{"id":"85d9b6ec-3e75-40ad-bf71-c826168c923f","arxiv_id":"2502.01291","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Rational rectangles and the three integrable triangles have strong inverse localization, while generic irrational rectangles and (claimed) generic ellipses fail it strongly.","lead":"This paper proves that high-energy eigenfunctions on some integrable billiards can locally approximate any wave pattern, while on most other integrable billiards they cannot. The results support the expectation that generic integrable quantum systems behave very differently from random waves.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem D's ellipse proof uses a nonlinear-coordinate identity that fails; the algebraic-variety argument for generic ellipses is unsupported as written.","rationale":"The reader's weakest-assumption pinpoints exactly the step I also find load-bearing: the identification in Section 5.3.2 of a Euclidean Planck-scale localization with a shift of the elliptic coordinate. Because the Cartesian-to-elliptic map is nonlinear, the point z0+F(I_{ξ0}(ξ),η0)/√λ does not have elliptic coordinates (ξ0+I_{ξ0}(ξ)/√λ, η0). The explicit major-axis computation in the concrete test demonstrates the mismatch already at first order in λ^{-1/2}. This invalidates the claim that the localized eigenfunction restricted to L_{z0} is a modified Mathieu function of the shifted argument, which is the sole mechanism producing the algebraic variety membership Q_M=0 and hence the codimension argument. Without this, Theorem D's negative result for generic ellipses is unsupported. I agree with the reader that the positive results for rational rectangles and integrable triangles, and the negative results for irrational rectangles and balls, are independent and may be salvageable; however, the paper's advertised conclusion that generic integrable billiards, including ellipses, do not match random waves depends critically on Theorem D. The false coordinate identity is an internal inconsistency in a central proof, not merely a disagreement with consensus, so a REJECT verdict is appropriate. If the authors can repair the ellipse argument, for example by replacing the exact Mathieu-product identity with a controlled asymptotic expansion and showing the algebraic-variety obstruction survives the error, a conditional acceptance could be reconsidered.","tokens_in":78617,"tokens_out":8032,"duration_ms":81427,"concrete_test":"Take z0 on the major axis with ξ0=1, choose any b with c=√(1-b^2)>0, and use the paper's I_{ξ0}(ξ)=ξ0(1-R)+ξ_b ξ. For ξ=0, compare the claimed coordinate ξ0+I_{ξ0}(0)/√λ with the actual elliptic coordinate of z0+F(I_{ξ0}(0),0)/√λ. For λ=100 and R=0.1, the claimed value is 1+0.9/10=1.09, while the actual coordinate is arccosh(cosh 1 + cosh(0.9)/10) ≈ 1.1219. More fundamentally, expanding in λ^{-1/2}, the actual shift of the elliptic coordinate is (cosh I_{ξ0}(0))/(sinh 1)·λ^{-1/2}, not I_{ξ0}(0)·λ^{-1/2}; these agree only if cosh I = I sinh ξ0. Recomputing Step II with the correct pushforward of the Cartesian localization to elliptic coordinates will show that the evaluation vector EM_M(EM of the localized restriction) is not a Mathieu-product sample, so the algebraic condition Q_M=0 no longer follows.","verdict_should_be":"REJECT","load_bearing_attack":"Section 5.3.2 defines L_{z0} as the coordinate curve {F(ξ,η0)} in Cartesian coordinates and asserts that for a localized eigenfunction, its restriction satisfies \\tilde R(\\tilde u)(ξ)=u_{mn}(ξ0+I_{ξ0}(ξ)/√λ, η0). This requires the equality F(ξ0+I_{ξ0}(ξ)/√λ, η0)=F(ξ0,η0)+F(I_{ξ0}(ξ),η0)/√λ. Since F(ξ,η)=(c cosh ξ cos η, c sinh ξ sin η) is not affine, this identity is false. On the major axis η0=0, the actual elliptic coordinate of z0+F(I,0)/√λ is arccosh(cosh ξ0+cosh I/√λ), whose first-order shift is (cosh I)/(sinh ξ0)·λ^{-1/2}, not I·λ^{-1/2}. Consequently the sampled values are not S_m^p((t+ξ)/s, q)E_n^p(η0,q), the claimed modified-Mathieu ODE need not hold for the composed function, and the inclusion EM_M(\\tilde R(\\tilde u^{z0}_{mn})) ∈ Q_M^{-1}({0}) does not follow. The proof of Theorem D, and therefore the assertion that generic ellipses fail inverse localization, rests on this inclusion. The rectangle, triangle, and ball results are separate and may remain valid, but the paper's central claim about smooth integrable billiards is not established as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the inverse localization property for high-energy Laplace eigenfunctions in integrable planar billiards: whether, around typical points and at Planck scale, eigenfunctions can approximate any given solution to the Helmholtz equation. The positive results show that rational rectangles and the three integrable triangles admit sequences of Dirichlet or Neumann eigenfunctions that approximate any monochromatic wave around a set of points of asymptotically full measure, and around all points under symmetry assumptions. The negative results claim strong failure of inverse localization for irrational rectangles, for generic ellipses and all balls, and for the square and equilateral triangle under small Robin boundary conditions. The paper also extends the positive results to rational almost integrable polygons and draws consequences for nodal sets and critical points. The central advertised conclusion is that, in a generic integrable billiard, local limits of eigenfunctions do not match random waves.","tokens_in":78782,"tokens_out":11564,"duration_ms":119188,"significance":"If the results were all correct, the paper would be a significant step in the rigorous understanding of Berry's random wave conjecture for integrable systems: it would provide the first broad family of non-chaotic billiards where the random-wave local limit provably fails, complementing the known positive example of the flat torus. The positive theorems for rational rectangles and integrable triangles are carefully structured, use explicit trigonometric eigenfunctions and lattice-point equidistribution, and appear sound; the negative theorem for irrational rectangles via jet spaces is also internally coherent and elegant. These parts are of independent value. However, the proof of the ellipse theorem (Theorem D) contains a concrete coordinate-change error that invalidates the argument as written, and because the paper's abstract and conclusion emphasize generic integrable billiards including ellipses, this is a load-bearing failure rather than a local blemish.","major_comments":[{"comment":"The central identity in the ellipse proof is false. The manuscript defines the restriction operator R^{ξ0}_{η0} by composing a function with the elliptic-coordinate curve F(ξ,η)=(c cosh ξ cos η, c sinh ξ sin η) at fixed η0, and then asserts that for a Planck-scale localized eigenfunction one has \\tilde R^{ξ0}_{η0}(\\tilde u^{z0}_{mn})(ξ)=u_{mn}(ξ0+I_{ξ0}(ξ)/√λ, η0). But \\tilde u^{z0}_{mn}(x)=u_{mn}(F(ξ0,η0)+x/√λ), so the left-hand side is u_{mn}(F(ξ0,η0)+F(I_{ξ0}(ξ),η0)/√λ). This equals u_{mn}(ξ0+I_{ξ0}(ξ)/√λ,η0) only if F(ξ0+ε,η0)=F(ξ0,η0)+F(ε,η0), which is not satisfied by the hyperbolic trigonometric parametrization. For example, on the major axis η0=0, the elliptic coordinate of F(ξ0,0)+F(I,0)/√λ is arccosh(cosh ξ0+cosh I/√λ), whose first-order shift is (cosh I)/(sinh ξ0)·λ^{-1/2}, not I·λ^{-1/2}. Consequently the sampled restriction is not the shifted modified-Mathieu product S_m^p((ξ0+I)/√λ,q)E_n^p(η0,q), the claimed ODE for f_{s,t,m,n} does not apply to the composed function, and the inclusion EM_M(\\tilde R(\\tilde u^{z0}_{mn}))∈Q_M^{-1}({0}) does not follow. Since the generic-ellipse conclusion of Theorem D rests on this inclusion, Theorem D is unsupported as written.","section":"§5.3.2, Step II"},{"comment":"The density step in both the ball and ellipse proofs uses the assertion that a union of codimension-M sets parametrized continuously by a d-dimensional parameter family has codimension at least M-d. This is not automatic: one needs a parametrized transversality or Sard-type statement, and the manuscript does not provide it. Without such a statement, the conclusion that the set of monochromatic waves outside N_{B,BC} (or N_{Eb,BC}) is dense does not follow from the fiberwise codimension estimates. This gap affects the negative results for balls and ellipses, not just the ellipse theorem.","section":"§5.3.1 and §5.3.2, Step IV"}],"minor_comments":[{"comment":"The statement that inverse localization fails for 'most' integrable polygons and ellipses should be revised if Theorem D is removed or substantially weakened; the current abstract overstates what is proved without the ellipse result.","section":"§1.3 and Abstract"},{"comment":"There are typographical issues such as 'mayor axes' for 'major axes' and inconsistent use of 'Ea,BC' versus 'Eb'; these should be corrected.","section":"§2.2.2"},{"comment":"The definition of R^{ξ0}_{η0} uses the arc {F(ξ,η0)} centered near ξ0 but not at ξ0; this shift is confusing and is connected with the error in the main identity. The authors should clarify the relationship between the coordinate parameter and the physical localization center.","section":"§5.3.2, Step II"}],"recommendation":"reject","confidential_remarks":"The paper contains substantial valid contributions: the positive inverse-localization results for rational rectangles and integrable triangles, the negative result for irrational rectangles, and the ball result are all plausibly correct and well motivated. However, the generic-ellipse theorem is a headline conclusion of the manuscript, and its proof fails at a specific, identifiable step that is not a minor presentational issue: the nonlinearity of elliptic coordinates invalidates the exact identity on which the algebraic-variety argument depends. Fixing this would require a substantially new argument for ellipses, not a local correction. A resubmission that either supplies such an argument or reframes the paper around the polygon and ball results could be viable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the polygon results are the real contribution; Theorem D's ellipse proof has a coordinate-change error that is not cosmetic, so the generic-ellipse claim is unsupported as written.\n\nWhat is new and good: Theorem A's positive part gives inverse localization for rational rectangles and the three integrable triangles, with quantitative error bounds using lattice-point equidistribution. The negative part for irrational rectangles, via a jet-space argument, is a genuinely new strong-failure result. The nodal set applications in Section 7 are a nice payoff. As far as I can tell, these results are likely correct and form a meaningful step toward the integrable side of Berry's conjecture.\n\nThe soft spot is Section 5.3.2. The authors restrict a Planck-scale localized ellipse eigenfunction to an arc of a hyperbola L_{z0}, parametrized in Cartesian coordinates, and then assert that this restriction equals the Mathieu product evaluated at the shifted elliptic coordinate (ξ0 + I_{ξ0}(ξ)/√λ, η0). That would be true only if the Cartesian-to-elliptic map were affine. It is not. On the major axis, for example, the point z0 + (F(I, η0) - F(ξ0, η0))/√λ has elliptic coordinate arccosh(cosh ξ0 + cosh I/√λ) to first order, not ξ0 + I/√λ. The modified Mathieu ODE and the inclusion into the algebraic variety Q^{-1}({0}) both depend on this exact identity. Without it, the argument does not apply to the actual eigenfunctions. This is a load-bearing error, not a typo: it removes the full-measure generic-ellipse statement. The disk case, handled separately in Section 5.3.1 with the radial Bessel argument, seems fine.\n\nThe reader's weakest-assumption is exactly right, and the stress-test note holds up on reading the paper. This is a genuinely substantial paper with a real flaw in one of its four main theorems. I would not reject the whole paper, but I would send it back for major revision, expecting the authors either to fix Theorem D or to rescope the paper to the polygon results. The rectangle and triangle theorems are worth publishing on their own.\n\nRecommendation: send to peer review. A serious editor should not desk-reject; the error is specific and the remaining content is strong enough that referee time is warranted, with the ellipse claim flagged as needing repair.","headline":"The polygon results are the real contribution; Theorem D's ellipse proof has a coordinate-change error that is not cosmetic, so the generic-ellipse claim is unsupported as written.","tokens_in":79428,"tokens_out":5390,"would_cite":true,"duration_ms":49659,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P20","81Q50","58J50"],"pacs":[],"model":"deepseek-v4-flash","headline":"High-energy eigenfunctions of generic integrable billiards fail to reproduce random monochromatic waves, while rational rectangles and integrable triangles can approximate any such wave.","keywords":["random wave conjecture","integrable billiards","inverse localization","Planck scale","eigenfunction local limits","nodal sets","Mathieu functions","lattice point equidistribution"],"falsifier":"The decisive check is to compute the restriction $\\tilde R^{\\xi_0}_{\\eta_0}(\\tilde u^{z_0}_{mn})(\\xi)=S^p_m((\\xi_0+I_{\\xi_0}(\\xi))/\\sqrt{\\lambda_{mn}},q_{mn})E^p_n(\\eta_0,q_{mn})$ directly for a non-circular ellipse and test whether the evaluation map still lands in a fixed proper algebraic variety. Concretely, expand $F(\\xi,\\eta)=(c\\cosh\\xi\\cos\\eta,\\,c\\sinh\\xi\\sin\\eta)$ around $(\\xi_0,\\eta_0)$; since $F$ is not affine, the curve $z_0+F(I_{\\xi_0}(\\xi),\\eta_0)/\\sqrt{\\lambda}$ does not have elliptic coordinates $(\\xi_0+I_{\\xi_0}(\\xi)/\\sqrt{\\lambda},\\,\\eta_0)$, and the extra $O(|\\xi|^2/\\lambda)$ terms change the differential equation that the restriction satisfies. If those extra terms drive the jets out of the zero set of the polynomial $Q_M$, the density argument for Theorem D fails; if the terms vanish or stay inside the variety, the theorem stands.","tokens_in":78293,"feed_emoji":"📐","tokens_out":14404,"duration_ms":119215,"temperature":0.7,"pith_summary":"Berry's random wave conjecture says that on a chaotic billiard, high-energy eigenfunctions, looked at in a window of size $\\lambda^{-1/2}$ (the Planck scale), should resemble random monochromatic waves; one consequence, called inverse localization, is that any solution of the Helmholtz equation $\\Delta\\varphi+\\varphi=0$ should be approximated by eigenfunctions at the Planck scale around many points. This paper tests that expectation on the opposite class of systems, integrable billiards, whose eigenfunctions are known explicitly. It proves a dichotomy: rational rectangles and the three integrable triangles (equilateral, isosceles right, and hemiequilateral) possess inverse localization, approximating any monochromatic wave around almost every point; irrational rectangles, balls, and almost all ellipses fail it so strongly that a dense open set of waves is never approximated, so their local limits cannot be random waves. The paper concludes that random-wave-like behavior at the Planck scale is exceptional among integrable systems rather than typical.","feed_headline":"Most integrable billiards fail the random-wave test","feed_subtitle":"Rational rectangles and three special triangles can mimic every wave; irrational rectangles and generic ellipses cannot.","key_machinery":"Two mechanisms carry the argument. The positive mechanism splits a localized eigenfunction, built from the polygon's explicit trigonometric basis, into a translation-invariant kernel plus an error; equidistribution of lattice points on ellipsoids turns the kernel into an integral of $\\cos(\\xi\\cdot(z-z_\\gamma))$ over the unit circle, while an $L^2$ bound on the error (Lemma 3.3) plus a Lipschitz interpolation lemma (Lemma 3.4) shows the error is uniformly small on most base points. The negative mechanism is a jet-space obstruction: jets of order 3 of every localized eigenfunction are shown to lie in a fixed algebraic variety with empty interior inside the jet space of all monochromatic waves, and the open mapping theorem converts that empty interior into density and openness of the set of waves that are never approximated. For ellipses and disks the same obstruction is fed by Mathieu-function and Bessel-function eigenfunctions, whose restrictions are sent by a grid evaluation map into the zero set of an explicit polynomial system; a codimension count over the continuum of base points then shows the approximable waves have empty interior.","core_discovery":"On the paper's own terms, the discovery is a dichotomy inside the class of integrable billiards. Using the complete trigonometric basis of eigenfunctions on rectangles and the three integrable triangles, the authors construct, for each monochromatic wave $\\varphi$, a sequence of Dirichlet or Neumann eigenfunctions $u_{\\lambda_n}$ and open sets $O_{\\lambda_n,\\varphi}$ of almost full measure such that $\\|u_{\\lambda_n}(z_0+\\cdot/\\sqrt{\\lambda_n})-\\varphi\\|_{C^k(B)}<\\varepsilon$ for every base point $z_0\\in O_{\\lambda_n,\\varphi}$; if $\\varphi$ has the reflection symmetries of the polygon, the approximation holds at every fixed point. The negative half shows the opposite for every irrational rectangle $Q_l$, for balls in any dimension, and for ellipses $E_b$ with $b$ in a full-measure set of parameters: the set $\\mathcal{N}_{\\Omega,BC}$ of monochromatic waves that stay at positive distance from every localized eigenfunction is dense and open in the space of waves with sharp decay, which rules out random waves as local limits of any eigenfunction sequence. The paper's stated conclusion is that in a generic integrable billiard, the local limits of Dirichlet and Neumann eigenfunctions do not match random waves, contrary to what Berry's conjecture might suggest.","pith_inferences":["If the dichotomy is read as arithmetic rather than dynamical, the rational-versus-irrational split of rectangles suggests inverse localization is governed by resonance of side ratios; a testable prediction is that the exceptional ellipses in Theorem D are exactly those with rationally related Mathieu parameters.","Should the proofs hold up, this supplies the first rigorous integrable counterexamples to random-wave-like local limits, complementing the single known positive example of the flat torus.","The jet-space obstruction is a transferable template: for any separable system with explicit eigenfunctions (higher-dimensional ellipsoids, other coordinate separations), checking whether localized eigenfunction jets satisfy an algebraic relation of empty interior would decide inverse localization there.","A numerical experiment could sharpen the negative result: fix an irrational rectangle and a generic wave, and measure the distance from the wave to the localized eigenfunction jets as $\\lambda\\to\\infty$; the theorem predicts a positive limiting distance, and its rate of stabilization indicates how the jet span fails to fill the Helmholtz jet space."],"forward_implications":["Rational rectangles and the three integrable triangles have inverse localization: every monochromatic wave is reproduced at the Planck scale by eigenfunctions on almost all of the domain, and at every point once the wave obeys the polygon's reflection symmetries.","No eigenfunction sequence on an irrational rectangle, a ball, or a generic ellipse can have Berry's random waves as a local limit, because the approximable waves form a meager set.","The positive results extend to rational almost integrable polygons, with a fixed finite number of eigenfunction sequences jointly covering almost every point of the domain.","As applications, these polygons admit eigenfunctions with prescribed complicated nodal topology, arbitrary nesting of nodal domains, and arbitrarily many nondegenerate critical points inside a Planck-scale ball."],"supporting_citations":[{"why":"States the random wave conjecture that the paper tests in the integrable setting.","marker":"[Ber77]"},{"why":"Derandomization technique that proves random-wave local limits on the flat torus, the baseline positive example.","marker":"[Bou14]"},{"why":"Supplies the inverse localization framework for flat tori and the factorization lemma used for eigenvalues of irrational rectangles.","marker":"[EGRPS22]"},{"why":"Equidistribution of lattice points on ellipsoids that converts translation-invariant kernels into integrals over the unit circle.","marker":"[Iwa97]"},{"why":"Identifies the even-parameter equidistribution sequence used in the error bound for the isosceles right triangle.","marker":"[Cil93]"},{"why":"Provides the local weak limit formulation of Berry's conjecture and the torus subsequence whose local limits are random waves.","marker":"[Ing21]"},{"why":"Classifies integrable polygons, defining the domain class on which the dichotomy is proved.","marker":"[Gut84]"},{"why":"Guarantees existence of Mathieu characteristic pairings that produce eigenfunctions of ellipses.","marker":"[Nev10]"},{"why":"Shows the ellipse spectrum is simple away from a countable set, so every eigenfunction has the product form $S^p_m E^p_n$.","marker":"[HJ24]"},{"why":"Gives the jet surjectivity lemma stating every third-order jet is realized by a locally defined Helmholtz solution, used in the density arguments.","marker":"[Dam97]"}],"fun_headline_variants":["Random-wave conjecture fails for generic integrable billiards","Most integrable billiards can't mimic all waves at small scales","Berry's random waves? Only special billiards qualify","Inverse localization fails for most integrable polygons"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"For the ellipse theorem, the proof assumes that localizing an eigenfunction at a point shifts the radial Mathieu factor exactly as if elliptic coordinates were affine; this identification ignores the curvature of the coordinate map, and the density argument collapses if the neglected curvature terms let the jets of localized eigenfunctions escape the algebraic variety used in the proof.","fun_headline_variants_meta":{"raw":{"variants":["Random-wave conjecture fails for generic integrable billiards","Most integrable billiards can't mimic all waves at small scales","Berry's random waves? Only special billiards qualify","Inverse localization fails for most integrable polygons"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000563,"raw_usage":{"total_tokens":2739,"prompt_tokens":1079,"completion_tokens":1660,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":695,"completion_tokens_details":{"reasoning_tokens":1592}},"tokens_in":695,"tokens_out":1660,"duration_ms":11282,"temperature":1.0,"reasoning_tokens":1592,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T15:49:21.568596+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The decisive check is to compute the restriction $\\tilde R^{\\xi_0}_{\\eta_0}(\\tilde u^{z_0}_{mn})(\\xi)=S^p_m((\\xi_0+I_{\\xi_0}(\\xi))/\\sqrt{\\lambda_{mn}},q_{mn})E^p_n(\\eta_0,q_{mn})$ directly for a non-circular ellipse and test whether the evaluation map still lands in a fixed proper algebraic variety. Concretely, expand $F(\\xi,\\eta)=(c\\cosh\\xi\\cos\\eta,\\,c\\sinh\\xi\\sin\\eta)$ around $(\\xi_0,\\eta_0)$; since $F$ is not affine, the curve $z_0+F(I_{\\xi_0}(\\xi),\\eta_0)/\\sqrt{\\lambda}$ does not have elliptic coordinates $(\\xi_0+I_{\\xi_0}(\\xi)/\\sqrt{\\lambda},\\,\\eta_0)$, and the extra $O(|\\xi|^2/\\lambda)$ terms change the differential equation that the restriction satisfies. If those extra terms drive the jets out of the zero set of the polynomial $Q_M$, the density argument for Theorem D fails; if the terms vanish or stay inside the variety, the theorem stands.","supporting_citations":[],"review_version":1}