assumes that under same-phase evolution the state remains in the same hemisphere as the instantaneous E+ eigenvector; for rapidly varying H(t) within a phase this is not proven. Thus the universal statement 'appears in every system ... regardless of spatial dimensionality' is only rigorously supported for piecewise-constant switches with a suitable initial state and a sufficiently short post-interface observation window, not for arbitrary time dependence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a framework for topological localisation in time in PT-symmetric two-level (or two-band) Schrödinger systems. It identifies two imaginary-gap topological phases through the hemispherical orientation of the E+ eigenvector, and shows that for an instantaneous quench between Hamiltonians in opposite phases, an initial state that dominantly excites the E+ eigenvector undergoes intensity growth before the interface and, for a short post-quench window, intensity decay after it—producing a temporal intensity peak. The authors extend this to translationally invariant lattices by treating each momentum sector as an independent PT-symmetric two-level model, and illustrate the effect numerically on a two-dimensional honeycomb lattice with balanced gain and loss. The Supplemental Material also contains an explicit four-level counterexample showing that the effect does not generalize to all imaginary-line-gapped models in the non-Hermitian AI class.","tokens_in":20486,"tokens_out":4734,"duration_ms":63906,"significance":"If the claims are properly scoped, this is a valuable and elegant contribution. The core quench calculation is analytic, parameter-free, and honest: it explicitly identifies the initial-state condition and the finite post-interface decay window, and it includes a counterexample to the natural higher-level generalisation. The use of the existing Kawabata et al. classification as an external benchmark, rather than as an input that assumes the result, is appropriate. The extension to momentum-space domains in higher-dimensional lattices is suggestive and could be of broad experimental relevance. However, the paper's headline claim of universality ('appears in every system that has parity-time symmetry and two coupled modes or bands') is stronger than what is proved. The proof requires a specific class of initial states and, more importantly, is only rigorous for piecewise-constant switches; the 'series of quenches' reduction for arbitrary time dependence is asserted, not proved. These issues are load-bearing for the central claim and warrant major revision.","major_comments":[{"comment":"The assertion 'We may always consider such an interface as a series of quenches between static Hamiltonians' is the only bridge from the exactly solved instantaneous-quench problem to the stated arbitrary time dependence. It is not proved, and the quench analysis explicitly requires both pre- and post-quench Hamiltonians to be imaginary-gapped. A continuous temporal interface between the two phases must pass through |γ|=c, where the Hamiltonian is defective and the eigenvectors coalesce; the quench decomposition is singular there. Moreover, the post-interface monotone decrease is established only for δt < t_r = (1/(2|E'|)) ln(|b_-|/|b_+|), a bound that depends on the overlap ratio at the quench instant. For a finite-rate ramp this ratio is not controlled by the static-quench argument. Please restrict the theorem to piecewise-constant switches, or supply a rigorous argument (e.g., a Trott","section":"Supplementary Information §1"},{"comment":"The abstract states that the phenomenon 'appears in every system that has parity-time symmetry and two coupled modes or bands, regardless of its spatial dimensionality.' This overstates the proved result. The derivation in SI §1 requires the initial state to satisfy |a_+|²>|a_-|², i.e., dominant E+ excitation, and the intensity decrease after the interface is only guaranteed for a finite time δt<t_r. The main text itself later acknowledges the initial-state requirement ('the eigenvector associated with E+(k) must be dominantly excited'), but the abstract and the conclusion omit it. As written, a reader could infer that a generic excitation will localise. This is a load-bearing gap between the result and the advertised universality.","section":"Abstract and Conclusion"},{"comment":"The spatial extension inherits both caveats of the zero-dimensional proof. For each momentum k in the excitation support S_k, the proof requires that the E+ eigenvector is dominantly excited and that dI_k/dt<0 is guaranteed only for t∈(t0,t0+t_r). With a continuous finite-rate switch, the state at t0+ for each k is not controlled by the static-quench analysis, and whether the momentum sectors remain in the appropriate hemisphere during same-phase evolution is not proved for rapidly varying H_k(t). The claim that the spatial profile can be chosen 'arbitrarily' is also contingent on these per-sector conditions being met. The numerical demonstrations in Fig. 3 use a switch of γ from 1 to -1; for such a finite-rate switch the same unsupported 'series of quenches' reduction is used.","section":"Main text 'Higher dimensions' and SI §2"}],"minor_comments":[{"comment":"In the expression for ψ(t) the second term is written with e^{-iE_+(t+δt)}; it should be e^{-iE_-(t+δt)}. This is a typographical error, since Eq. (6) and the surrounding argument assume E_±=±i|E|.","section":"SI §1, Eq. (5)"},{"comment":"The caption says the time dependence 'may be arbitrary.' Given the major comment above, please clarify whether the numerical simulation uses a piecewise-constant switch or a continuous ramp, and state which claim the figure is intended to support.","section":"Fig. 2 caption"},{"comment":"The phrase 'two levels or bands' is used interchangeably; in the zero-dimensional case it is strictly two levels. This is not an error, but the abstract's 'two coupled modes or bands' should be aligned with the theorem statement after revision.","section":"Main text, 'Topological phases of two parity-time symmetric levels'"}],"recommendation":"major_revision","confidential_remarks":"I think the core physics is sound and the paper is likely publishable after a careful revision. The main issue is not correctness of the quench calculation but the gap between the proved statement and the advertised universality. The authors should either prove the 'series of quenches' reduction under explicit assumptions or temper the abstract/conclusion to the piecewise-constant setting. The four-level counterexample is a strong point and should be retained. I would not recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a good paper with a real new result, and it should go to peer review. The core theorem—an intensity peak at a temporal interface between the two imaginary-gap phases of a PT-symmetric two-level dimer—is proven cleanly in the SI, and the four-level counterexample is a welcome piece of honesty. But the abstract overstates what is shown: the effect requires an imaginary gap, a dominant-E+ initial state, and an abrupt (piecewise-constant) interface; the proof does not justify the 'arbitrary time dependence' language for continuous ramps through the exceptional point.\n\nWhat's genuinely new: previous temporal topological localization relied on periodic driving or on a space-time swapped spatial band structure. This paper shows that the simple PT dimer already has Z2 imaginary-gap topology, with the hemisphere of the E+ eigenvector as a physical invariant. The derivation from overlap inequalities is elegant and the geometric lemma (the Bloch-sphere hemisphere condition controlling |a+| vs |a-|) is solid. The extension to 2D via momentum conservation is straightforward but useful, and the honeycomb gain/loss proposal should be experimentally realizable. The four-level counterexample is important: it shows the localization is not a generic feature of the AI class, which is exactly the kind of caveat that should be in the literature.\n\nSoft spots: First, the abstract says localization 'appears in every system that has parity-time symmetry and two coupled modes or bands.' Read literally, that is false: you also need an imaginary gap (|gamma|>c) and a suitable initial state, and the localization is a transient bump, not a sustained bound state. The main text mostly makes these conditions clear, but the abstract does not. Second, the SI's 'series of quenches' reduction justifies arbitrary time dependence only for piecewise-constant H(t). If the interface is realized by a continuous ramp, the Hamiltonian must pass through the exceptional point where the imaginary gap closes; the quench analysis cannot be applied there, and the paper gives no control of that regime. So the strong claim about arbitrary time dependence is not established. This is a moderate gap rather than a fatal one: most experiments switch parameters abruptly, and the mechanism survives that case.\n\nWho should read it: anyone working on non-Hermitian topological photonics, time-varying media, or photonic time crystals. It gives a unifying framework and a concrete experimental direction. I would be happy to referee it; with a tightened abstract and either a proof or a restriction of the continuous-time claim, it would be a strong contribution.","headline":"Solid, genuinely new mechanism for temporal topological localization in PT-symmetric two-level systems; the abstract overclaims universality, and the proof only covers abrupt interfaces, but this deserves a real referee.","tokens_in":20924,"tokens_out":6989,"would_cite":true,"duration_ms":86164,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Any parity-time-symmetric system with two coupled modes shows topological wave localisation at a temporal interface, regardless of spatial dimension.","keywords":["PT symmetry","topological localisation in time","non-Hermitian photonics","imaginary gap","temporal interface","two-level systems","photonic time crystals","Bloch sphere"],"falsifier":"Simulate a PT-symmetric two-level system with H(t) = i gamma(t) sigma_z + c sigma_x, where gamma(t) sweeps smoothly from +1 to -1 at a rate slow compared to c, and measure the intensity near the crossing. If the intensity either fails to peak at the crossing or peaks with a delay inconsistent with the quench-series prediction, the static-quench assumption fails. An experiment using coupled waveguides with time-varying loss/gain would settle it directly.","tokens_in":20086,"feed_emoji":"⏱","tokens_out":3997,"duration_ms":42237,"temperature":0.7,"pith_summary":"The paper establishes that wave intensity necessarily peaks at a temporal interface in every Schrödinger system with PT symmetry and two levels or bands, in any spatial dimension. The core idea is a new classification of imaginary-gapped two-level PT-symmetric Hamiltonians into two topological phases, labelled by the hemisphere in which the upper eigenvector sits. Because switching between phases forces the pre-interface growth of intensity into post-interface decay, a localised temporal state appears. This framework covers prior photonic time-crystal observations without needing periodic driving and makes the temporal localisation a generic consequence of PT symmetry.","feed_headline":"PT symmetry makes wave peaks at time interfaces universal","feed_subtitle":"Two coupled modes under PT symmetry always amplify before a phase switch and decay after, in any dimension.","key_machinery":"The central object is the two-level PT-symmetric Hamiltonian H = i gamma sigma_z + (c e^{i phi} sigma_+ + h.c.), with imaginary gap |gamma| > c, whose eigenvectors are forced by PT symmetry to be reflections of each other across the Bloch-sphere equator. This yields two disjoint gapped phases identified by the sign of gamma and by whether the E+ eigenvector lies in the +z or -z hemisphere. The dynamical argument is a quench-overlap lemma: for a same-phase quench the pre-quench E+ eigenvector best overlaps the post-quench E+ eigenvector, while for a cross-phase quench it best overlaps the post-quench E- state, reversing the intensity slope. A temporal interface is treated as a series of same-","core_discovery":"The paper's central claim is that the Z2 imaginary-gap classification of zero-dimensional non-Hermitian Hamiltonians in the AI class has a physical, dynamical consequence for two-level PT-symmetric systems: when the Hamiltonian crosses from one topological phase to the other, the intensity of an initially dominant upper-eigenvector state must rise monotonically before the interface and fall monotonically immediately after it, producing a localised intensity peak in time. The topological phase is diagnosed by sign(gamma), equivalently by which Bloch-sphere hemisphere contains the E+ eigenvector. The same mechanism applies to each momentum sector of a translationally invariant multi-band latti","pith_inferences":["A direct test of the quench-series assumption would be to switch between phases at a finite, tunable rate; if a slow continuous sweep through the exceptional point changes the peak timing or suppresses it, the static-quench treatment is not the whole story.","The same geometric mechanism likely applies to other classical wave systems whose dynamics are Schrödinger-like, such as acoustic or mechanical metamaterials with balanced gain and loss, opening an experimental path beyond photonics.","Because the spatial profile of the temporally localised wave is unconstrained, the effect could be used to carve out space-time-localised wave packets, potentially useful for signal isolation or time-domain switching.","The identification of momentum-gap domains as carriers of orientational order suggests that other non-Hermitian symmetry classes with imaginary-line gaps may host analogous but distinct temporal localisation phenomena."],"forward_implications":["Every PT-symmetric two-band model, in zero, one, two, or three spatial dimensions, can host topological temporal localisation at an interface between the two gapped phases.","No periodic driving is needed; the time dependence within each phase may be arbitrary, including non-commuting Hamiltonians, and the localisation survives.","In one dimension, all momenta inside a momentum gap can host the localisation, and the gap itself carries eigenvector orientational order as a topological property.","In two dimensions, the analogue is a path-connected domain of imaginary energy (a momentum hole) whose orientational order can be switched at a temporal interface to create a localised wave of freely chosen spatial profile.","The framework unifies existing experimental observations in photonic time crystals and predicts the effect in any platform realising two coupled PT-symmetric modes.","The effect is not universal in the entire non-Hermitian AI class: a four-level counterexample shows a topological phase change without intensity localisation."],"supporting_citations":[{"why":"Supplies the Z2 imaginary-line-gap classification of zero-dimensional non-Hermitian models in the AI class that the two topological phases realise.","marker":"[38]"},{"why":"First proposed momentum-gap topology and temporal interface localisation in photonic time crystals, which this paper generalises to a broad class.","marker":"[27]"},{"why":"Contains the detailed quench-series proofs, the higher-dimensional generalisation, and the four-level counterexample supporting the universality claim.","marker":"[46]"},{"why":"Provides the experimentally realised two-dimensional PT-symmetric honeycomb lattice used to illustrate the higher-dimensional case.","marker":"[47]"}],"fun_headline_variants":["PT symmetry forces wave peaks at time interfaces","Topological time peaks without periodic driving","Universal time localization from PT-symmetric pairs","Two coupled modes under PT symmetry yield time peaks","PT symmetry makes time-interface peaks universal in any dimension"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The argument assumes that a temporally varying Hamiltonian can be treated as a series of instantaneous quenches, so that arbitrarily slow or continuous passage through the phase boundary does not change the intensity dynamics.","fun_headline_variants_meta":{"raw":{"variants":["PT symmetry forces wave peaks at time interfaces","Topological time peaks without periodic driving","Universal time localization from PT-symmetric pairs","Two coupled modes under PT symmetry yield time peaks","PT symmetry makes time-interface peaks universal in any dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000178,"raw_usage":{"total_tokens":1109,"prompt_tokens":696,"completion_tokens":413,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":440,"completion_tokens_details":{"reasoning_tokens":344}},"tokens_in":440,"tokens_out":413,"duration_ms":5038,"temperature":1.0,"reasoning_tokens":344,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T23:16:23.254614+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a PT-symmetric two-level system with H(t) = i gamma(t) sigma_z + c sigma_x, where gamma(t) sweeps smoothly from +1 to -1 at a rate slow compared to c, and measure the intensity near the crossing. If the intensity either fails to peak at the crossing or peaks with a delay inconsistent with the quench-series prediction, the static-quench assumption fails. An experiment using coupled waveguides with time-varying loss/gain would settle it directly.","supporting_citations":[{"cited_title":"Lustig , author Y","cited_arxiv_id":null,"evidence_quote":"First proposed momentum-gap topology and temporal interface localisation in photonic time crystals, which this paper generalises to a broad class."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Contains the detailed quench-series proofs, the higher-dimensional generalisation, and the four-level counterexample supporting the universality claim."},{"cited_title":"Kremer , author T","cited_arxiv_id":null,"evidence_quote":"Provides the experimentally realised two-dimensional PT-symmetric honeycomb lattice used to illustrate the higher-dimensional case."}],"review_version":1}