{"id":"076a8a9f-df1d-46cf-9132-67cc65301bb2","arxiv_id":"2506.14174","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A new local spectral gap condition and relative operator bounds guarantee a spectral gap in the spectral localizer, with an improved tapering constant.","lead":"This math paper strengthens the conditions under which the spectral localizer, a numerical tool for computing topological invariants, is guaranteed to work. It shows that only a local spectral gap and relative operator bounds are needed, and it improves a key constant, making the method more reliable for disordered and heterostructure materials.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central existence theorem is sound; the load-bearing soft spot is the numerical tapering constant CF≈4.56/CF≈2, which is only numerically/heuristically supported, not rigorously proven, yet it enters the advertised quantitative criterion (12).","rationale":"I checked the proof of Theorem 6 and found the algebra sound: the local gap condition, the commutator bound (11), and the first inequality in (12) together produce the lower bound μ≥b gρ, and the homotopy argument for box-independence of the half-signature is valid. The weak-locality assumption is explicit and not hidden; the 'easy criterion' after Definition 1 appears too permissive (for power-law hopping with decay (1+|n-m|)^{-(1+δ)}, the commutator [X,H] requires δ>d rather than δ>0), but it is not used in the main theorem and can be corrected without affecting the central claim. The one genuinely load-bearing soft spot is the numerical constant CF. The paper's headline improvement and the explicit criterion (2) rely on CF≈2, which is heuristic, and on CF≤4.56, which is supported only by a numerical evaluation of an integral. A rigorous theorem using CF=8 exists in the same appendix, so the qualitative result (local gap ⇒ gapped localizers at all larger volumes) is not threatened. However, the quantitative criterion that a practitioner would use is only as reliable as the certified value of CF. Since the reader already accepted with high confidence and the central claim stands, I do not change the verdict.","tokens_in":27196,"tokens_out":24899,"duration_ms":282750,"concrete_test":"Use interval arithmetic or a verified numerical integration (e.g., Arb or Taylor-model based quadrature) to prove a rigorous upper bound for I = 12/π ∫_{-∞}^{∞} |4 sin(p/2)-2p cos(p/2)| |sin(p/2)| / |p|^3 dp. If the certified bound is ≤ 4.56, the CF≈4.56 claim is validated for the k=1 family. Separately, for the nearest-neighbor discrete case, compute certified bounds on ρ‖[Fρ(X),H]‖/‖[X,H]‖ for increasing finite ρ (e.g., ρ=100, 200, 400) to test convergence to a constant ≤ 2; if the observed constant approaches a value > 2, the heuristic CF≈2 is wrong and the constants in Eq. (2) and Remark 7 must be revised upward.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 6 is an implication: if H is weakly local, has a ρ-local gap gρ, and κ satisfies (12) with a constant CF for which (11) holds, then the localizer gap is ≥ b gρ. The rigorous content is therefore only as strong as the proven value of CF. Appendix A proves (11) with CF=8 analytically, then claims CF≤4.56 by evaluating an explicit integral numerically ('Mathematica then gives'), and CF≈2 by heuristics/numerical tests; the latter value is used in the headline criterion (2). Since (12) and (2) are inequalities whose right-hand side shrinks as CF grows, an underestimated CF would make the advertised quantitative criterion invalid: a reader could choose κ satisfying (12) with CF=2 while the actual constant is larger, and the claimed μ≥b gρ would not be guaranteed. This does not falsify the qualitative theorem—one may always use the rigorously proven CF=8—but it leaves the paper's third advertised improvement, and the specific criterion (2), on non-rigorous footing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops three improvements to the spectral localizer framework for detecting local topological invariants. The main mathematical result, Theorem 6, states that if a weakly local Hamiltonian has a rho-local spectral gap g_rho and the tuning parameter kappa satisfies the two inequalities in (12), then all finite-volume spectral localizers of radius rho' >= rho have a localizer gap at least b g_rho, and the local Chern marker is independent of rho'. The proof uses a commutator/tapering estimate (11) whose constant C_F enters the criterion. The paper also introduces and studies the rho-local gap notion, proves a stability result for the localizer gap under distant perturbations (Proposition 12), establishes a spectral-flow stability result near phase boundaries (Proposition 13), and illustrates the theory with numerical simulations on the Haldane model and disordered variants. The central implication theorem is proven from the stated assumptions, but part of the advertised quantitative improvement—the constant C_F ≈ 4.56 and the headline criterion (2) with C_F ≈ 2—rests on numerical or heuristic evaluations rather than rigorous estimates.","tokens_in":27397,"tokens_out":11956,"duration_ms":125643,"significance":"If accepted in full, the paper would make a valuable contribution to the spectral localizer literature. The replacement of a global gap by a rho-local gap is a genuinely useful weakening of the hypothesis, and the relative operator bounds in (12) give a concrete quantitative form to the locality of the localizer. The arbitrary-shape stability statement in Proposition 12 and the spectral-flow stability in Proposition 13 are clean and go beyond prior work. The proofs of Theorem 6 and Proposition 12 are, apart from the constant issue discussed below, coherent and self-contained. I also credit the paper for being unusually explicit about the status of the numerical constant: Appendix A distinguishes the rigorously proven C_F = 8, the numerically evaluated C_F ≈ 4.56, and the heuristic C_F ≈ 2. However, the presentation of Eq. (2) in the introduction and the claimed factor-4.5 improvement in Remark 7 rely on the unproven heuristic value, so the advertised quantitative criterion is not yet on rigorous footing.","major_comments":[{"comment":"The claim that the tapering estimate (11) holds with C_F ≤ 4.56 is not proven rigorously. After the explicit integral representation (26), the paper states 'Mathematica then gives' the values 9.16, 4.56, 5.12, 5.75, with no numerical error control for the improper integral. This matters because C_F enters the denominator of the sufficient condition (12), so the quantitative content of Theorem 6 is only as strong as the proven value of C_F. The rigorous C_F = 8 bound from the earlier function keeps the theorem valid, but the advertised improvement of the tapering constant is conditional. I ask the authors to either supply a rigorous bound on the integral (for example, by interval arithmetic or by a convergent majorant) or to state explicitly in all theorem statements, remarks, and the abstract that the improved constants C_F ≈ 4.56 and C_F ≈ 2 are numerical/heuristic and not part of the proven results.","section":"Appendix A and Eq. (12)"},{"comment":"The criterion (2) is presented in the introduction as a main result, but it implicitly uses the heuristic value C_F ≈ 2. Comparing (2) with (17) and Remark 11 shows that (2) is obtained only after setting C_F = 2 and identifying ||[D(x),H](i 1 + 1/rho |X|)^-1|| with ||[X1+iX2,H](i 1 + 1/rho |X|)^-1||. If the actual constant is larger, a kappa chosen from (2) need not satisfy (12), and the lower bound mu >= b g_rho is not guaranteed. The exact relation between (2), (17), and the rigorous value of C_F should be spelled out; in particular, the introduction should not state (2) as a proven quantitative criterion unless the constant issue is resolved.","section":"Introduction, Eq. (2), and Remark 7"}],"minor_comments":[{"comment":"There are several typos that should be corrected: 'criterium' should be 'criterion' in the abstract and introduction, the affiliation contains stray spaces ('L aboratories'), and 'F AU Erlangen-N¨ urnberg' should be cleaned up.","section":"Throughout"},{"comment":"The formula for rho_c near the end of Section 6 contains a typo: 'CF ||[H||' should presumably be 'C_F ||[H,D]||' or similar; please correct the norm notation.","section":"Section 6"},{"comment":"In the proof of Proposition 14, the formula '⟨φ−|W|φ−⟩ − ⟨φ−|W|φ−⟩' should read '⟨φ+|W|φ+⟩ − ⟨φ−|W|φ−⟩'.","section":"Proposition 14 proof"},{"comment":"The sentence 'CF = 2 implies that 4/3 C_F = 8/3, notably an improvement by a factor 4.5' should be explicitly marked as relying on the heuristic value C_F ≈ 2, not on the proven estimate, to avoid misleading readers.","section":"Remark 7"}],"recommendation":"major_revision","confidential_remarks":"The central theorem is sound and the paper makes a useful contribution, so rejection is not warranted. The main issue is the use of non-rigorous constants in the advertised quantitative criteria; this is fixable either by supplying rigorous bounds or by transparently labeling the improved constants as numerical/heuristic. I would be satisfied with a revision that clearly separates the proven C_F = 8 results from the numerical/heuristic C_F ≈ 4.56 and C_F ≈ 2 claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main take: this is a real advance over prior spectral-localizer work. The new ρ-local gap (Definition 2) and the relative operator bounds in criterion (12) are not cosmetic relabelings; they change what the framework can guarantee for heterostructures and disordered systems. Theorem 6 is proven cleanly from its stated assumptions, and the homotopy argument for constancy of the local index is sound. The local-gap notion is a genuine contribution in itself, and Proposition 4 (a partial converse back to a global gap) is a nice touch.\n\nThe soft spot is real but narrow. Theorem 6 is an implication: if (12) holds for a constant CF for which the tapering estimate (11) holds, then the localizer gap is bounded below. Appendix A proves (11) rigorously with CF = 8. The claimed improvement to CF ≤ 4.56 rests on a Mathematica evaluation of an explicit integral with no rigorous error control, and the CF ≈ 2 used in the headline criterion (2) is explicitly heuristic and numerically motivated. So the paper's third advertised improvement is not fully rigorous. This matters for a reader who takes (2) literally with a small CF and picks a κ that would not satisfy (12) with the true constant. It does not break the central theorem, since one can always fall back to the proven CF = 8, and the paper is transparent about the numerical basis of the improvement. But the introduction and abstract frame the constant improvement as a settled result, which overstates the case.\n\nThe weak-locality assumption (Definition 1) is a genuine boundary of applicability, but it is stated clearly and is natural in this context. The numerics illustrate the claims well, though no code is shipped. Self-citation is heavy, but the new definitions and theorem are distinctly new, not a restatement of prior work.\n\nThis paper deserves a serious referee. I would recommend acceptance after the authors mark the rigorous status of CF more carefully—separating the proven CF = 8 theorem from the numerical/heuristic improvements, and reporting CF ≈ 2 as a conjecture.","headline":"A genuinely useful improvement to the spectral localizer criterion, with a solid central theorem and one soft spot: the advertised constant improvement is only numerically supported.","tokens_in":27929,"tokens_out":1794,"would_cite":true,"duration_ms":22473,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L80","47A53","81Q10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a positive local spectral gap is enough, under two explicit inequalities on a tuning parameter, to guarantee a protected spectral-localizer gap and a box-independent local Chern marker.","keywords":["spectral localizer","local spectral gap","local Chern marker","topological phase boundary","spectral flow","relative operator bounds","tapering estimate","disorder localization"],"falsifier":"Compute, for a concrete weakly local Hamiltonian whose parameters satisfy (12), the smallest singular value $\\mu_{\\kappa,\\rho'}(H,x)$ over boxes $\\rho'\\ge\\rho$; Theorem 6 predicts $\\mu_{\\kappa,\\rho'}(H,x)\\ge b\\,g_\\rho(H,x)$, so a numerical instance below this bound would refute the statement. A sharp adversarial test is a model with hopping decaying as $(1+|n-m|)^{-(1+\\delta)}$ for tiny $\\delta>0$ plus a strong impurity just outside $B_\\rho(x)$, checking whether the impurity's effect indeed decays with its distance as the resolvent bounds predict.","tokens_in":26981,"feed_emoji":"🧭","tokens_out":9645,"duration_ms":96070,"temperature":0.7,"pith_summary":"The paper's aim is to replace the global, often impractical hypotheses behind the spectral localizer by genuinely local ones. It introduces the $\\rho$-local gap $g_\\rho(H,x)$, the smallest eigenvalue of the Dirichlet restriction of $H^2$ to a box of length $\\rho$ around $x$, and proves in Theorem 6 that if this gap is positive and a tuning parameter $\\kappa$ obeys the explicit inequalities in (12), then the finite-volume spectral localizer has a gap of at least $b\\,g_\\rho(H,x)$ for every larger box, so the local Chern marker is independent of the box. The criterion uses only relative norms of $H$ and of $[D(x),H]$ with respect to the Dirac operator, which makes distant perturbations harmless and explains the observed robustness of local invariants in heterostructures and disordered systems. A careful reader would care because this turns the spectral localizer from a heuristic probe into a quantitatively checkable local certificate of topology.","feed_headline":"A local gap is enough to certify local topology","feed_subtitle":"Only a local gap and two inequalities now protect the local Chern marker.","key_machinery":"The load-bearing objects are the even spectral localizer $L_\\kappa(H,x)=-H\\Gamma+\\kappa D(x)$, the $\\rho$-local gap $g_\\rho(H,x)=\\inf\\mathrm{spec}\\big((H^2)_\\rho(x)\\big)$, and the tapering estimate $\\big\\|[F_\\rho(D(x)),H](\\imath 1+\\delta^{-1}D(x))^{-1}\\big\\|\\le \\frac{C_F}{\\rho}\\,\\big\\|[D(x),H](\\imath 1+\\delta^{-1}D(x))^{-1}\\big\\|$ for a smoothed indicator $F_\\rho$ of a $\\rho$-ball. The proof inserts the smoothed ball through $1_\\rho\\ge F_\\rho^2$, applies the local-gap inequality $F_\\rho H^2 F_\\rho\\ge g_\\rho^2F_\\rho^2$, and controls the commutator terms through the tapering estimate with relative resolvent damping. The resolvents of $D(x)$ are what make the criterion depend on $H$ only through relative norms, and the constant $C_F$, improved to $4.56$ with numerical support for roughly $2$, sets the quantitative scale of the admissible $\\kappa$ and hence the minimal system size needed for certification.","core_discovery":"The central discovery is that local topology can be certified locally: no global gap of the Hamiltonian is needed. For a weakly local Hamiltonian with $\\rho$-local gap $g_\\rho(H,x)>0$, the paper proves that two inequalities on $\\kappa$—the lower bound $2g_\\rho/\\rho<\\kappa$ and an upper bound built from $C_F\\|H R_\\kappa\\|+g_\\rho$ times $\\|[D(x),H]R_\\kappa\\|$, with $R_\\kappa=(\\imath 1+c\\,\\kappa\\,g_\\rho^{-1}D(x))^{-1}$ and constants $a,b,c$ satisfying $1-a-b^2>0$—force the squared localizer $L_{\\kappa,\\rho'}(H,x)^2$ to be at least $b^2g_\\rho^2\\,1_{\\rho'}(x)$ on every box $\\rho'\\ge\\rho$. Hence the localizer gap satisfies $\\mu_{\\kappa,\\rho'}(H,x)\\ge b\\,g_\\rho(H,x)$, and the half-signature $\\mathrm{Ch}_{\\kappa,\\rho'}(H,x)$ is constant under continuous changes inside the admissible region, and even for any enclosing set that contains $B_\\rho(x)$. The mechanism is that a smoothed indicator $F_\\rho(D(x))$ localizes the gap estimate, while resolvent factors suppress the Hamiltonian and its commutator with position far from $x$; a separate estimate improves the tapering constant from $C_F=8$ to $C_F\\le 4.56$, with numerical evidence for $C_F\\approx 2$ in short-range models.","pith_inferences":["Editorial inference: the ratio $g_\\rho(H,x)/\\mu_{\\kappa,\\rho}(H,x)$ could serve as a spatially resolved confidence map, because regions where the local gap closes are exactly where the ratio diverges; scanning $x$ at fixed $\\kappa,\\rho$ would locate topological phase boundaries directly from numerical or experimental localizer data.","Editorial inference: because Theorem 6 needs only the finite-volume quantity $g_\\rho(H,x)$, it suggests an adaptive protocol in which one estimates the local gap from measured local spectra and then chooses $\\kappa$ to satisfy (12), rather than assuming a known global gap.","Editorial inference: a rigorous reduction of the tapering constant to $C_F\\approx 2$ would make the sufficient condition nearly tight, aligning the predicted minimal volumes with the sizes at which local Chern markers are already observed to stabilize.","Editorial inference: the same proof scheme should extend recognisably to odd and real versions of the localizer, giving local-gap validity criteria for $\\mathbb{Z}_2$ and spin-Chern invariants; the paper indicates the even case but leaves those extensions implicit."],"forward_implications":["A positive $\\rho$-local gap at one point, together with the explicit inequalities (12), guarantees that the spectral localizer stays gapped in every larger enclosing region, with $\\mu_{\\kappa,\\rho'}(H,x)\\ge b\\,g_\\rho(H,x)$.","The local Chern marker $\\mathrm{Ch}_{\\kappa,\\rho'}(H,x)$ is box-independent: it is the same for all $\\rho'\\ge\\rho$ and for any finite set containing $B_\\rho(x)\\cap\\mathbb{Z}^d$.","Perturbations supported far from $x$ cannot close the localizer gap; their effect enters the bounds divided by a power of the distance from their support to $B_\\rho(x)$, so the local index is genuinely local.","The spectral flow of the localizer along a path crossing a topological phase boundary is stable under weakly local perturbations whose support avoids the two endpoints, even if the perturbation cuts across the path.","The improved tapering constant makes the criterion quantitatively realistic: with $C_F\\approx 2$, the bounds predict that a few tens of unit cells per direction suffice to certify a stable local gap."],"supporting_citations":[{"why":"introduces the even spectral localizer and proves the half-signature equals the bulk Chern number; Theorem 6 refines its validity criterion.","marker":"[22]"},{"why":"provides the prior global-gap criterion and the simplified spectral-flow proof whose hypotheses Theorem 6 relaxes.","marker":"[11]"},{"why":"introduces the spectral localizer as a finite-volume tool for computing K-theory invariants.","marker":"[21]"},{"why":"supplies the original tapering-estimate argument that Appendix A improves to the constant $C_F=4.56$.","marker":"[4]"},{"why":"supplies the resolvent perturbation theory used in the two-parameter expansion for eigenvalue-slope stability in Section 8.","marker":"[18]"},{"why":"provides the honeycomb-lattice Chern-insulator model used as the numerical testbed for local gaps, $\\kappa$ bounds, and local Chern markers.","marker":"[15]"},{"why":"reported that the spectral localizer remains gapped in a disorder-localized random superconductor, which motivates the local-gap analysis of Section 6.","marker":"[23]"}],"fun_headline_variants":["Local gap suffices to certify local topology","No global gap needed: local topology from local gap","Spectral localizer: relative bounds tighten local gap proof","Improved tapering constant sharpens local topology bounds","Local gap alone pins down the local Chern marker"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is weak locality of $H$: the Hamiltonian must keep the domain of the Dirac operator $D(x)$ invariant and make $[D(x),H]$ a bounded operator; if hopping is so long-range that this commutator is unbounded, the theorem's proof and conclusion do not apply even when a local spectral gap exists.","fun_headline_variants_meta":{"raw":{"variants":["Local gap suffices to certify local topology","No global gap needed: local topology from local gap","Spectral localizer: relative bounds tighten local gap proof","Improved tapering constant sharpens local topology bounds","Local gap alone pins down the local Chern marker"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000212,"raw_usage":{"total_tokens":1435,"prompt_tokens":976,"completion_tokens":459,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":592,"completion_tokens_details":{"reasoning_tokens":386}},"tokens_in":592,"tokens_out":459,"duration_ms":5021,"temperature":1.0,"reasoning_tokens":386,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:18:38.191875+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a concrete weakly local Hamiltonian whose parameters satisfy (12), the smallest singular value $\\mu_{\\kappa,\\rho'}(H,x)$ over boxes $\\rho'\\ge\\rho$; Theorem 6 predicts $\\mu_{\\kappa,\\rho'}(H,x)\\ge b\\,g_\\rho(H,x)$, so a numerical instance below this bound would refute the statement. A sharp adversarial test is a model with hopping decaying as $(1+|n-m|)^{-(1+\\delta)}$ for tiny $\\delta>0$ plus a strong impurity just outside $B_\\rho(x)$, checking whether the impurity's effect indeed decays with its distance as the resolvent bounds predict.","supporting_citations":[{"cited_title":"Loring, H","cited_arxiv_id":null,"evidence_quote":"introduces the even spectral localizer and proves the half-signature equals the bulk Chern number; Theorem 6 refines its validity criterion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the prior global-gap criterion and the simplified spectral-flow proof whose hypotheses Theorem 6 relaxes."},{"cited_title":"Loring, H","cited_arxiv_id":null,"evidence_quote":"introduces the spectral localizer as a finite-volume tool for computing K-theory invariants."},{"cited_title":"Bratteli, D","cited_arxiv_id":null,"evidence_quote":"supplies the original tapering-estimate argument that Appendix A improves to the constant $C_F=4.56$."},{"cited_title":"Kato, Perturbation theory of linear operators , 2nd edition, (Springer, Berlin, 2012)","cited_arxiv_id":null,"evidence_quote":"supplies the resolvent perturbation theory used in the two-parameter expansion for eigenvalue-slope stability in Section 8."},{"cited_title":"Parity Anomaly","cited_arxiv_id":null,"evidence_quote":"provides the honeycomb-lattice Chern-insulator model used as the numerical testbed for local gaps, $\\kappa$ bounds, and local Chern markers."},{"cited_title":"Lozano Viesca, J","cited_arxiv_id":null,"evidence_quote":"reported that the spectral localizer remains gapped in a disorder-localized random superconductor, which motivates the local-gap analysis of Section 6."}],"review_version":1}