{"id":"7e5a31ce-a8c5-4800-a609-801f8a35f7ac","arxiv_id":"1908.07712","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Bulk wave-packet dynamics, through the drift-velocity dependence of the Lyapunov exponent, can reveal the non-Hermitian skin effect and non-Bloch symmetry-breaking transitions.","lead":"The paper shows that the growth rate (Lyapunov exponent) of a wave packet moving in the bulk of a non-Hermitian crystal can detect the non-Hermitian skin effect and phase transitions that are invisible in ordinary Bloch bands. This gives experiments a way to probe these exotic effects without looking at the edges.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Core bridge to OBC spectra rests on the unproven Appendix B conjecture that every saddle point of Q belongs to the generalized Brillouin zone; a counterexample would break the claimed general probing scheme.","rationale":"The reader's weakest assumption is exactly the load-bearing concern: the link between the steepest-descent saddle points and the OBC/non-Bloch spectrum is the unproven conjecture in Appendix B. I agree with the CONDITIONAL verdict because the paper's own text flags the conjecture, admits a non-rigorous \"likely\" step, and documents exceptional cusp cases in Appendix C where the criterion fails. The analytic derivation of Eq. (25) is standard and the numerical agreement in Figs. 3 and 4 for the four models is genuine independent support, but those checks do not establish the \"rather generally\" claim in the abstract. The proposed random-polynomial test would settle whether the conjecture holds beyond the four model families, and would thereby determine whether the central interpretation survives as a general principle or must be delimited to non-cuspidal, saddle-controlled cases. No stronger internal inconsistency is apparent, so the verdict should remain CONDITIONAL rather than being upgraded or rejected.","tokens_in":20795,"tokens_out":8061,"duration_ms":91108,"concrete_test":"Perform a numerical search over random finite-range Laurent polynomials Q(β) with, for example, coefficients σ_n for |n|≤R drawn from a generic complex distribution, excluding cases where Q=0 on the unit circle. For each instance, compute all saddle points β_s of dQ/dβ=0, then check whether each β_s lies on the generalized Brillouin zone (using the root-ordering condition |β_M|=|β_{M+1}| for the characteristic equation at E=±√Q(β_s)) and whether E(β_s) is a turning point of the numerically obtained OBC spectrum for a large chain (N≈200). Repeat over several thousand samples. If any counterexample appears, the Appendix B conjecture and the claimed general probing interpretation are falsified; if none appears, the conditional acceptance is strengthened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central chain is: steepest descent gives λ(v)=Im E(k_s)−v Im k_s (Eq. 25); at v=0, k_s is a saddle point of Q(β). To conclude that λ(0) probes the OBC/non-Bloch spectrum, the paper needs every saddle point of Q to lie on the generalized Brillouin zone and to be a turning point of the OBC spectral arcs. This is precisely the Appendix B conjecture, stated as \"We conjecture that such properties... are rather general ones.\" The supporting argument around Eqs. (B1)–(B4) only exhibits pairs of β branches with equal modulus for energies near the saddle; it does not prove that these are the ordered middle roots |β_M|=|β_{M+1}| required by the GBZ construction, and the text itself says such pairs are \"likely\" to belong to the GBZ. The paper also concedes that the saddle-point criterion is not necessary: Appendix C and the cusp/Bloch-point examples show systems with NHSE but vm=0, so λ(v) can fail to signal the skin effect even when the saddle link holds. Because properties (i)–(iii) and the abstract's \"rather generally\" claim all depend on this bridge, the load-bearing weakness is the unproved universality of the saddle-point/GBZ correspondence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies one-dimensional non-Hermitian two-band tight-binding models and proposes that the long-time Lyapunov exponent lambda(v) of bulk real-space wave-packet dynamics along the space-time ray n=vt can probe non-Bloch (OBC) spectral features, namely the non-Hermitian skin effect (NHSE) and non-Bloch symmetry-breaking phase transitions. For a dominant saddle point k_s of E(k)-vk, the steepest-descent result lambda(v)=Im(E(k_s))-v Im(k_s) is derived (Eq. (25)) and verified numerically for four non-Hermitian SSH-type models. The paper further claims that lambda(v) peaks at nonzero drift velocity v_m iff the model displays the NHSE (property (iii)), and that a non-Bloch PT transition in model III is visible as lambda(0)=0 below the threshold and lambda(0)=sqrt(delta^2-t^2) above it (Eq. (36)), with numerical agreement.","tokens_in":21104,"tokens_out":3564,"duration_ms":134091,"significance":"If the central claims hold, the paper offers an experimentally relevant bulk-dynamics probe of OBC spectra and non-Bloch phase transitions in non-Hermitian lattices, without requiring access to edge states. The analytical steepest-descent derivation is standard, and the concrete predictions for models I-III, especially the parameter-free formula Eq. (36), are confirmed by direct numerical simulation with no fitting parameters. These are genuine strengths. However, the claimed 'rather general' link between the Lyapunov exponent and OBC spectra depends on an unproved conjecture about saddle points and the generalized Brillouin zone, and one step in the proof of property (iii) is asserted rather than demonstrated. The paper's concrete model results are valuable, but the universality claim needs substantial qualification or proof.","major_comments":[{"comment":"The statement that every saddle point beta_s of Q(beta) belongs to the generalized Brillouin zone, and that OBC spectral arcs turn at such saddle points, is explicitly conjectural ('We conjecture that such properties... are rather general ones'). The supporting argument around Eqs. (B1)-(B4) only shows that near a saddle point one can find pairs of beta branches with equal modulus and equal Q; it does not prove that these branches are the adjacent ordered roots |beta_M|=|beta_{M+1}| required by the GBZ construction, and the text itself says such pairs are 'likely' to belong to C_tilde_beta. Because property (iii) and the interpretation of lambda(0) as an OBC-spectral probe rely on this bridge, the central universality claim is not established. The authors should either provide a proof of the saddle-point/GBZ correspondence or explicitly restrict the claims to the verified model classes.","section":"Appendix B and Sec. III"},{"comment":"In the paragraph after Eq. (32), the non-NHSE case is handled by asserting that at the real Bloch wave number k0 maximizing Im(E(k)), the Bloch energy has a dominant saddle point with Re(E'(k0))=0. This assertion is not proved and is not an immediate consequence of PBC and OBC spectra coinciding. The implication 'no NHSE implies v_m=0' is load-bearing for the diagnostic claim, so the proof is incomplete as written. A derivation or an explicit condition under which Re(E'(k0))=0 holds is needed.","section":"Sec. IV, proof of property (iii)"},{"comment":"Appendix C provides an explicit model with NHSE in which all saddle points lie on the unit circle and the Lyapunov exponent takes its maximum at v_m=0. Thus property (iii) is only a sufficient condition and is not a universal signature of the NHSE. The paper's phrasing in Sec. IV that v_m is a 'clear signature of the existence of the NHSE' overstates the result. The authors should state precisely the generic conditions (e.g., absence of cusp singularities and Bloch-point-saddle coincidences) under which the diagnostic applies, and clearly separate the sufficient criterion from the universal claim.","section":"Appendix C and Sec. IV, property (iii)"}],"minor_comments":[{"comment":"There is a typo in the first paragraph: 'Lypaunov exponent' should be 'Lyapunov exponent'.","section":"Sec. V"},{"comment":"The acronym 'NSHE' is used once in the Introduction and should be 'NHSE' for consistency with the rest of the paper.","section":"Introduction"},{"comment":"The steepest-descent result assumes that the spectral amplitude G_+(k_s) is nonzero at the dominant saddle point. The claim that the measured lambda(v) is insensitive to the initial condition should be qualified, since a specially chosen initial wave packet with G_+(k_s)=0 could remove the dominant saddle contribution and change the asymptotic exponent.","section":"Sec. IV, Eq. (31)"},{"comment":"The statement that models II and III are 'basically equivalent' is explained only briefly; a one-sentence clarification of the unitary transformation connecting them would help readers who want to compare Eqs. (A4) with the real-space models in Figs. 1(a) and 1(b).","section":"Appendix A, model II/III"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the unproved Appendix B conjecture, which carries the paper's headline claim of generality. The concrete model calculations and the parameter-free prediction Eq. (36) are sound and publishable after the scope is narrowed and the proof of property (iii) is completed or explicitly conditioned. I would also ask the authors to check whether subsequent literature has proven or disproven the saddle-point/GBZ correspondence, since that would affect how the conjecture is presented."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper gives a genuinely new and useful diagnostic — the Lyapunov exponent of bulk wave-packet dynamics, computed from ordinary Bloch data, carries information about the non-Hermitian skin effect and non-Bloch phase transitions. The steepest-descent derivation of λ(v) is standard, but the specific application to two-band non-Hermitian lattices, with properties (i)–(iii), is new, and the numerical checks are convincing. The analytic prediction for Model III, λ(0)=sqrt(δ²−t²), is a parameter-free formula confirmed by simulation; that is the kind of evidence that makes the paper worth taking seriously.\n\nWhere it gets delicate is the bridge from the Lyapunov exponent to the OBC non-Bloch spectrum. That bridge is the Appendix B conjecture: every saddle point of Q(β) lies on the generalized Brillouin zone and the OBC spectral arcs turn at those saddles. The paper is honest that this is a conjecture, and the supporting argument near Eqs. (B1)–(B4) only shows equal-modulus branches can be found near a saddle; it does not show they are the ordered middle roots used in the standard GBZ construction. So the abstract's 'rather generally' is stronger than what is proven. The paper also concedes in Appendix C that the saddle-point criterion is not necessary: there are NHSE systems with all saddle points on the unit circle, hence vm=0, so property (iii) can fail as a diagnostic. That is not a fatal objection to what is actually derived, but it should be clearly delimited in the final version.\n\nA smaller soft spot: the proof of property (iii) for the no-NHSE direction is fairly quick. It essentially asserts that when OBC and PBC spectra coincide, the dominant saddle at v=0 sits on the real axis. That probably holds for the models considered, but it deserves a cleaner argument.\n\nOverall the paper is solid, well written, and the limitation is flagged rather than hidden. I would send it to a serious referee. The referee should ask for a precise statement of the conjecture's scope and, ideally, either a proof for a reasonably broad class or at least a sharper characterization of when the saddle/GBZ correspondence breaks down. The reader's conditional verdict is fair. I disagree with any reading that treats Appendix C as a minor aside; it is a genuine counterexample to the universal reading of the abstract, but it does not undermine the specific models and the analytic prediction. The citation pattern looks appropriate, including the self-citations to previous work on convective/absolute PT breaking, which are relevant. This is not a desk reject.","headline":"A clean steepest-descent result with an honest conjecture at the hinge; the numerics are good, but the abstract's 'rather generally' is stronger than what is proven.","tokens_in":21574,"tokens_out":2875,"would_cite":true,"duration_ms":25552,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A single Lyapunov exponent extracted from bulk wave-packet dynamics reveals both the non-Hermitian skin effect and non-Bloch phase transitions.","keywords":["non-Hermitian skin effect","non-Bloch band theory","Lyapunov exponent","non-Hermitian SSH model","generalized Brillouin zone","non-Bloch phase transition","bulk wave-packet dynamics","saddle point criterion"],"falsifier":"Measure $\\lambda(v)$ from bulk dynamics in the cusp model at the special parameters where all saddle points lie on the unit circle: the paper predicts the maximum at $v=0$ despite the skin effect. If a nonzero $v_m$ appears there, the claimed exception and the criterion's domain are wrong; more broadly, finding a non-Hermitian two-band chain with a saddle point off the unit circle but no skin effect would refute the sufficient criterion.","tokens_in":20598,"feed_emoji":"📈","tokens_out":8847,"duration_ms":167856,"temperature":0.7,"pith_summary":"This paper asks whether two boundary-sensitive features of non-Hermitian crystals—the skin effect, which pushes bulk eigenstates to an edge, and non-Bloch symmetry-breaking transitions, which alter the open-boundary spectrum—can be read from wave motion far inside the crystal, where ordinary Bloch bands are supposed to control everything. The author's answer is yes: the Lyapunov exponent $\\lambda(v)$, which measures the long-time growth of a wave packet moving along the ray $n=vt$, carries an imprint of the non-Bloch bands. A peak of $\\lambda(v)$ at a nonzero drift velocity $v$ signals the skin effect, and the zero-drift value $\\lambda(0)$ jumps from zero to a positive value across the non-Bloch parity-time transition of model III. This matters because bulk wave-packet experiments could detect edge-related physics without ever approaching the edges.","feed_headline":"Bulk wave packets reveal the non-Hermitian skin effect","feed_subtitle":"Long-time growth along a drifting ray flags edge localization and non-Bloch transitions without touching the edges.","key_machinery":"The central object is the Lyapunov exponent $\\lambda(v)$ of bulk propagation along the space-time ray $n=vt$. It is evaluated by steepest descent after analytically continuing the Bloch bands $E_\\pm(k)$ to complex $k$; the dominant saddle $k_s$ satisfies $dE_\\pm/dk=v$ and $\\lambda(v)=\\mathrm{Im}(E_\\pm(k_s))-v\\,\\mathrm{Im}(k_s)$. The same saddle points, written as $\\beta_s=\\exp(ik_s)$, are saddle points of $Q(\\beta)=E^2$, and the paper argues, conjecturally in general, that they lie on the generalized Brillouin zone and are precisely the turning points of the open-boundary spectrum. The coalescence of such saddles marks a non-Bloch phase transition.","core_discovery":"The central claim is that long-time bulk wave-packet dynamics, although built from ordinary extended Bloch states, is asymptotically controlled by the saddle points of the analytically continued band dispersion, and those saddle points coincide with the turning points of the open-boundary (non-Bloch) spectrum. For a path $n=vt$, the Lyapunov exponent is $\\lambda(v)=\\mathrm{Im}(E(k_s))-v\\,\\mathrm{Im}(k_s)$, where $k_s$ solves $dE/dk=v$ at the dominant saddle; its maximum occurs at drift velocity $v_m$. If $v_m\\neq 0$, the system exhibits the non-Hermitian skin effect; in model III the zero-drift exponent is $\\lambda(0)=0$ below $\\delta=t$ and $\\lambda(0)=\\sqrt{\\delta^2-t^2}$ above it, revealing the non-Bloch parity-time transition. The author states the saddle-point criterion as sufficient and rather general: a saddle point of $Q(\\beta)=E^2$ off the unit circle implies the skin effect, while noting the criterion is not strictly necessary because exceptional cusp models can place all saddles on the unit circle yet still show the effect.","pith_inferences":["If the saddle-point connection to the generalized Brillouin zone holds in higher-dimensional or synthetic-dimensional lattices, the same $\\lambda(v)$ protocol could probe non-Bloch transitions there; the paper raises this as an open problem, not a proven result.","The exceptional cusp cases imply that $v_m=0$ is not proof of the absence of the skin effect; a null result from the bulk probe should be interpreted alongside the known spectrum shape, not as a standalone verdict.","One testable extension is to tune deliberately into the cusp condition of the appendix model and monitor $\\log|\\psi(t)|$: the predicted $\\lambda(v)$ curve with its peak at $v=0$ would directly test the boundary of the method's validity.","A practical extension is to replace the asymptotic fit by transient-time measurements over a propagation time of roughly five to ten hopping units, which the numerics show is already enough to extract $\\lambda$ with good accuracy."],"forward_implications":["The drift velocity at which $\\lambda(v)$ is maximal provides a boundary-free diagnostic: $v_m=0$ means no skin effect in the generic case, while $v_m\\neq 0$ means the open-boundary spectrum deviates from the Bloch bands.","The zero-drift Lyapunov exponent gives an order parameter for non-Bloch symmetry-breaking transitions; in model III it is $0$ in the unbroken phase and $\\sqrt{\\delta^2-t^2}$ in the broken phase, so a single bulk measurement locates $\\delta=t$.","The saddle-point criterion reduces the existence of the skin effect to a property of the periodic-boundary dispersion $Q(\\beta)$: a saddle point off the unit circle is sufficient, independent of boundary details.","Because the prediction is insensitive to the initial excitation and matches direct numerical integration, the same protocol transfers to experimental platforms where these non-Hermitian SSH models are already realized."],"supporting_citations":[{"why":"Supplies model III, the parity-time-symmetric non-Hermitian SSH chain whose open-boundary transition is probed by $\\lambda(0)$.","marker":"[4]"},{"why":"Introduces the non-Bloch band theory and generalized Brillouin zone used to compute open-boundary spectra.","marker":"[18]"},{"why":"Analyzes skin modes and the periodic-to-open spectral flow that underlies the open-arc picture.","marker":"[21]"},{"why":"Defines the generalized Brillouin zone for non-Hermitian Bloch band theory, a basis for the saddle-point argument.","marker":"[26]"},{"why":"Introduces Bloch points and real-space non-Hermitian topological invariants, needed for model IV and the exceptional cusp analysis.","marker":"[27]"},{"why":"Reports experimental observation of bulk-boundary correspondence breakdown in topolectrical circuits, motivating the probe's experimental relevance.","marker":"[37]"}],"fun_headline_variants":["Lyapunov exponent reveals skin effect via bulk drift","Bulk wave packets expose non-Bloch phase transitions","Drifting rays detect edge localization without edges","Saddle points in bulk govern non-Bloch transitions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is a conjecture: at the complex wave numbers where the band derivative vanishes, the open-boundary spectrum always bends around, and those points always lie on the generalized Brillouin zone. The paper admits it cannot prove this generally and records exceptional cusp cases where the saddle points stay on the unit circle while the skin effect still exists.","fun_headline_variants_meta":{"raw":{"variants":["Lyapunov exponent reveals skin effect via bulk drift","Bulk wave packets expose non-Bloch phase transitions","Drifting rays detect edge localization without edges","Saddle points in bulk govern non-Bloch transitions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000455,"raw_usage":{"total_tokens":2273,"prompt_tokens":918,"completion_tokens":1355,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":534,"completion_tokens_details":{"reasoning_tokens":1292}},"tokens_in":534,"tokens_out":1355,"duration_ms":416703,"temperature":1.0,"reasoning_tokens":1292,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:58:33.007014+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure $\\lambda(v)$ from bulk dynamics in the cusp model at the special parameters where all saddle points lie on the unit circle: the paper predicts the maximum at $v=0$ despite the skin effect. If a nonzero $v_m$ appears there, the claimed exception and the criterion's domain are wrong; more broadly, finding a non-Hermitian two-band chain with a saddle point off the unit circle but no skin effect would refute the sufficient criterion.","supporting_citations":[{"cited_title":"Lee and R","cited_arxiv_id":null,"evidence_quote":"Analyzes skin modes and the periodic-to-open spectral flow that underlies the open-arc picture."},{"cited_title":"Yokomizo and S","cited_arxiv_id":null,"evidence_quote":"Defines the generalized Brillouin zone for non-Hermitian Bloch band theory, a basis for the saddle-point argument."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces Bloch points and real-space non-Hermitian topological invariants, needed for model IV and the exceptional cusp analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports experimental observation of bulk-boundary correspondence breakdown in topolectrical circuits, motivating the probe's experimental relevance."}],"review_version":1}