{"id":"eab3648f-8bd3-4d0c-ab56-27606fb59383","arxiv_id":"1908.01756","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On a closed symplectic manifold, the low-lying eigenvalues of the Bochner Laplacian on L^p are pτ0 + μj + O(p^{-1/2}), where μj are eigenvalues of model Toeplitz operators at the wells.","lead":"This paper proves that low-energy eigenvalues of a magnetic Schrödinger-type operator on a curved space, built from high powers of a line bundle, are asymptotically described by eigenvalues of model oscillators at the points where the magnetic field is weakest. A generalist might read it because it gives a precise semiclassical expansion for a natural family of spectral problems and ties together geometry, analysis, and Toeplitz operator theory.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof's lower bound rests on the author's unpublished [13, Thm 1.5]; the displayed two-term formula (3.25) alone would not justify the full expansion (1.9), so this external input is the weakest load-bearing step.","rationale":"The paper is well organized and the internal microlocal computations (Agmon estimates, reduction to the lowest Landau level, Toeplitz symbol matching) are careful. The only point at which the proof of Theorem 1.1 depends on an input not proved here is the asymptotic expansion for generalized Toeplitz operators with discrete wells, imported from the author's preprint [13]. The reader's verdict already identified this as the weakest assumption. I agree with that identification; no independent internal gap emerged. The extra nuance about (1.9) is that the quoted form (3.25) is only a two-term expansion, so the paper should either cite the all-orders version explicitly or state it. This does not change the accept recommendation: the concern is a verification item, not an observed error, and the core result (1.8) is plausibly well-supported conditional on [13].","tokens_in":18281,"tokens_out":20302,"duration_ms":179811,"concrete_test":"Obtain the full statement and proof of [13, Theorem 1.5] and check three items: (a) it covers generalized Toeplitz operators T_p = P_Hp(Σ_{l≥0} p^{-l}g_l)P_Hp with P_Hp the low-energy projection of Δ_p, as used in Section 3.3; (b) when the model eigenvalue μ_j is simple, it supplies an asymptotic expansion to all orders in p^{-1/2} (not just the two terms of (3.25)); (c) all its hypotheses hold for D_p with h=τ-τ0 and g1=J_{1,2}. If (a)-(c) all hold, the concern is resolved; if any fails, the lower bound (3.30) and/or the complete expansion (1.9) remain unproven.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.3 obtains the decisive lower bound (3.30) by applying [13, Theorem 1.5] to the generalized Toeplitz operator D_p = p^{-1}P_Hp Δ_p P_Hp + P_Hp(τ-τ0)P_Hp. That theorem is cited without proof, and it is not self-contained in the submitted manuscript: the only displayed statement from it, (3.25), gives λ_m^p = p^{-1}μ_m + p^{-3/2}φ_m + O(p^{-2}), i.e., just two terms. The paper then claims the complete expansion (1.9) for simple μ_j with 'The expansion (1.9) is proved similarly', without stating an all-orders version of the imported theorem. If [13] has an unstated hypothesis (for instance, a restriction on the subprincipal symbol or on the class of Bergman projections) or proves only (3.25), then (3.30) and especially (1.9) are not established by this paper. This is a load-bearing reliance on an external preprint, not a demonstrated flaw in the argument itself.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the Bochner Laplacian Δ^{L^p} acting on sections of the p-th tensor power of a positive Hermitian line bundle over a closed symplectic manifold, equivalently a semiclassical magnetic Schrödinger operator with non-degenerate magnetic field. Under Assumption 1, which says that the minima of the magnetic intensity τ are non-degenerate and hence form a finite set of discrete wells, the paper claims in Theorem 1.1 that the low-lying eigenvalues satisfy λ_j(Δ^{L^p}) = p τ_0 + μ_j + O(p^{-1/2}) as p → ∞, where μ_j are the eigenvalues of a model operator D obtained as a direct sum of Toeplitz operators D_{x_i} associated with the wells; moreover, if μ_j is a simple eigenvalue of D, then λ_j(Δ^{L^p}) has a complete asymptotic expansion in powers of p^{-1/2}. The upper bounds are proved by an explicit quasimode construction in §2, using rescaled normal coordinates and a formal eigenfunction expansion for the rescaled operator. The lower bounds in §3 combine Agmon-type estimates with a localization argument and a reduction to the lowest Landau level, followed by an application of the author's theorem [13] on eigenvalue asymptotics for Toeplitz operators with discrete wells.","tokens_in":18421,"tokens_out":7226,"duration_ms":70490,"significance":"If the result is correct, it provides a natural and fairly general extension to arbitrary dimension and to a symplectic rather than Kähler setting of earlier two-dimensional magnetic-well asymptotics, and it gives a complete asymptotic expansion along a simple eigenvalue branch. The paper's strengths are its explicit construction of approximate eigenfunctions with remainder O(p^{-1/2}) and with arbitrary-order formal expansions for simple eigenvalues, the clean statement of the model operator in terms of Bergman kernels, and the absence of any fitted parameters. The proof is internally consistent and the strategy is coherent: upper bounds come from quasimodes, and lower bounds from Agmon localization, reduction to a Toeplitz operator, and an imported Toeplitz spectral theorem. The main fragility is that the decisive lower-bound and full-expansion steps rest on the author's own unpublished preprint [13], whose needed statements and hypotheses are only partially reproduced in the manuscript.","major_comments":[{"comment":"The lower-bound proof applies [13, Theorem 1.5] to the Toeplitz operator D_p, but the only statement of that theorem given in the paper is (3.25), which asserts just the two-term expansion λ_m^p = p^{-1} μ_m + p^{-3/2} φ_m + O(p^{-2}). This is sufficient for the O(p^{-1/2}) bound in (3.30), but it does not support the complete expansion (1.9) for simple eigenvalues. The sentence 'The expansion (1.9) is proved similarly' does not state an all-orders version of the imported theorem, nor its hypotheses. Since (1.9) is a central claim of the paper, the author should either state and prove the required all-orders Toeplitz eigenvalue expansion, provide a precise reference to a published version, or restrict (1.9) to the two-term statement that is actually established here.","section":"§3.3, Eqs. (3.25)–(3.30) and Theorem 1.1(1.9)"},{"comment":"The manuscript verifies that the principal symbol of D_p is h = τ - τ_0 and that the first two terms of its Toeplitz symbol expansion match the model operator D_{x_0}, via (3.28)–(3.29). It does not, however, spell out the remaining hypotheses of [13, Theorem 1.5] (for example any conditions on the subprincipal symbol, on the class of Bergman projections, or on uniformity of the symbol expansion) and it does not check them explicitly for D_p. Because [13] is a preprint and is not reproduced, the referee cannot certify that the application is legitimate. The author should list the hypotheses of the imported theorem and verify each one for the operator D_p, or include the full theorem statement in an appendix.","section":"§3.3, application of [13, Theorem 1.5]"}],"minor_comments":[{"comment":"The abstract contains typographical errors such as 'd iscrete' and 'eigenv alue asymptotics'; these should be corrected.","section":"Title and abstract"},{"comment":"The notation for indices is inconsistent: φ^p_{j2}, φ^p_{jN}, C_{j,2}, C_{j,N}, and p_{j,2} are used without a uniform convention. Please use commas consistently, e.g. φ^p_{j,2} and φ^p_{j,N}.","section":"§2, Theorem 2.1"},{"comment":"The infimum in the definition of μ_0 is written as inf over u ∈ T_x X, x ∈ X but the vector u = 0 should be excluded; please make the domain explicit.","section":"Eq. (3.12)"},{"comment":"The asymptotic notation '∼=' and the norm 'C_l(X)' are used without definition; please define the relevant seminorms or specify the exact meaning of the estimate.","section":"Eq. (3.26)"},{"comment":"In the formula for the eigenvalues in the case n = 1, the term 'A2' should presumably read 'A^2'; please correct the typography.","section":"Remark 1.3"},{"comment":"Reference [13] is cited as a preprint; if it has been published or revised, the reference should be updated with the final publication data.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The decisive lower-bound and full-expansion arguments depend on the author's own unpublished preprint [13], and the manuscript reproduces only the two-term version (3.25) of the needed theorem. This is not a novelty concern, but it makes verification of the central claim contingent on an external document that the referee cannot inspect. I would suggest that the editor require either the full statement and proof of the all-orders Toeplitz eigenvalue expansion, or a published reference for it, before final acceptance. The rest of the paper is careful and the upper-bound construction is particularly clear."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper proves what the title promises: for the Bochner Laplacian on high tensor powers of a positive line bundle over a closed symplectic manifold, with discrete wells, the low-lying eigenvalues are pτ0 + μ_j + O(p^{-1/2}), with a full asymptotic expansion when μ_j is simple. That result is new in this generality. Previous work covered the 2D case and the trivial-bundle case; here the setting is an arbitrary positive line bundle on a symplectic manifold, and the model operator is a direct sum of Toeplitz operators on harmonic-oscillator kernels at the wells. That is a real step forward, not a repackaging.\n\nThe proof is thorough and mostly self-contained. The upper bounds are explicit quasi-modes with a careful iterative construction and honest remainders. The lower bounds use Agmon estimates and a reduction to the lowest Landau level, and the localization argument is written in enough detail to check. The paper also correctly credits prior work: the Bergman kernel machinery from Ma–Marinescu and the author's own Toeplitz framework are cited precisely.\n\nThe soft spots are real but not fatal. The decisive lower bound comes from applying the author's Theorem 1.5 in the unpublished preprint [13]. The paper states only the two-term form (3.25) of that theorem, and then claims the complete expansion (1.9) with “proved similarly.” So the full expansion rests on an all-orders version of [13] that is never stated. That is a gap in self-containedness, though not a demonstrated flaw; the ingredient is plausible and the author is presumably in a position to justify it. A referee should ask for the precise statement of [13, Thm 1.5] or an appendix that closes this.\n\nA smaller note: the paper mentions that the lower-bound localization under Assumption 2 is deferred to a forthcoming paper. That is a side claim and does not affect Theorem 1.1, which is the core result. There are also a few typos and minor slips, but nothing that obscures the argument.\n\nWho is this for? People working in semiclassical analysis or geometric spectral theory who want a clean statement in the positive-line-bundle setting. It deserves a serious referee and, after addressing the [13] dependence, publication. I would take it if I were editing the journal.\n\nRecommendation: send to peer review.","headline":"A careful, general proof of the expected semiclassical eigenvalue expansion for the Bochner Laplacian, whose main soft spot is a load-bearing reliance on the author's own unpublished Toeplitz theorem.","tokens_in":19041,"tokens_out":1313,"would_cite":true,"duration_ms":14958,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["58J50","53D50","58J37"],"pacs":[],"model":"deepseek-v4-flash","headline":"Eigenvalues of Bochner Laplacians are governed by local well models up to $O(p^{-1/2})$.","keywords":["Bochner Laplacian","semiclassical eigenvalue asymptotics","symplectic manifold","magnetic Schrödinger operator","discrete wells","Toeplitz operators","Bergman kernel","lowest Landau level"],"falsifier":"On a flat two-torus with a periodic magnetic field whose intensity $\\tau$ has two non-degenerate minima, compute the first few eigenvalues of $\\Delta^{L^p}$ for large $p$ and compare each with $p\\tau_0+\\mu_j$; if the difference does not tend to zero at the rate $O(p^{-1/2})$, or if the next coefficient in the expansion (1.9) disagrees with a direct calculation, Theorem 1.1 is false. A sharper check is to apply the imported Toeplitz theorem to the symbol $h=\\tau-\\tau_0$ with subprincipal symbol $J_{1,2}$ and verify that its predicted model eigenvalues $\\mu_j$ are the ones seen numerically.","tokens_in":17989,"feed_emoji":"🧲","tokens_out":15436,"duration_ms":141195,"temperature":0.7,"pith_summary":"This paper tries to establish that, on a closed even-dimensional manifold carrying a symplectic form, the low-lying eigenvalues of the Bochner Laplacian on the $p$-th tensor power of a positive line bundle are determined, up to an error of order $p^{-1/2}$, by finitely many local model operators attached to the minima, or wells, of the magnetic intensity function $\\tau$. With $\\tau_0=\\min\\tau$, the claim is that the $j$-th eigenvalue satisfies $\\lambda_j(\\Delta^{L^p})=p\\tau_0+\\mu_j+O(p^{-1/2})$, where $\\mu_j$ is the $j$-th eigenvalue of a direct sum of model operators at the wells, and that a complete asymptotic expansion in powers $p^{-k/2}$ exists when $\\mu_j$ is simple. A sympathetic reader would care because this turns a curved, global spectral problem into local algebraic data at finitely many points and extends to arbitrary even dimension the magnetic-well expansions previously known in dimension two. In the trivial-bundle case, the same statement covers the semiclassical magnetic Schrödinger operator with non-degenerate magnetic field and discrete wells.","feed_headline":"Low-lying Bochner spectrum is a well model plus a square-root error","feed_subtitle":"Each eigenvalue is p times the minimum magnetic intensity plus a model-well value, up to a square-root error.","key_machinery":"The central object is the model operator at a well, $D_{x_0}=P_{x_0}(Q_{x_0}(Z)+J_{1,2}(x_0))$, acting on the kernel of the harmonic-oscillator-type operator $L_{x_0}$ on the tangent space at $x_0$. Here $P_{x_0}$ is the Bergman projection onto that kernel, $Q_{x_0}$ is the quadratic form from the Hessian of $\\tau$, and the scalar $J_{1,2}(x_0)$ is the value at the well of the first subprincipal coefficient in the near-diagonal expansion of the generalized Bergman kernel; this term enters because the effective second-order operator on low-energy sections is $F_{1,2}+P Q_{x_0}$, with $F_{1,2}$ a scalar operator. The argument forces $D_{x_0}$ into view by rescaling near the well with $t=p^{-1/2}$, expanding the rescaled Bochner Laplacian as $L_{x_0}+\\tau_0+tO_1+t^2(O_2+Q_{x_0})+\\cdots$, and using $PO_1P=0$ so the linear term cannot shift the ground energy. The lower-bound half is carried by exponential localization and the reduction of the full problem to the Toeplitz operator $P_{H_p}\\Delta^{L^p}P_{H_p}$ on the lowest Landau level.","core_discovery":"Under Assumption 1, which says that each minimum of the magnetic intensity $\\tau$ is non-degenerate, the paper proves that the low-lying spectrum of $\\Delta^{L^p}$ is asymptotically governed by the finite direct sum $D=D_{x_1}\\oplus\\cdots\\oplus D_{x_N}$ of model operators. Theorem 1.1 states that for every fixed $j$, $\\lambda_j(\\Delta^{L^p})=p\\tau_0+\\mu_j+O(p^{-1/2})$ as $p\\to\\infty$, and that when $\\mu_j$ is a simple eigenvalue of $D$, the eigenvalue has the complete expansion $\\lambda_j=p\\tau_0+\\mu_j+\\sum_{k=1}^{\\infty}a_{k,j}p^{-k/2}$. The proof has an upper half, which constructs approximate eigenfunctions concentrated in $p^{-1/2}$-neighborhoods of the wells, and a lower half, which combines exponential localization estimates, reduction to the lowest Landau level through the generalized Bergman projection, and the Toeplitz-operator description of the renormalized Bochner Laplacian.","pith_inferences":["When $\\mu_j$ is multiple, the natural extension is a matrix-valued effective Hamiltonian built from the blocks $D_{x_i}$; individual eigenvalues would expand only after diagonalizing that finite matrix, a step the paper does not perform.","The same reduction suggests that a theorem for non-degenerate submanifold wells would replace the point model by an operator on the normal bundle of the well, with $J_{1,2}$ as the leading subprincipal correction; the paper leaves that setting to a later proof.","Because the model operator is explicitly solvable in the one-well example, a numerical diagonalization on a two-dimensional torus with two wells could test whether consecutive low eigenvalues split by the predicted constant as $p$ grows."],"forward_implications":["For every fixed $j$, the entire low-energy cluster of $\\Delta^{L^p}$ is asymptotically given by $p\\tau_0+\\mu_j+O(p^{-1/2})$, so the manifold enters the leading asymptotics only through the Hessian of $\\tau$ and the Bergman-kernel coefficient $J_{1,2}$ at the wells.","When $\\mu_j$ is simple, the eigenvalue has a full asymptotic series in powers $p^{-k/2}$, so no logarithmic or other non-power terms appear.","The upper-bound construction yields approximate eigenfunctions supported in a $p^{-1/2}$-neighborhood of a single well, so each low-lying eigensection is effectively concentrated at one minimum of $\\tau$.","In the trivial-bundle case, the theorem applies verbatim to the semiclassical magnetic Schrödinger operator with Planck constant $\\hbar=1/p$ and non-degenerate magnetic field with discrete wells."],"supporting_citations":[{"why":"Supplies the theorem on low-lying eigenvalues of Toeplitz operators with discrete wells that produces the lower bound (3.25) and hence (3.30).","marker":"[13]"},{"why":"Provides the rescaled Bochner Laplacian expansion, the model operator $L_{x_0}$, and the coefficient $F_{1,2}$ used to identify $D_{x_0}$.","marker":"[16]"},{"why":"Proves that $P_{H_p}\\Delta_p P_{H_p}$ is a Toeplitz operator and gives the near-diagonal expansion of its kernel whose leading coefficient is $J_{1,2}$.","marker":"[12]"},{"why":"Furnishes the quantization framework for eigenstates of the Bochner Laplacian on symplectic manifolds, underwriting the Toeplitz algebra step.","marker":"[11]"},{"why":"Introduces the renormalized Bochner Laplacian $\\Delta_p=\\Delta^{L^p}-p\\tau$ and the space $H_p$ of low-lying eigensections used in the reduction.","marker":"[4]"},{"why":"Provides the universal lower bound (3.1) from the Weitzenböck formula that starts the exponential localization estimates.","marker":"[15]"},{"why":"Gives the exponential off-diagonal decay of the generalized Bergman kernel used to control the commutator term in Proposition 3.3.","marker":"[14]"}],"fun_headline_variants":["Discrete wells dictate low-lying Bochner spectrum","Magnetic minima set eigenvalues, square-root error","Semiclassical eigenvalues: wells plus model terms","Bochner eigenvalues: well model plus square-root error"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The lower-bound half imports, without proof, a theorem on the low eigenvalues of Toeplitz operators with several discrete wells; if that theorem carries an unstated condition or has a gap, the estimate $\\lambda_j(\\Delta^{L^p})\\ge p\\tau_0+\\mu_j-Cp^{-1/2+\\delta}$ is not secured.","fun_headline_variants_meta":{"raw":{"variants":["Discrete wells dictate low-lying Bochner spectrum","Magnetic minima set eigenvalues, square-root error","Semiclassical eigenvalues: wells plus model terms","Bochner eigenvalues: well model plus square-root error"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000471,"raw_usage":{"total_tokens":2290,"prompt_tokens":838,"completion_tokens":1452,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":454,"completion_tokens_details":{"reasoning_tokens":1390}},"tokens_in":454,"tokens_out":1452,"duration_ms":11353,"temperature":1.0,"reasoning_tokens":1390,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:04:37.515125+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a flat two-torus with a periodic magnetic field whose intensity $\\tau$ has two non-degenerate minima, compute the first few eigenvalues of $\\Delta^{L^p}$ for large $p$ and compare each with $p\\tau_0+\\mu_j$; if the difference does not tend to zero at the rate $O(p^{-1/2})$, or if the next coefficient in the expansion (1.9) disagrees with a direct calculation, Theorem 1.1 is false. A sharper check is to apply the imported Toeplitz theorem to the symbol $h=\\tau-\\tau_0$ with subprincipal symbol $J_{1,2}$ and verify that its predicted model eigenvalues $\\mu_j$ are the ones seen numerically.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the theorem on low-lying eigenvalues of Toeplitz operators with discrete wells that produces the lower bound (3.25) and hence (3.30)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the rescaled Bochner Laplacian expansion, the model operator $L_{x_0}$, and the coefficient $F_{1,2}$ used to identify $D_{x_0}$."},{"cited_title":"On asymptotic expansions of generalized Bergman kernels on symplectic manifolds","cited_arxiv_id":"1703.04107","evidence_quote":"Proves that $P_{H_p}\\Delta_p P_{H_p}$ is a Toeplitz operator and gives the near-diagonal expansion of its kernel whose leading coefficient is $J_{1,2}$."},{"cited_title":"Berezin-Toeplitz quantization for eigenstates of the Bochner-Laplacian on symplectic manifolds","cited_arxiv_id":"1703.06420","evidence_quote":"Furnishes the quantization framework for eigenstates of the Bochner Laplacian on symplectic manifolds, underwriting the Toeplitz algebra step."},{"cited_title":"Guillemin, A","cited_arxiv_id":null,"evidence_quote":"Introduces the renormalized Bochner Laplacian $\\Delta_p=\\Delta^{L^p}-p\\tau$ and the space $H_p$ of low-lying eigensections used in the reduction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the universal lower bound (3.1) from the Weitzenböck formula that starts the exponential localization estimates."}],"review_version":1}