{"id":"ffba577f-e9ff-43fc-a46f-c1db0d216d0f","arxiv_id":"1908.01372","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Non-Hermitian losses in anomalous Floquet insulators let boundary states detach from bulk bands and be engineered independently, enabling new chiral and directional edge transport.","lead":"This paper shows that in non-Hermitian, periodically driven topological insulators, boundary transport can be adjusted independently of the bulk. This enables new effects such as chiral motion in the same direction on opposite edges, which ordinary topological systems forbid.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The topological-protection claim rests on a point-gap winding argument that ignores the non-Hermitian skin effect; the asserted bulk-boundary correspondence for open systems is not established.","rationale":"Good-faith reading: The paper introduces a genuinely interesting capability—boundary-site-specific losses detach Floquet edge states from the bulk in the complex quasienergy plane, enabling selective manipulation of chiral/helical transport. The analytic loss counting at perfect coupling and the real-space propagation simulations support the 'engineering' aspect. What would have to be true for the central claim to hold is that the detached boundary states are still topologically protected, i.e., robust against perturbations that preserve the relevant symmetry and the gap. The only argument offered for this is the invertibility of U and the preservation of the winding number. The load-bearing assumption is that this point-gap winding is a valid invariant for the open-boundary spectrum. In non-Hermitian systems this assumption fails in general: the OBC spectrum is determined by the generalized Brillouin zone, and a nonzero point-gap winding of the PBC propagator is the very condition for the non-Hermitian skin effect. If the skin effect is present, bulk eigenstates migrate to the boundaries, the OBC spectrum no longer matches the PBC bands, and a point gap around the origin in the PBC spectrum does not imply a gap in the OBC spectrum. The detached boundary state could then hybridize with skin modes, and its winding number would not confer protection. The manuscript contains no GBZ or skin-effect analysis, and its conclusion explicitly flags the status of bulk-boundary correspondence in non-Hermitian Floquet systems as open. The numerical disorder study (Fig. 6E,F) provides preliminary evidence of robustness for specific parameters, but it does not establish the general topological statement. This is not a matter of disagreement with current consensus; it is a missing proof step in the paper's own argument. The concrete test—comparing torus and cylinder spectra and checking eigenstate IPRs—would settle whether the skin effect is actually present in the specific protocol. If the check shows no skin effect (bulk states extended, OBC spectrum tracking PBC bands), the concern is resolved and the conditional acceptance can stand. If it shows a skin effect, the central claim's proof collapses and a CONDITIONAL verdict requiring a GBZ-based invariant would be the appropriate outcome. The reader's weakest_assumption identified exactly this point; our analysis agrees. Since the reader's verdict is already CONDITIONAL, we recommend no change to the verdict.","tokens_in":15419,"tokens_out":17486,"duration_ms":191506,"concrete_test":"Compute the non-Hermitian Floquet propagator for the bulk with periodic boundary conditions in both directions on a torus, U(k_x,k_y), for the parameters of Fig. 5 (J=1.5, J'=0.4, TRS∗ or TRSt losses). Simultaneously compute the spectrum of the same protocol on a cylinder of width L_y = 20, 40, 80 with open boundaries in y and periodic in x. Plot all complex quasienergies in the same plane, and compute the inverse participation ratio (IPR) of each eigenstate in the transverse direction. If a finite fraction of eigenstates have IPR ~ O(1) (localized at edges) and the cylinder bulk spectrum does not converge to the torus bands as L_y increases, the non-Hermitian skin effect is present; the point-gap winding number of the torus is then not a valid invariant for the open-boundary system, and the topological-protection claim in the paragraph after Fig. 1 is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that non-Hermiticity modifies boundary states 'without sacrificing their topological nature' (abstract)—is justified in the text by one sentence: 'Since the propagator U is invertible, its spectrum cannot move through the origin, and the winding number is preserved' (paragraph after Fig. 1). This argument invokes a point-gap winding number of the Floquet propagator, but applies it to a system with open boundaries. In non-Hermitian systems, the open-boundary spectrum is governed by the generalized Brillouin zone, not the real-momentum Bloch spectrum; a nonzero point-gap winding of the periodic-boundary propagator generically produces the non-Hermitian skin effect, which relocates bulk eigenstates to the edges and can close the very point gap around the origin that is assumed to protect the detached boundary states. The paper never computes the periodic-boundary spectrum of U(k_x,k_y), never checks whether the strip's bulk eigenstates are extended, and never evaluates the generalized Brillouin zone. The authors' own conclusion concedes: 'further theoretical research regarding the status of topological invariants and the bulk-boundary correspondence in non-Hermitian Floquet systems' is needed. Thus the claim that the detached boundary states retain their topological nature is an assumption, not a demonstrated result; the numerical disorder robustness in Fig. 6(E,F) is encouraging but does not replace a bulk-boundary correspondence. The engineering of boundary-state decay rates (Eqs. (2)-(3)) is sound, but the 'topological' status of the engineered states is the weakest link.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper introduces the concept of 'boundary state engineering' (BSE) for non-Hermitian Floquet topological insulators. The authors argue that in an anomalous Floquet phase, boundary-state quasienergies can be moved in the complex plane by adding losses/gains, thereby detaching boundary states from bulk bands while retaining their winding around the origin. They analyze a six-step Floquet protocol with fermionic time-reversal symmetry, derive an analytic loss budget at perfect coupling (Eqs. (2)-(3)), and identify two non-Hermitian TRS variants (TRS* and TRSt) with distinct constraints on bulk and boundary loss parameters (Eqs. (4)-(6)). Numerical quasienergy spectra and real-space propagation are used to demonstrate enhanced boundary transport, transport with a preferred direction, and same-direction chiral transport on opposite boundaries, including robustness against disorder. The supplemental material provides additional examples and a driven Kane-Mele model.","tokens_in":15741,"tokens_out":11194,"duration_ms":123582,"significance":"If the claims hold, BSE would be a new design principle for topological photonics: boundary transport is no longer fixed by bulk topology, enabling transport scenarios impossible in Hermitian systems. The paper's strengths are the clean analytic loss accounting at perfect coupling, the explicit enumeration of non-Hermitian TRS constraints, and the direct real-space simulations showing the predicted transport phenomena. The experimental relevance to photonic waveguide lattices is well argued, and the supplemental material generalizes the construction to several other protocols. However, the central topological-protection claim is only supported by a short winding-number argument and by disorder numerics; the bulk-boundary correspondence for the open non-Hermitian system is not established, which is a significant gap.","major_comments":[{"comment":"The central claim that the boundary states remain 'topologically protected' rests on the sentence 'Since the propagator U is invertible, its spectrum cannot move through the origin, and the winding number is preserved.' This argument is not sufficient for open-boundary non-Hermitian systems. In such systems, the open-boundary spectrum and eigenstates can differ dramatically from the periodic-boundary Bloch spectrum due to the non-Hermitian skin effect, and point-gap winding numbers of U(k) do not by themselves guarantee the existence or protection of detached boundary states. The manuscript never computes the periodic-boundary spectrum U(k_x,k_y), never examines the generalized Brillouin zone, and never checks whether the bulk eigenstates of the strip are extended. The conclusion itself acknowledges that further theoretical research on invariants and bulk-boundary correspondence is needed. Since the abstract's claim 'without sacrificing their topological nature' is load-bearing, the authors should either (i) provide a rigorous justification for the absence of skin effect in this reciprocal protocol, or (ii) qualify the protection claim as a numerically supported robustness statement for the specific model. This is a required revision.","section":"Conceptual argument after Fig. 1"},{"comment":"The disorder robustness central to the protection claim is shown for a single disorder configuration (δ=0.2). Topological protection is a property of phases and should be demonstrated over an ensemble of disorder realizations; a single snapshot cannot rule out a rare backscattering event. The authors should show ensemble-averaged observables, such as the average intensity transmitted along the boundary or the mean backscattered fraction, over many realizations and for a range of disorder strengths. This would substantially strengthen the claim that the boundary states remain topologically protected.","section":"Fig. 6 (E,F)"},{"comment":"The statement that 'Regular boundary states have to remain attached to the bulk bands, since otherwise the continuous dependence on momentum would be violated' is not self-evident for complex spectra. In a non-Hermitian system, a boundary-state dispersion over a periodic momentum variable is a closed loop in the complex plane, and closed loops can detach from bulk bands without violating continuity. Since the distinction between regular and anomalous phases is the conceptual basis of BSE, the authors should provide a proof or a reference establishing this dichotomy, or at least clarify the precise sense in which 'attachment' is required in the non-Hermitian setting.","section":"Fig. 1 and surrounding text"}],"minor_comments":[{"comment":"In the TRS* constraint for the horizontal couplings, 'B*_f6' appears to be a typo for 'B*_f5'; only parameters with subscript f5 are defined for step 5.","section":"Supplemental Material, Eq. (36b)"},{"comment":"The statement that normalized intensities I(r)=|ψ(r)|^2/max_r' |ψ(r')|^2 cancel the factor Γ is only valid for a global maximum at a fixed time; this should be stated more precisely, since the maximum can change in time and the cancellation is not a full gauge invariance.","section":"Shift σ(t) and normalized intensities"},{"comment":"Because the color scale is normalized to the maximum intensity at each time, the absolute decay of the signal is not visible. A color scale tied to absolute intensity, or a separate panel showing the total norm, would help the reader assess the claimed enhancement of boundary transport relative to bulk motion.","section":"Fig. 6"},{"comment":"The combined notation '(TRS∗,t)' in Eq. (8) is compact but may confuse readers; writing the two cases explicitly would improve readability.","section":"Eq. (8)"},{"comment":"The final suggestion that experiments should investigate robustness is vague; the authors could list a specific measurable signature, such as the ratio of boundary to bulk intensity as a function of the loss parameter γ, to make the experimental proposal more concrete.","section":"Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The central physics is interesting and the numerics are convincing for the specific model, but the topological-protection claim as stated is too strong given the missing bulk-boundary analysis. I would encourage the editor to allow a revision that either adds the generalized Brillouin zone analysis or carefully qualifies the claim; the paper would then be a solid contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline is simple: this is a genuinely new design principle, and the paper earns serious attention. The authors show that in an anomalous Floquet topological insulator with non-Hermitian losses, you can move boundary states around in the complex quasienergy plane independently of the bulk. That gives you effects Hermitian systems forbid—same-direction chiral transport on opposite edges, preferred-direction helical transport—and they demonstrate the idea in three extra models, including a driven Kane-Mele. The analytic loss counting at perfect coupling (Eq. 3) is clean and parameter-free: each boundary state's imaginary quasienergy is just the number of lossy steps it experiences per cycle. The TRS* vs TRSt analysis is careful and predicts real spectral features. The numerics, especially the real-space propagation and the disorder robustness, back the main claims. No fitting, no hand-waving in the mechanism itself.\n\nThe soft spot is the word “topological.” The paper justifies protection with one sentence: because U is invertible, its spectrum can't cross the origin, so the winding number is preserved. That's a homotopy argument about an existing boundary-state curve, not a proof that the detached states are protected by a bulk-boundary correspondence. Non-Hermitian systems make this subtle: the skin effect can relocalize bulk states and close point gaps, so PBC winding does not automatically control the OBC spectrum. The paper never checks whether strip bulk states are extended, never computes the generalized Brillouin zone, and never asks whether the effective Floquet Hamiltonian is non-reciprocal. The authors themselves concede this in the conclusion. So the abstract's 'without sacrificing their topological nature' is an assumption supported by numerics, not a demonstrated fact.\n\nDoes that sink the paper? I don't think so. The engineering principle stands on its own: detached boundary states with tunable relative damping are demonstrated in a concrete model. And the disorder robustness in Fig. 6 gives you practical protection even without a rigorous invariant. But it's a major soft spot for a paper that puts topology in the title. A serious referee should push for a proper non-Hermitian bulk-boundary analysis, or at least a numerical check of the skin effect.\n\nFor the right reader—someone in non-Hermitian photonics or Floquet topological phases—this is a useful and citable paper. I'd bring it to the reading group.\n\nRecommendation: accept for peer review. It deserves referee time; the revision should center on the topological claim.","headline":"Genuinely new boundary-state engineering in non-Hermitian Floquet insulators, with clean analytics and convincing numerics; the topological-protection claim is undertheorized but not fatal.","tokens_in":16255,"tokens_out":8926,"would_cite":true,"duration_ms":97757,"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":"Boundary states in anomalous Floquet topological insulators can be detached from the bulk bands and repositioned by losses without losing topological protection.","keywords":["non-Hermitian topology","Floquet topological insulator","boundary state engineering","anomalous Floquet phase","time-reversal symmetry","photonic waveguide","bulk-boundary correspondence","topological boundary transport"],"falsifier":"Diagonalize the full open-boundary Floquet propagator for a finite strip with loss rates as in Eq. (7) and track the complex quasienergies of the two counterpropagating edge states as $\\gamma_*$ or $\\gamma_t$ is varied; if their imaginary parts do not follow the predicted values (for example $-2\\gamma_*$ and $-6\\gamma_*$ on opposite boundaries) with the winding number unchanged, or if a TRS*-separated channel shows backscattering under TRS-breaking disorder comparable to the TRSt case, the central claim of independent boundary-state engineering is falsified.","tokens_in":15246,"feed_emoji":"🌀","tokens_out":17431,"duration_ms":146115,"temperature":0.7,"pith_summary":"The paper sets out to show that the strict bulk-boundary correspondence of Hermitian topological systems can be lifted in non-Hermitian Floquet insulators. Its central claim is that when the insulator is in an anomalous Floquet topological phase, adding loss or gain lets you move the boundary states in the complex quasienergy plane independently of the bulk bands, while the winding number of the one-period propagator keeps them topologically protected. The authors demonstrate this boundary state engineering in a six-step driven lattice with fermionic time-reversal symmetry, where counterpropagating helical boundary states can be selectively damped to produce helical transport with a preferred direction and chiral transport in the same direction on opposite boundaries. This matters because boundary transport in topological systems is normally fixed by the bulk, and the paper opens a practical route to controlling it in photonic waveguide lattices, where losses are intrinsic.","feed_headline":"Detach topological edge states from bulk bands with added loss","feed_subtitle":"In anomalous Floquet insulators, loss can make edge currents on opposite sides flow the same way.","key_machinery":"The load-bearing object is the one-period Floquet propagator $U(T)$, whose eigenvalues $e^{-i\\varepsilon}$ lie on the unit circle in the Hermitian case and move off it once losses are added. An anomalous Floquet boundary state is one whose dispersion winds around the unit circle; because $U$ is invertible its spectrum cannot cross the origin, so the winding number survives non-Hermitian deformation and keeps the detached boundary states protected. Boundary state engineering is implemented by assigning independent loss rates $\\gamma_r,\\gamma_b$ to the two sublattices in the bulk and $\\breve{\\gamma}_r,\\breve{\\gamma}_b$ to isolated boundary sites, constrained by the non-Hermitian time-reversal conditions TRS* ($\\gamma_r+\\gamma_b=\\breve{\\gamma}_b+\\breve{\\gamma}_r$) or TRSt ($\\gamma_r=\\gamma_b$, $\\breve{\\gamma}_b=\\breve{\\gamma}_r$). These losses shift the imaginary parts of the quasienergies, letting boundary states detach from the bulk bands while the preserved winding number maintains their topological character.","core_discovery":"The central discovery is that an anomalous Floquet topological phase combined with non-Hermiticity breaks the usual bulk-boundary lockstep. Because the Floquet propagator $U(T)$ is invertible, its spectrum cannot pass through the origin, so a boundary state that winds around the origin keeps its winding number even when losses pull its quasienergy off the unit circle; the boundary state can therefore detach from the bulk bands and be repositioned independently. In the $Z_2$-protected six-step protocol, choosing different loss rates on the two sublattices and on the two boundaries detaches the counterpropagating boundary states from the bulk, either with equal damping on both channels (the TRSt case, which preserves true bidirectional transport) or with separated imaginary parts (the TRS* case, which suppresses one direction). The result is boundary transport that is enhanced relative to bulk motion, helical transport with a preferred direction, and chiral transport in the same direction on opposite boundaries, with the TRS*-separated channel surviving symmetry-breaking disorder.","pith_inferences":["A natural testable extension is to check whether the detachment persists when nonreciprocal hopping (a non-Hermitian skin effect) is added; those systems can invalidate conventional bulk-boundary arguments, and the paper does not analyze them.","The same mechanism could be used as a reconfigurable topological switch: changing the boundary loss parameters during operation would reroute boundary current without altering the bulk phase, something the paper does not discuss.","In a balanced gain/loss (PT-type) setting, the relative-shift interpretation means the suppressed channel could become lasing or amplifying; exploring that regime goes beyond the paper's loss-only examples.","The argument based on invertibility and winding is not specific to fermionic TRS, so it may extend to other non-Hermitian Floquet phases with different symmetries, though the paper only classifies TRS* and TRSt in its example."],"forward_implications":["Boundary transport in an anomalous Floquet insulator can be enhanced relative to bulk motion by choosing bulk losses larger than the boundary losses.","With TRS* symmetry, one of the two counterpropagating boundary states can be damped more strongly, producing helical transport with a preferred direction that no longer needs time-reversal symmetry for protection.","Choosing different boundary losses on opposite boundaries makes chiral transport flow in the same direction on both edges, a configuration forbidden in Hermitian systems by the bulk-boundary correspondence.","Because the winding number of the invertible Floquet propagator is preserved, these modifications remain topologically protected even when the boundary is imperfect or disorder is present.","The same loss engineering transfers directly to photonic waveguide lattices, where gain and loss are intrinsic and can be implemented by bent waveguides."],"supporting_citations":[{"why":"Supplies the concept of anomalous Floquet topological phases, whose boundary states wind around the quasienergy Brillouin zone.","marker":"[32]"},{"why":"Provides the six-step driving protocol with fermionic time-reversal symmetry used as the paper's main example.","marker":"[25]"},{"why":"Experimental realization of the TRS-protected Floquet phase, showing the boundary-state configuration is physically accessible.","marker":"[15]"},{"why":"Defines the non-Hermitian time-reversal symmetries TRS* and TRSt that the paper uses to classify the loss configurations.","marker":"[39]"},{"why":"Shows photonic waveguide lattices as a platform for anomalous Floquet topological phases with intrinsic losses.","marker":"[29]"},{"why":"Extends the topological classification to non-Hermitian systems, supporting the use of invariants away from the unit circle.","marker":"[37]"}],"fun_headline_variants":["Loss lifts bulk-boundary lockstep in Floquet insulators","Edge states stand alone: non-Hermitian Floquet engineering","Non-Hermitian Floquet insulators unfetter boundary states","Loss gives opposite edges identical chiral flow in Floquet insulators"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the winding number of the one-period time-evolution operator remains a valid topological invariant in a non-Hermitian system with boundaries, so that boundary states moved off the bulk bands are still protected.","fun_headline_variants_meta":{"raw":{"variants":["Loss lifts bulk-boundary lockstep in Floquet insulators","Edge states stand alone: non-Hermitian Floquet engineering","Non-Hermitian Floquet insulators unfetter boundary states","Loss gives opposite edges identical chiral flow in Floquet insulators"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001292,"raw_usage":{"total_tokens":5257,"prompt_tokens":908,"completion_tokens":4349,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":524,"completion_tokens_details":{"reasoning_tokens":4274}},"tokens_in":524,"tokens_out":4349,"duration_ms":32983,"temperature":1.0,"reasoning_tokens":4274,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:14:46.253994+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Diagonalize the full open-boundary Floquet propagator for a finite strip with loss rates as in Eq. (7) and track the complex quasienergies of the two counterpropagating edge states as $\\gamma_*$ or $\\gamma_t$ is varied; if their imaginary parts do not follow the predicted values (for example $-2\\gamma_*$ and $-6\\gamma_*$ on opposite boundaries) with the winding number unchanged, or if a TRS*-separated channel shows backscattering under TRS-breaking disorder comparable to the TRSt case, the central claim of independent boundary-state engineering is falsified.","supporting_citations":[{"cited_title":"B 99, 245102 (2019)","cited_arxiv_id":null,"evidence_quote":"Provides the six-step driving protocol with fermionic time-reversal symmetry used as the paper's main example."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows photonic waveguide lattices as a platform for anomalous Floquet topological phases with intrinsic losses."}],"review_version":1}