{"id":"72706c30-ea36-4750-a8a9-1ae3bfa1fa27","arxiv_id":"1908.05687","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Zero-energy corner states in a chiral, C4v-symmetric higher-order topological insulator are protected bound states in the continuum; breaking either symmetry turns them into topological resonances.","lead":"The authors show that a two-dimensional topological lattice without a bulk gap still hosts four corner states that remain sharply localized even though ordinary bulk states exist at exactly the same energy. The result brings bound states in the continuum into closed condensed matter systems and enlarges the search space for topological phases.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The loss-probe diagnostic cannot work as stated: for H0 − iκP_R, Im E = −κ||P_Rψ||^2, so a fixed lossy region R gives a fixed nonzero decay for any non-compact BIC; Fig. 2(b)'s exponential decay is inconsistent with the claimed exponential penetration.","rationale":"The reader correctly identified the loss-probe equivalence as the weakest assumption, and I agree; the issue is sharper than 'unproven.' The exact identity Im E = −κ ||P_Rψ||^2 makes the diagnostic's logic internally inconsistent when S is fixed: a non-compact localized state has a constant overlap with R as n grows. The paper's own claim that BICs penetrate exponentially into the bulk is incompatible with the exponential decay of Im E in Fig. 2(b). This is not a disagreement with consensus; it is a mathematical property of the non-Hermitian construction. A direct Hermitian diagonalization of the zero-energy subspace would settle whether a corner-localized eigenstate of H0 exists. Because the central numerical demonstration is invalid as presented, the verdict should move from CONDITIONAL to REJECT, although a revised demonstration with the proposed check could restore the claim.","tokens_in":14431,"tokens_out":17215,"duration_ms":189609,"concrete_test":"Fix ns=3 and κ = −5×10^−2. For n = 8, 16, 32, 64, diagonalize both H0 and H = H0 − iκP_R. For the four corner-selected states of Fig. 2(b), compute p_R = ||P_Rψ||^2/||ψ||^2, both for the non-Hermitian eigenvectors and for the zero-energy H0 eigenvectors with maximal corner weight. If p_R is constant or grows with n, the vanishing Im E is an artifact of eigenvalue selection and the BIC claim fails. If p_R decreases with n, verify directly that the H0 eigenvector has exactly zero amplitude on every R site adjacent to S; otherwise no such decay mechanism exists. Repeat with ns=6: a saturation value of |Im E| that scales as e^{−ns/ξ} would show that BIC status is an artifact of the arbitrary choice of S.","verdict_should_be":"REJECT","load_bearing_attack":"The load-bearing issue is the non-Hermitian identification in Eq. (2) and Fig. 2(b). For any right eigenvector of H = H0 − iκP_R, with H0 and P_R Hermitian, Im E = −κ ||P_Rψ||^2 / ||ψ||^2. Thus a vanishing imaginary part requires the eigenstate to have vanishing support in the lossy environment R. The authors fix S to four ns×ns corner regions while the total size n grows. An exponentially localized BIC has a tail extending into R; the integrated weight in R is dominated by the region immediately outside S and is independent of n once n exceeds the localization length. Therefore Im E should saturate at a nonzero value of order κ e^{−ns/ξ}, not decay to zero as n grows. The observed decay in Fig. 2(b) can only occur if the selected states are exactly compact in S, which contradicts the stated exponential penetration and would make them eigenstates of a decoupled finite cluster rather than of the infinite lattice, or if the limit is taken with ns or κ varying, which is not described. The method therefore does not establish that the four states are eigenstates of the original Hermitian H0 at zero energy, and the central claim is unsupported by the presented numerics.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that two-dimensional higher-order topological insulators can host corner-localized bound states in the continuum despite having gapless zero-energy bulk bands. To identify such states, the authors propose a non-Hermitian loss-probe method: a small lossless region S (four corner squares) is kept intact while a large environment R is given uniform local loss, and eigenstates of the resulting non-Hermitian Hamiltonian whose imaginary energies tend to zero as the system grows are declared to be BICs of the original Hermitian lattice. Applying this to a C4v- and chiral-symmetric lattice, they find four zero-energy corner states in the topological phase, argue that they are protected by the simultaneous presence of C4v and chiral symmetries, and show that breaking either symmetry transforms them into corner-localized resonances. The paper also argues that their BICs are not due to Hamiltonian separability and that the filling anomaly alone is insufficient for BIC protection.","tokens_in":14744,"tokens_out":15082,"duration_ms":158370,"significance":"If established, the result would be significant: it would extend the notion of higher-order topological corner states to systems without a bulk gap, broaden the design space for topological metamaterials, and offer a generic diagnostic for BICs in closed crystalline systems. The paper includes direct spatial-density evidence and a symmetry-based protection argument that goes beyond earlier separability-based BIC mechanisms. However, the central diagnostic is formally flawed, and the symmetry argument for two of the four claimed BICs is incomplete. The core claim is therefore not supported by the presented numerics, although the underlying idea may be salvageable with a different analysis.","major_comments":[{"comment":"The loss-probe criterion is not a valid test for BICs of the original Hermitian Hamiltonian. For H = H0 - iκ P_R, any right eigenvector ψ satisfies Im E = -κ ||P_R ψ||^2 / ||ψ||^2. Thus a vanishing imaginary part requires the eigenstate to have exactly zero support in the lossy environment R. For a fixed lossless region S of size ns and an exponentially localized state with penetration length ξ, the integrated weight in R is dominated by the region just outside S and is independent of the total lattice size n once n exceeds ξ; it is of order e^{-2ns/ξ}, a nonzero constant. The imaginary part should therefore saturate at a nonzero value as n grows, not decay exponentially. The decay shown in Fig. 2(b) and the exactly zero values for fixed ns=4 in Fig. S2(a) imply that the selected states have no support in R, i.e., they are either exactly compact within S or some parameter such as ns or κ is being varied without being reported. This contradicts the paper's stated exponential penetration of the BICs into the bulk. The presented numerics therefore do not establish that the four corner states are eigenstates of the original Hermitian H0 embedded in the continuum.","section":"Bound states in the continuum; Eq. (2); Fig. 2(b); Supplemental Fig. S2"},{"comment":"The protection argument for the two E-representation corner states is incomplete. The claim that the hybridized states |ψ1,2⟩ are merely arbitrary choices in the highly degenerate zero-energy subspace does not prove that the localized corner states are eigenstates. A Hermitian Hamiltonian diagonalizes within the degenerate subspace, and generically the true eigenstates are mixtures of corner and bulk components unless an additional symmetry or a vanishing coupling forces a localized eigenstate to exist. To establish that the E corner states are BICs, one must show that there is an eigenstate whose overlap with all bulk states is zero, for example by symmetry or by a direct inverse-participation-ratio calculation in the thermodynamic limit. The loss-probe was intended to supply this evidence, but as argued above it is invalid.","section":"Symmetry protection of the BICs"}],"minor_comments":[{"comment":"The loss term is defined with 0<κ≪1 in Eq. (2), but the captions of Figs. 2, 3, and 4 state κ=-5×10^{-2}; a negative κ would produce gain rather than loss, so the sign convention is inconsistent and should be corrected.","section":"Eq. (2) and figure captions"},{"comment":"The caption describes the horizontal axis as 'system size' without stating whether the varied parameter is the total lattice size n or the size ns of the lossless corner regions; this ambiguity is material to the claimed exponential decay and should be clarified.","section":"Fig. 2(b)"},{"comment":"The statement that indirect gap closings 'start to occur at t=0.5' appears to conflict with the earlier assertion that the topological transition is at |t|=1; please clarify which quantity is being described.","section":"BICs as a signature of the topological phase"},{"comment":"The insets in Fig. S2 lack labeled axes and scales, making it difficult to verify the claim that the BIC imaginary energies are exactly zero as a function of n; adding labels would improve the presentation.","section":"Supplemental Fig. S2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses a timely question and the model is plausible, but the main demonstration rests on a loss-probe criterion that is formally inconsistent with the stated exponential localization. I encourage the authors to re-analyze their data by directly computing eigenstates of the Hermitian H0 and reporting the inverse participation ratio as a function of system size, and to provide a rigorous symmetry proof for the E-representation corner states. The paper may be publishable after such a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere's my read of Benalcazar and Cerjan's BIC-in-HOTI paper. The concept is genuinely interesting: they argue that corner states can survive as exact bound states even when the bulk is gapless at the same energy, and they propose a symmetry protection mechanism (C4v + chiral) that goes beyond the known separability argument. That's a meaningful conceptual step, and the symmetry analysis is plausible. The paper also does a decent job situating itself relative to the acoustic HOTI work of Chen et al. (Ref. [57]).\n\nThe problem is in the diagnostic. The method is: add a uniform loss -iκ to the environment R (everything outside the four ns×ns corner squares), diagonalize H = H0 - iκ P_R, and pick states with Im E → 0 as the lattice size n grows. But there is an exact identity: for any eigenstate ψ of H, Im E = -κ ||P_R ψ||^2/||ψ||^2. So a vanishing imaginary part requires the eigenstate to have identically zero support in R. An exponentially localized BIC has a tail that penetrates into R, and for fixed ns that tail's weight in R is independent of n once n exceeds the localization length. So Im E should saturate at a non-zero value of order κ e^{-ns/ξ}, not decay to zero with increasing n. The observed decay in Fig. 2(b) is inconsistent with the stated exponential penetration. Either the selected states are exactly compact (and thus eigenstates of a decoupled cluster, not the infinite lattice), or ns is secretly varying. The paper doesn't resolve this. This is load-bearing: the loss probe is the main evidence that these are BICs rather than resonances.\n\nThe symmetry argument is a better foundation, but it's not a proof. It shows that the corner and bulk zero-energy states of the same E irrep cannot generically hybridize into split eigenstates while chiral symmetry holds, but it doesn't guarantee that exact corner-localized eigenstates of H0 exist. The numerics directly on H0 (e.g., spatial decay, participation ratio) would have settled the matter, but those are not presented. Also, there's a minor sign inconsistency: Eq. (2) says 0<κ<<1, but the figures use κ=-5×10^-2, which would be gain, not loss. And there's no code/data deposit.\n\nVerdict: I would not reject the idea. The BIC existence in this model may well be true—the symmetry argument is suggestive—but the current manuscript's main diagnostic is flawed. A serious referee should ask for a direct numerical demonstration on the Hermitian lattice, or a corrected complex-energy analysis that accounts for the identity above. This deserves peer review, but as a major-revision candidate, not acceptance.\n\nWho is this for? Researchers in topological photonics/metamaterials and HOTI theory. It's a think-piece that, if fixed, could change how people look for BICs in closed systems. As written, treat the central claim with caution.","headline":"Loss-probe diagnostic is mathematically unsound, so the central BIC claim is not established; the symmetry idea is still worth exploring.","tokens_in":15228,"tokens_out":13097,"would_cite":false,"duration_ms":127274,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A two-dimensional higher-order topological lattice hosts four zero-energy corner-bound states inside its gapless bulk continuum, protected by the simultaneous presence of C4v and chiral symmetry.","keywords":["bound states in the continuum","higher-order topological insulators","corner states","chiral symmetry","C4v symmetry","topological resonances","filling anomaly","non-Hermitian probing"],"falsifier":"Diagonalize the original Hermitian lattice for increasing $n$ and isolate the zero-energy subspace: if the four corner-localized states have bulk weight that decays only algebraically, or if the loss-probe imaginary energies scale as $\\kappa$ rather than exponentially with $n$, the BIC claim fails. A direct calculation that exhibits a basis of the degenerate zero-energy subspace in which the four corner states are not exact eigenstates would also falsify the paper's central assertion.","tokens_in":14215,"feed_emoji":"⚛️","tokens_out":7813,"duration_ms":67713,"temperature":0.7,"pith_summary":"The paper claims that a two-dimensional lattice with higher-order topology can pin four zero-energy states to its corners even though the bulk spectrum is gapless at that same energy, making these states genuine bound states in the continuum (BICs). The authors propose a practical diagnostic: add a small uniform loss to every site outside the corner regions, then identify BICs as eigenstates whose complex energy stays real as the system grows. Using this test on a $C_{4v}$- and chiral-symmetric four-site lattice, they find four corner BICs in the topological phase ($|t|<1$) and none in the trivial phase. They argue that the BICs are protected by the simultaneous presence of $C_{4v}$ and chiral symmetry, and that breaking either symmetry converts the BICs into corner-localized resonances. If correct, the result shows that topological boundary states can survive without spectral isolation, widening the search space for crystalline topological phases.","feed_headline":"Gapless topological lattice still pins four corner states","feed_subtitle":"A loss-probe experiment shows corner-bound states surviving in a gapless bulk, protected by two symmetries.","key_machinery":"The central mechanism is the loss-probe diagnostic: the lattice is split into four corner 'system' regions $S$ and the rest as the 'environment' $R$, and a uniform on-site loss $-i\\kappa$ is added to $R$ (Eq. 2). Bulk and edge states acquire negative imaginary energies of order $\\kappa$, whereas BIC wave functions, being exponentially confined to $S$, have imaginary energies that vanish exponentially with system size. The supporting symmetry mechanism is the representation analysis of the corner states ($A_1\\oplus B_2\\oplus E$) against the bulk zero-energy states (pure $E$), combined with chiral symmetry, which forces any $E$-type hybridization to remain at zero energy and therefore cannot produce a physical avoided crossing that would destroy the BICs.","core_discovery":"For the Bloch Hamiltonian $h(k)$ in Eq. (1), the paper's central claim is that the topological phase with dimerized hopping $|t|<1$ hosts four degenerate zero-energy corner-localized bound states in the continuum, embedded in the zero-energy continuum of the central bulk band, while the trivial phase $|t|>1$ does not. The four corner states are exact eigenstates in the thermodynamic limit whose penetration into the bulk decays exponentially, and they are unaffected by the loss-probe's environmental loss because they have no support in the lossy region. The symmetry argument is that under $C_{4v}$ the four corner states form the representations $A_1\\oplus B_2\\oplus E$, while all degenerate bulk states at zero energy transform as the two-dimensional $E$ representation; chiral symmetry pins every $E$-type state to zero energy, so a corner-bulk hybrid cannot move off zero energy and split. Thus the simultaneous preservation of $C_{4v}$ and chiral symmetry protects an identifiable BIC subspace, and breaking either symmetry turns the corner states into higher-order topological resonances.","pith_inferences":["Extending the symmetry logic, any chiral- and $C_n$-symmetric higher-order topological insulator whose corner states and bulk states carry incompatible irreps, or whose $E$-type partners are pinned to zero by chirality, should host corner BICs even with a gapless bulk; the paper does not test this general criterion.","The same loss-probe protocol could be applied to hinge modes of three-dimensional HOTIs or to other boundary-localized states, as long as the lossless region fully contains the exponentially confined mode — a straightforward but untested extension.","Because the loss probe maps onto gain-loss contrast, a photonic or acoustic metamaterial realization of this lattice should show four corner modes with near-zero linewidth while the surrounding bulk remains lossy; observing this would confirm the BIC character experimentally."],"forward_implications":["Corner bound states can exist without a bulk gap, so higher-order topological boundary states are not contingent on spectral isolation.","The BICs are protected by $C_{4v}$ and chiral symmetry together; preserving the symmetries that protect the topological phase but breaking either of these two converts them into higher-order topological resonances.","The loss-probe method provides a direct numerical way to identify BICs in closed crystalline lattices, complementing open-system radiative definitions of BICs.","BIC protection holds for nonseparable Hamiltonians, so the separability of $h(k_x,k_y)$ into $k_x$ and $k_y$ parts is not the origin of these BICs.","The corner filling anomaly is a necessary onset condition but not sufficient: additional symmetries decide whether the corner states emerge as BICs or as resonances."],"supporting_citations":[{"why":"Establishes quantized electric multipole insulators and the notion of boundary-localized corner states that this paper extends.","marker":"[5]"},{"why":"Supplies the corner-induced filling anomaly indices $Q^{(4)}$ and $Q^{(2)}$ used to identify which bands have the higher-order topological onset for the BICs.","marker":"[11]"},{"why":"Gives the foundational definition of bound states in the continuum as discrete eigenvalues embedded in a continuous spectrum.","marker":"[21]"},{"why":"Provides the environmental-design idea of adding loss to an environment to reveal BICs, which the present loss-probe method adapts to closed lattices.","marker":"[50]"},{"why":"Earlier numerical search for corner states as BICs in an acoustic second-order topological insulator, which the paper argues cannot distinguish BICs from resonances.","marker":"[57]"},{"why":"Introduces the separable-Hamiltonian mechanism for BICs that this paper demonstrates is not needed for its corner BICs.","marker":"[58]"},{"why":"Another instance of the separability route to BICs, used as the baseline the paper goes beyond.","marker":"[59]"},{"why":"Introduces the concrete four-site lattice model of Eq. (1) on which the whole BIC demonstration is built.","marker":"[60]"}],"fun_headline_variants":["Corner-bound states in continuum survive in gapless bulk","Symmetry-protected BICs at corners of topological insulator","Higher-order topology pins corner states inside bulk continuum","Gapless bulk fails to break corner BICs due to symmetry","Corner BICs resist hybridization in topological lattice"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that adding a small uniform loss to the environment and declaring 'BIC' any eigenstate whose imaginary energy vanishes as the lattice grows is a complete and valid way to certify bound states in the continuum of the original lossless lattice.","fun_headline_variants_meta":{"raw":{"variants":["Corner-bound states in continuum survive in gapless bulk","Symmetry-protected BICs at corners of topological insulator","Higher-order topology pins corner states inside bulk continuum","Gapless bulk fails to break corner BICs due to symmetry","Corner BICs resist hybridization in topological lattice"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000247,"raw_usage":{"total_tokens":1519,"prompt_tokens":900,"completion_tokens":619,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":516,"completion_tokens_details":{"reasoning_tokens":551}},"tokens_in":516,"tokens_out":619,"duration_ms":6680,"temperature":1.0,"reasoning_tokens":551,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:07:31.588406+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Diagonalize the original Hermitian lattice for increasing $n$ and isolate the zero-energy subspace: if the four corner-localized states have bulk weight that decays only algebraically, or if the loss-probe imaginary energies scale as $\\kappa$ rather than exponentially with $n$, the BIC claim fails. A direct calculation that exhibits a basis of the degenerate zero-energy subspace in which the four corner states are not exact eigenstates would also falsify the paper's central assertion.","supporting_citations":[{"cited_title":"Excess charges as a probe of one-dimensional topological crystalline insu- lating phases,","cited_arxiv_id":null,"evidence_quote":"Establishes quantized electric multipole insulators and the notion of boundary-localized corner states that this paper extends."},{"cited_title":"Braiding Majorana corner modes in a second-order topological superconductor","cited_arxiv_id":"1904.07822","evidence_quote":"Gives the foundational definition of bound states in the continuum as discrete eigenvalues embedded in a continuous spectrum."},{"cited_title":"Zero-Index Bound States in the Contin- uum,","cited_arxiv_id":null,"evidence_quote":"Provides the environmental-design idea of adding loss to an environment to reveal BICs, which the present loss-probe method adapts to closed lattices."},{"cited_title":"Topological protection of bound states against the hybridization,","cited_arxiv_id":null,"evidence_quote":"Earlier numerical search for corner states as BICs in an acoustic second-order topological insulator, which the paper argues cannot distinguish BICs from resonances."},{"cited_title":"Corner states in a second-order acoustic topological insulator as bound states in the con- tinuum,","cited_arxiv_id":null,"evidence_quote":"Introduces the separable-Hamiltonian mechanism for BICs that this paper demonstrates is not needed for its corner BICs."},{"cited_title":"A simple separable hamiltonian having bound states in the continuum,","cited_arxiv_id":null,"evidence_quote":"Another instance of the separability route to BICs, used as the baseline the paper goes beyond."},{"cited_title":"Resonances in quantum-dot transport,","cited_arxiv_id":null,"evidence_quote":"Introduces the concrete four-site lattice model of Eq. (1) on which the whole BIC demonstration is built."}],"review_version":1}