{"id":"c8b2ab96-6ff5-40be-a219-c510e5dceadc","arxiv_id":"1908.09438","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A modified periodic boundary condition with a decay parameter b makes the bulk-edge correspondence work for a non-Hermitian SSH model in an enlarged parameter space.","lead":"This paper proposes a new kind of periodic boundary condition for non-Hermitian quantum wires that preserves the skin effect, allowing topological edge states to be predicted in the usual way. It could make it easier to classify and design non-Hermitian topological materials, such as topological lasers.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Generalized BEC is not established for arbitrary points in the w=1 region because the proof uses circular mpbc contours, while OBC edge states are controlled by the generically non-circular generalized Brillouin zone.","rationale":"The reader identified the circularity of the generalized Brillouin zone as the weakest assumption, and this stress-test agrees that it is the load-bearing point. The concern is not that the numerical results are wrong; the mpbc spectra and the optimal-path evolution in Sec. V are coherent and supportive. Rather, the paper's generalized claim is stronger than what is proven: the proof by smooth deformation connects circular mpbc contours to the Hermitian limit, but OBC edge states are governed by the actual generalized Brillouin zone, which the paper itself admits is generically non-circular. The optimal path probes only one slice of the red region. Thus the verdict remains CONDITIONAL: the central claim is plausible and well supported for the perturbative regime and for the optimal path, but the full finite-region statement requires either a non-Bloch proof or an explicit demonstration that every red-region point has an OBC zero mode. The proposed test would settle whether the concern is realized or merely formal.","tokens_in":12754,"tokens_out":12918,"duration_ms":144831,"concrete_test":"Pick a point inside the w=1 (red) region of Fig. 2(a) that is not on the optimal path b = |β(Ebot)|, for example t1 = 0.2 with b = 0.8 if it lies in the red region. Compute the OBC spectrum at L = 200 and search for a zero-energy eigenstate with edge localization. Independently compute the non-Bloch winding number along the actual generalized Brillouin zone Cβ of Ref. 12 at the same τ. If no OBC zero mode exists, or if the winding along Cβ differs from (w+, w-) = (1, -1) computed on the circle |β| = b, then the circular mpbc criterion is not sufficient for the generalized correspondence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that every τ whose projection lies in the w=1 region of Fig. 2(a) exhibits a zero-energy OBC edge state. The argument in Sec. IV shows only that w± computed on the circular contour β = b e^{ik} is deformation-invariant and that this contour is smoothly connected, in the enlarged parameter space τ~ = {τ, b}, to the Hermitian η0 contour. It does not show that the OBC gap-closing condition coincides with the zero set of R±(β) on that circle for arbitrary b. Indeed, the authors explicitly restrict the proof to the perturbative regime, defer a general proof to future work (footnote 64), and acknowledge that the generalized Brillouin zone is generically non-circular (Sec. IV, discussion of Refs. 11 and 12). The optimal path b = |β(Ebot)| in Sec. V is a special choice; edge-state existence under OBC cannot depend on the auxiliary parameter b, so every other point of the w=1 region inherits the edge-state prediction only by an unproven continuity assumption. If a red-region point with b differing from |β(Ebot)| lacks an OBC zero mode, the generalized bulk-edge correspondence as stated is not valid.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a modified periodic boundary condition (mpbc) parameterized by b for non-Hermitian one-dimensional tight-binding models with skin effect. For a non-Hermitian SSH model with third-nearest-neighbor hopping and chiral symmetry, the authors define chiral winding numbers w± on the circular contour β = b e^{ik} (Eq. (19)), compute an analytic phase diagram in the enlarged parameter space (t1, b), and claim a generalized bulk-edge correspondence: points with (w+, w−) = (1, −1) in the region smoothly connected to the Hermitian topological phase host zero-energy edge states under open boundary conditions. The paper also introduces an optimal path b = |β(Ebot)|, chosen so that the mpbc spectrum reproduces the bottom of the OBC band, and numerically follows the evolution from mpbc to OBC.","tokens_in":12934,"tokens_out":8381,"duration_ms":81171,"significance":"If fully established, the mpbc construction would be a useful and practical way to restore the bulk-edge correspondence for non-Hermitian systems with skin effect, and the explicit analytic phase diagram for a concrete model is a valuable pedagogical and methodological contribution. The winding-number definitions, the similarity transformation in Sec. III, and the phase-boundary equations in Sec. IV are explicit and correct as far as they go, and the numerical spectra along the optimal path support the existence of protected zero modes in that specific setting. However, the central proof is only a deformation sketch, the authors explicitly restrict it to the perturbative regime, and they defer the general proof to Ref. 64, so the abstract's finite-region claim is considerably stronger than what is demonstrated.","major_comments":[{"comment":"The proof of the generalized bulk-edge correspondence is not completed for arbitrary points in the w = 1 region. The winding numbers are evaluated on the circular contour β = b e^{ik}, but the authors themselves note in the discussion of Refs. 11 and 12 that the OBC generalized Brillouin zone is generically non-circular, and they restrict the proof to the perturbative regime, deferring a more general proof to Ref. 64. The smooth deformation of the circular-contour phase diagram as γ1, γ2 → 0 only shows that these circular-contour winding numbers vary continuously; it does not identify the OBC gap-closing condition with the zero set of R±(β) on that circle for every b. Thus the abstract's claim that a finite region in (t1, b) indicates a topologically protected OBC edge state is not established for all points in the region.","section":"Sec. IV, Eq. (19) and the paragraph after Eq. (23)"},{"comment":"The optimal path ηbar is defined by b = |β(Ebot)|, which is extracted from the OBC generalized Brillouin zone and hence from OBC spectral data. The subsequent numerical evolution from mpbc to OBC along this path therefore partially builds in the presence of the edge state. Because the OBC spectrum is independent of the auxiliary parameter b, the observation of zero modes on ηbar does not prove that every other red-region point with a different value of b hosts an OBC zero mode. A concrete test would be to compute the OBC spectrum for a point in the w = 1 region with b far from |β(Ebot)|; the stated correspondence requires such a point to display the zero mode.","section":"Sec. V, Eq. (24) and Fig. 6"},{"comment":"The assertion that 'on any path η smoothly connected to η0, the correspondence ... is guaranteed' is asserted rather than proved. The smooth connection between phase boundaries in the (τ, b) space does not by itself transfer the usual PBC/OBC bulk-edge correspondence to the non-Hermitian setting, because the applicability of Ref. 61 to the mpbc reference geometry is precisely the point at issue. This is a load-bearing gap: the central result of the paper depends on this transfer, and the manuscript's own sentence restricting the proof to the perturbative regime and citing an in-preparation general proof makes the gap explicit.","section":"Sec. IV, paragraph following Eq. (23)"}],"minor_comments":[{"comment":"Equation (21) is ambiguous as printed: the text requires w = (w+ − w−)/2, since w = 1 for (w+, w−) = (1, −1) and w = 1/2 for (1, 0) or (0, −1), but the displayed expression appears to be a half-sum with an overall sign. Please correct or clarify.","section":"Eq. (21)"},{"comment":"The caption should state explicitly that the plotted |β| is |β(Ebot)| and that the gray bars outline the w = 1 region in the (t1, log b) plane, so that the reader can connect the figure to Eq. (24) without inferring it from the main text.","section":"Fig. 4 caption"},{"comment":"There is a typo in the phrase 'edge geometrygedge'; it should read 'edge geometry g_edge'. Similar spacing artifacts appear elsewhere in the typeset text and should be corrected.","section":"Sec. I, paragraph 2"}],"recommendation":"major_revision","confidential_remarks":"The paper's advertised result is a proof of a generalized bulk-edge correspondence, but the proof is explicitly restricted to the perturbative regime and the general case is deferred to Ref. 64, which is 'in preparation'. This is a significant mismatch between the abstract and the content. The authors should either include the missing general argument or carefully rescope the claims to the perturbative regime and the optimal path, while making explicit that the w = 1 region as a whole is a conjecture supported by numerics rather than a proved prediction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's core idea is genuinely new: impose a modified periodic boundary condition parameterized by b, so that the bulk geometry explicitly absorbs the skin effect via a similarity transformation, and then define winding numbers on the circle β = b e^{ik}. This is a clean and useful device. For the non-Hermitian SSH model they actually analyze, the winding-number computation and the phase diagram in (t1, b) space are correct, and the numerical spectra support the edge-state predictions along the optimal path. The half-integer winding regions (w = 1/2) are a nice byproduct, though the paper honestly leaves their edge interpretation open.\n\nThe main soft spot is the gap between the model-specific demonstration and the generalized bulk-edge correspondence claimed in the title and abstract. The proof in Sec. IV is a deformation argument that leans on the Hermitian limit and requires paths in the enlarged space (τ, b) to stay smoothly connected to η0. The authors themselves restrict the proof to the perturbative regime and defer a general proof to footnote 64. That is an honest limitation, but it means the central claim is narrower than the abstract suggests.\n\nThe stress-test concern about the circularity of b is worth taking seriously but not fatal for this model. On the optimal path b = |β(Ebot)|, the choice of b uses OBC data, so the correspondence along that path is partly built in. However, the paper does not show that every point in the red region of Fig. 2(a) corresponds to an OBC zero mode for arbitrary b. For a fixed physical Hamiltonian, b is not a tunable parameter; it is a property of the skin effect. The authors treat it as an independent coordinate, which is fine for constructing a phase diagram, but the edge-state prediction for a given physical system should not depend on which b you pick. They do show that the optimal path stays in the w = 1 region until the transition point, which is compelling for this model, but the general inference is exactly where the proof is weakest.\n\nThe citation pattern looks fine; the relation to the generalized Brillouin zone method of Yao-Wang and Yokomizo-Murakami is acknowledged, and the differences are explained reasonably. The paper does not oversell the half-integer phases.\n\nWho is this for? People working on non-Hermitian topological models, especially those interested in practical recipes for computing topological numbers in the presence of skin effect. It is a useful contribution, but it is a model study with a suggestive generalization, not a fully general proof.\n\nI would send it to peer review. A good referee should push the authors to clarify the status of the generalized claim, perhaps by testing a second model with a non-circular generalized Brillouin zone or by proving the deformation argument under fewer restrictions. Even if the general claim needs trimming, the mpbc construction is worth publishing.","headline":"A fresh boundary-condition trick that makes non-Hermitian bulk-edge correspondence intuitive for a chiral SSH model, but the general proof is explicitly restricted to the perturbative regime and the role of the free parameter b remains under-examined.","tokens_in":13492,"tokens_out":712,"would_cite":true,"duration_ms":9819,"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":"This paper restores bulk-edge correspondence in non-Hermitian chains by replacing the ordinary periodic boundary condition with a modified one parameterized by a skin-effect scale $b$, and proves the correspondence in the enlarged…","keywords":["non-Hermitian topology","bulk-edge correspondence","non-Hermitian skin effect","modified periodic boundary condition","chiral winding number","non-Hermitian SSH model","generalized Brillouin zone"],"falsifier":"Numerically diagonalize the open-boundary SSH chain at a point where the mpbc phase diagram gives $(w_+,w_-)=(1,-1)$ but which is not smoothly connected to the Hermitian line; if no zero-energy edge state appears, the deformation argument is insufficient. Alternatively, compute the actual generalized Brillouin zone by the method of Ref. 12 and compare its winding number with the mpbc winding number at $b=|\\beta(E_{\\mathrm{bot}})|$; a disagreement marks the boundary of the proof's validity.","tokens_in":12470,"feed_emoji":"🔗","tokens_out":10113,"duration_ms":91671,"temperature":0.7,"pith_summary":"Non-Hermitian systems with asymmetric hopping pile their bulk wave functions against one boundary, so ordinary periodic boundary conditions do not reproduce the open-boundary physics and the standard bulk-edge correspondence appears to fail. This paper restores it by a modified periodic boundary condition (mpbc) in which wave functions obey $\\psi(j+L)=b^L\\psi(j)$, introducing a real parameter $b$ that encodes the non-Hermitian skin effect. On this boundary condition the system has a chiral Bloch-like Hamiltonian in $\\beta = b e^{ik}$, and the paper proves that the pair of winding numbers $w_\\pm$ computed in the enlarged parameter space $\\tilde{\\tau}=\\{\\tau,b\\}$ controls the existence of protected zero-energy edge states under open boundary. A reader should care because the result turns the apparent failure of bulk-edge correspondence in non-Hermitian systems into a bookkeeping problem with a single extra parameter, and because the same construction exposes genuinely non-Hermitian topological phases with half-integral winding.","feed_headline":"Modified boundary condition restores non-Hermitian bulk-edge link","feed_subtitle":"Winding numbers on a circle of radius b predict edge states that ordinary periodic boundaries miss.","key_machinery":"The load-bearing object is the modified periodic boundary condition (mpbc), defined by boundary hopping terms $\\Delta H = b^{-L} t |1\\rangle\\langle L| + b^L t |L\\rangle\\langle 1|$ for the single-band chain and by the analogous terms for the SSH-type chain; it imposes $\\psi(j+L)=b^L\\psi(j)$. A similarity transformation $S=\\mathrm{diag}[1,1,b,b,b^2,b^2,\\dots]$ turns the mpbc Hamiltonian into an ordinary periodic Hamiltonian with rescaled hoppings, leaving the spectrum unchanged. This machinery replaces the ordinary Brillouin zone by the circle $\\beta=b e^{ik}$, on which the chiral winding numbers $w_\\pm = (1/2\\pi)[\\arg R_\\pm(\\beta)]_{k=0}^{2\\pi}$ are evaluated. Its essential work is to make the bulk reference geometry compatible with exponential localization of skin modes, so that the deformation argument of the Hermitian proof can be carried over to the enlarged parameter space $\\tilde{\\tau}=\\{\\tau,b\\}$.","core_discovery":"The central claim is that the modified periodic boundary condition (mpbc), in which the wave function is multiplied by $b^L$ across the boundary, is the correct reference bulk geometry for non-Hermitian systems with skin effect. Under mpbc the eigenstates take the generalized Bloch form with $\\beta=b e^{ik}$, and the bulk Hamiltonian $H_{\\mathrm{mpbc}}(\\beta)$ is explicitly chiral. The winding numbers $w_\\pm$ of the off-diagonal entries $R_\\pm(\\beta)$ around the origin are well defined, and the paper proves a bulk-edge correspondence in $\\tilde{\\tau}=\\{\\tau,b\\}$: the region with $(w_+,w_-)=(1,-1)$ signals a zero-energy edge state under open boundary, and this region is smoothly connected to the Hermitian topological-insulator region. The proof adapts the chiral-symmetry bulk-edge proof of Ref. 61, with the additional parameter $b$ absorbing the skin effect. The paper also exhibits an optimal path $b=|\\beta(E_{\\mathrm{bot}})|$ along which the mpbc spectrum reproduces the bottom of the open-boundary bulk spectrum, and shows that switching the mpbc off continuously evolves a detached spectral branch into the zero-energy edge state.","pith_inferences":["Read $b$ as the radial scale of the generalized Brillouin zone; where that zone is genuinely non-circular, the natural extension is to let $b$ depend on $k$, making the present construction the constant-$b$ slice of a more general contour formulation.","The same strategy should transfer to other symmetry classes and higher dimensions, replacing the scalar $b$ by a boundary-condition object matched to the anisotropic skin effect; a testable consequence is that topological edge-state existence is governed by winding numbers computed on such skin-effect-matched bulk geometries.","In engineered directional-gain or lossy lattices, the claim predicts that closing the ring with an amplitude-amplifying link $b^L$ should make the measured bulk spectrum follow the open-boundary gap closings, and the $w=1/2$ windows should appear as parameter ranges with mid-gap states and anomalous winding."],"forward_implications":["On any path in the $(t_1,b)$ phase diagram that is smoothly connected to the Hermitian line $b=1$, the $w=1$ region guarantees a zero-energy edge state under open boundary, whereas the ordinary periodic-boundary spectrum misplaces the gap closings.","The central $w=1$ region is smoothly connected to the Hermitian topological-insulator phase, while the $w=1/2$ regions with $(w_+,w_-)=(1,0)$ or $(0,-1)$ are new topological phases with no Hermitian analogue.","Along the optimal path $b=|\\beta(E_{\\mathrm{bot}})|$, the mpbc spectrum reproduces the bottom of the open-boundary bulk spectrum, and gradually switching off the mpbc turns a detached spectral branch into the zero-energy edge state.","The generalized bulk-edge correspondence is at present proven only in the perturbative non-Hermitian regime, namely for phases smoothly connected to a Hermitian topological phase; the authors state that a more general proof is left to future work."],"supporting_citations":[{"why":"Supplies the single-band model of asymmetric hopping whose open-boundary wavefunction is localized as $\\psi_j\\propto b^j$, motivating the modified periodic boundary condition.","marker":"[2,3]"},{"why":"Provides the non-Hermitian SSH-type tight-binding model with anisotropic hopping whose open- and periodic-boundary spectra differ, setting up the problem.","marker":"[10]"},{"why":"Introduces the non-Hermitian SSH model and demonstrates that ordinary periodic boundary conditions fail to capture the skin-effect-modified topological phases; the baseline the mpbc is designed to fix.","marker":"[11]"},{"why":"Gives the generalized Brillouin-zone construction with a non-circular contour $C_\\beta$ for open boundary conditions, which the paper contrasts with its own mpbc formulation.","marker":"[12]"},{"why":"Provides the chiral-symmetry bulk-edge correspondence proof in the Hermitian case that the paper adapts to the enlarged parameter space.","marker":"[61]"},{"why":"Documents the standard winding-number characterization of the Hermitian SSH chain, which motivates the definition of $w_\\pm$.","marker":"[63]"}],"fun_headline_variants":["Non-Hermitian bulk-edge saved by modified periodic boundary","Generalized Bloch form enables bulk-edge in non-Hermitian","One extra parameter b reconnects non-Hermitian bulk to edge","Chiral winding numbers survive skin effect via modified boundaries","Modified boundaries make non-Hermitian topology behave"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that a single number $b$ captures the whole non-Hermitian skin effect, i.e. that the generalized Brillouin zone is the circle $|\\beta|=b$; in general it is not.","fun_headline_variants_meta":{"raw":{"variants":["Non-Hermitian bulk-edge saved by modified periodic boundary","Generalized Bloch form enables bulk-edge in non-Hermitian","One extra parameter b reconnects non-Hermitian bulk to edge","Chiral winding numbers survive skin effect via modified boundaries","Modified boundaries make non-Hermitian topology behave"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000307,"raw_usage":{"total_tokens":1746,"prompt_tokens":922,"completion_tokens":824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":742}},"tokens_in":538,"tokens_out":824,"duration_ms":8910,"temperature":1.0,"reasoning_tokens":742,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:11:34.244070+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically diagonalize the open-boundary SSH chain at a point where the mpbc phase diagram gives $(w_+,w_-)=(1,-1)$ but which is not smoothly connected to the Hermitian line; if no zero-energy edge state appears, the deformation argument is insufficient. Alternatively, compute the actual generalized Brillouin zone by the method of Ref. 12 and compare its winding number with the mpbc winding number at $b=|\\beta(E_{\\mathrm{bot}})|$; a disagreement marks the boundary of the proof's validity.","supporting_citations":[],"review_version":1}