{"id":"70e07cf7-72f7-4ef1-b40f-b4d0e43a7bb0","arxiv_id":"2505.13256","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A soft local bound for triple product matrix coefficient integrals makes the implication 'subconvexity implies effective quantum unique ergodicity' hold uniformly for GL(2) forms in all spectral aspects.","lead":"This math paper proves a new, general upper bound for the local pieces in the triple product formula for GL(2) automorphic forms. It shows that a standard analytic hypothesis (subconvexity) yields effective quantum unique ergodicity, including for smooth test functions that earlier methods could not handle.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 8 overclaims: subconvexity-to-QUE gives weak-* convergence, not convergence for every bounded measurable observable.","rationale":"I reviewed the proof of Theorem 1 in detail. The reduction through Proposition 13, Lemma 14, Lemma 16, and the Rankin–Selberg interpolation is coherent; Lemma 16 is a cited standard identity, and the Phragmén–Lindelöf step in Lemma 21 is written imprecisely for complex s but is easily repaired for the real s needed in Proposition 22. I therefore did not find a fatal gap in the central local estimate. The load-bearing problem I did find is in the global formulation: Theorem 8 asserts equidistribution against every bounded measurable function, but the proof's spectral approximation argument can only establish convergence against bounded continuous functions (or functions continuous almost everywhere for the limit measure). The factor dim_K Ψ in Theorem 6 is infinite for a general bounded measurable observable, and sup-norm approximation by K-finite smooth functions is impossible for such an observable. This overclaim is not present in the reader's weakest-assumption analysis, but it affects a headline consequence of the paper. The main theorem and the smooth-observable consequence in Theorem 11 remain valid, so the appropriate verdict is conditional acceptance with Theorem 8 corrected to the standard QUE class of observables.","tokens_in":15079,"tokens_out":26300,"duration_ms":269080,"concrete_test":"Take Ψ = 1_E, the indicator of a measurable set E ⊂ [PGL2] whose boundary has positive Haar measure (e.g., a fat Cantor set). Trace the 'standard approximation argument' in the proof of Theorem 8: any approximation Ψ_T by K-finite smooth functions satisfies ||Ψ−Ψ_T||∞ ≥ 1 on a set of positive measure, so sup-norm convergence fails, while Theorem 6's bound is linear in dim_K Ψ_T and gives no uniform control as T→∞. If the proof cannot handle this Ψ, the theorem's 'bounded measurable' wording overclaims; replacing it with 'bounded continuous' (or Haar-a.e. continuous) is required.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The main local bound (Theorem 1) appears sound; the cited identity in Lemma 16 and the Phragmén–Lindelöf step are standard and repairable. The genuine soft spot is the global statement of Theorem 8, which asserts ∫_{[PGL2]} |φ_j|^2 Ψ → ∫ Ψ for each bounded measurable Ψ. The proof cites a 'standard approximation argument and spectral expansion' (after Theorem 6). This cannot work for arbitrary bounded measurable Ψ. The bound (9) in Theorem 6 has an explicit factor dim_K Ψ; for a non-K-finite function, dim_K Ψ = ∞, so the local bound gives no finite control. Moreover, weak-* convergence of the probability measures |φ_j|^2 to Haar gives convergence only for bounded continuous functions (or functions whose discontinuity set has Haar measure zero). A general bounded measurable function, e.g. the indicator of a set with positive-measure boundary, need not satisfy this. Spectral approximation of such Ψ by K-finite smooth functions holds in L^2, not in L∞; the resulting error is controlled by ||Ψ−Ψ_T||∞, which need not tend to 0, and the factor dim_K Ψ_T grows. Thus the asserted 'bounded measurable' conclusion goes beyond what subconvexity (or any weak-* equidistribution theorem) supplies. The correct statement should be for bounded continuous observables, which is the standard QUE formulation and is what the spectral method can prove.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a uniform upper bound for local triple product integrals on PGL_2(F) for generic irreducible unitary representations, with explicit polynomial dependence on K-types and analytic conductors, valid uniformly in the local field. The proof combines matrix coefficient bounds, a positivity reduction to the complementary series, the Michel–Venkatesh linearization identity for matrix coefficient integrals, and a Phragmén–Lindelöf convexity argument for local Rankin–Selberg integrals. The author then derives global consequences: subconvexity in the pi_1-aspect for L(1/2, pi_1 x tilde pi_1 x sigma) implies equidistribution of |phi_j|^2 (Theorem 8) and an effective, smooth-observable version of quantum unique ergodicity on the phase space (Theorem 11).","tokens_in":15350,"tokens_out":3403,"duration_ms":36801,"significance":"If the results are correct, this is a valuable soft contribution: it replaces case-by-case evaluation of local triple product factors with a single uniform estimate, removes the exponential K-type loss that limited earlier approaches to K-finite observables, and makes the subconvexity-to-QUE implication clean and all-aspect. The derivation is parameter-free and relies on standard, citable tools; the main local bound is plausible and well structured. The global applications, especially Theorem 11, are significant and would be of broad interest. However, the advertised generality of Theorem 8 is not supported as stated, and the key linearization identity is cited rather than proved; these points need to be addressed before the paper is in final form.","major_comments":[{"comment":"The conclusion (11) is asserted for every bounded measurable function Ψ on [PGL_2], but the proof cannot supply this. The cited bound (9) in Theorem 6 contains the factor dim_K Ψ, which is infinite for a general bounded measurable Ψ, so the local estimate gives no finite control for such observables. The 'standard approximation argument and spectral expansion' can approximate Ψ by K-finite smooth functions in L^2, not in L^∞, and the error term ‖Ψ−Ψ_T‖_∞ need not tend to zero; the dimension factor for the approximants grows and is not compensated. At best, subconvexity gives weak-* convergence against bounded continuous functions (or functions whose discontinuity set has Haar measure zero). Please restate Theorem 8 for bounded continuous observables, or prove the measurable case by an argument that actually controls the approximation error.","section":"§1.5, Theorem 8"},{"comment":"The identity ∫_{PGL_2(F)} |⟨gW,W⟩|² Ξ_s(g) dg = c_F I(s,W)I(1−s,W) is the bridge from the matrix coefficient integral to the Rankin–Selberg bound (21), and the entire proof of Proposition 15 depends on it. The lemma is cited to [MV10, Lemma 3.4.2] and several later references, with analytic continuation from Re(s)=1/2, but no proof or precise statement of the hypotheses (Whittaker model normalization, measure conventions, range of s) is given. Since this is load-bearing, please either include a proof in an appendix or state the exact version being used and explain how the cited references cover it, including the normalization of c_F ≍ 1.","section":"§3.2, Lemma 16"}],"minor_comments":[{"comment":"The assertion that ‖φ_q‖=1 for all q, obtained via raising and lowering operators, is stated without derivation or reference; please supply a short justification or a precise citation, since (15) relies on it.","section":"§1.6, proof of Theorem 11"},{"comment":"In the interpolation step, the exponent on the right is written as (d1d2)^{(1/2−α)/(1−2ϑ)+ε′}; the definition of ε′ and its passage to ε in the final bound should be made explicit for clarity.","section":"§3.1, Lemma 14"},{"comment":"The phrase 'by Cauchy's formula' at the end of Lemma 20(iv) is terse; a one-sentence explanation of how the bound away from poles is extended to all s with Re(s)≥0 would improve readability.","section":"§5.2, Lemma 21"},{"comment":"The typesetting 'GL 2' appears in several places where 'GL_2' is intended (e.g., the abstract and §1.1); please correct these for the final version.","section":"§1.2 and throughout"}],"recommendation":"major_revision","confidential_remarks":"The local bound and the overall structure appear sound, and the paper is likely to be accepted after the authors weaken Theorem 8 to the standard bounded-continuous formulation and give a more complete account of Lemma 16. The overclaim in Theorem 8 is the main issue; it is not merely cosmetic, but it is easily fixable within the manuscript's scope. I would not require a full proof of the Michel–Venkatesh identity if the authors provide the precise statement and a convincing derivation from the cited references."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result here is real. Theorem 1 gives a uniform upper bound for local triple product periods with polynomial dependence on the K-type of the observable, valid for arbitrary weight triples and uniformly across local fields and all aspects. That is genuinely new. Previous work either evaluated the archimedean factor in special cases (Watson) or got K-finite bounds with exponential loss in weight (BHM+24). Removing that exponential dependence is what allows the paper to extend the subconvexity-to-QUE implication to smooth observables, which is a meaningful step.\n\nThe proof is also clean. It recycles a handful of standard ingredients—matrix coefficient bounds from CHH88/Venkatesh, the Michel–Venkatesh linearization identity, the convexity principle for local Rankin–Selberg integrals—and the combination is transparent. The paper is honest about what it does not do: it only bounds the local factors, it does not evaluate them, and it compares its weaker t-dependence with the sharper, K-finite bound of BHM+24.\n\nThe soft spot is the global Theorem 8. It states that |φ_j|^2 equidistributes against every bounded measurable function on [PGL2]. That goes beyond what the proof supplies. The local bound in Theorem 6 carries the factor dim_K Ψ, which is infinite for non-K-finite Ψ. The standard approximation/spectral expansion argument works in L^2, and for arbitrary bounded measurable Ψ the tail ∥Ψ − Ψ_T∥∞ need not tend to zero. The correct statement is for bounded continuous observables (or for bounded measurable functions whose discontinuity set has Haar measure zero). This is a genuine overclaim, not a cosmetic one, but the fix is simple and does not affect the local results.\n\nThe reader's flagged dependence on Lemma 16 is fair. The identity is cited from Michel–Venkatesh rather than proved, and the central reduction rests on it. It is a standard and well-tested input, and I see no reason to doubt it, but a referee with local representation-theory expertise should check the measure normalization and the range of s.\n\nWho is this for? Anyone working on the subconvexity-to-QUE implication for GL2. It deserves a serious referee. I would accept after a revision that restricts Theorem 8 to bounded continuous observables.","headline":"The local bound in Theorem 1 is a genuine and useful new uniformity result, but Theorem 8 overclaims by asserting equidistribution against every bounded measurable observable.","tokens_in":15891,"tokens_out":4762,"would_cite":true,"duration_ms":48528,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["58J51","11F67","11F12"],"pacs":[],"model":"deepseek-v4-flash","headline":"For GL₂, a soft local bound makes subconvexity imply effective quantum unique ergodicity for automorphic forms.","keywords":["triple product L-functions","quantum unique ergodicity","subconvexity","Rankin-Selberg integrals","matrix coefficients","analytic conductor","GL2 automorphic forms","local bounds"],"falsifier":"For a ramified principal series representation over a non-archimedean local field, take a K-finite unit vector and compare both sides of the identity in Lemma 16 numerically or symbolically for several s in the stated range; a single mismatch, or a proof that the identity needs a hypothesis not stated in the paper, would falsify the main theorem.","tokens_in":2092,"feed_emoji":"🧮","tokens_out":2377,"duration_ms":120660,"temperature":0.7,"pith_summary":"This paper establishes a uniform upper bound for the local integrals appearing in the triple product formula for GL₂, with explicit dependence on the analytic conductors of the first two representations and polynomial or better dependence on the K-type of the third. The bound is uniform in the local field, which is what makes the proof soft: no exact evaluation of the local factors is needed. The intended consequence, stated as Theorems 8 and 11, is that a subconvex bound for the triple product L-function in the π₁-aspect automatically yields effective and general forms of quantum unique ergodicity for GL₂, including for smooth observables and across eigenvalue, weight, and level aspects. If the main theorem is correct, this explains the standard subconvexity-to-QUE mechanism while removing the exponential loss in weight that earlier local computations suffered.","feed_headline":"For GL2, a soft local bound makes subconvexity imply effective QUE","feed_subtitle":"Uniform over local fields and observable K-types, it also covers smooth test functions.","key_machinery":"The load-bearing object is the local matrix-coefficient integral on the left of (5), and the mechanism is a chain of reductions ending at local Rankin–Selberg integrals. After applying matrix coefficient bounds and a positivity argument that reduces the observable to a spherical complementary-series matrix coefficient, the paper invokes a linearization identity (Lemma 16): $\\int_{\\mathrm{PGL}_2(F)}|\\langle gW,W\\rangle|^2\\Xi_s(g)\\,dg = c_F I(s,W)I(1-s,W)$, where $I(s,W)=\\int_{N\\backslash\\mathrm{PGL}_2(F)}|W(g)|^2 f_s(g)\\,dg$ is a Rankin–Selberg integral. The remaining estimate is obtained by the Phragmén–Lindelöf convexity principle for these integrals, with the analytic conductor entering through the Stirling bound for the local γ-factor; a polynomial correction factor cancels finitely many poles near the strip. The uniformity in F comes from carrying the conductor through this convexity argument without evaluating any integral exactly.","core_discovery":"Let F be any local field and let π₁, π₂, π₃ be generic irreducible unitary representations of GL₂(F) with trivial product of central characters. If π_j is ϑ_j-tempered and $2\\max(\\vartheta_1,\\vartheta_2)+\\vartheta_3<1/2$, then for unit vectors $v_j, v_j'$ the integral $\\int_{\\mathrm{PGL}_2(F)}\\prod_{j=1}^3 \\langle g v_j, v_j'\\rangle\\, dg$ is bounded by $O_{\\vartheta,\\varepsilon}((\\dim_K v_3)^{\\Theta+\\varepsilon}(\\dim_K v_3')^{\\Theta+\\varepsilon}/(C(\\pi_1\\times\\tilde\\pi_1)^{1/4}C(\\pi_2\\times\\tilde\\pi_2)^{1/4}))$, uniformly in F, with $\\Theta=(\\vartheta_1+\\vartheta_2)/(1-2\\vartheta_3)<1/2$. This is Theorem 1. The paper feeds this local estimate into the global triple product formula, obtaining a global inequality (Theorem 6), and derives two consequences: subconvexity in the π₁-aspect forces the L² mass of any non-dihedral sequence of cusp forms to equidistribute (Theorem 8), and under the standard subconvex bound it yields effective convergence of Wigner distributions against smooth observables (Theorem 11). The proof deliberately avoids computing the local factors exactly, using only matrix coefficient bounds, a linearization identity for matrix coefficient integrals, and the convexity principle for local Rankin–Selberg integrals.","pith_inferences":["A natural extension beyond this paper: the same soft route should apply to generalized triple products on higher-rank groups whenever an analogue of the linearization identity and a conductor-weighted convexity bound are available; this paper proves the GL₂ case only.","Implicit in the proof is a recipe for upgrading existing QUE results: replace explicit hypergeometric evaluations of archimedean local factors by Theorem 1, sacrificing sharp dependence on one spectral parameter but gaining uniformity in the observable's weight.","A testable prediction is that the exponent $\\Theta=(\\vartheta_1+\\vartheta_2)/(1-2\\vartheta_3)$ exactly controls the price of non-temperedness in the observable, so any improvement over the known 7/64 temperedness bound would improve the implied constants in Theorems 8 and 11."],"forward_implications":["Theorem 8: if $L(1/2,\\pi_j\\times\\tilde\\pi_j\\times\\sigma)$ decays faster than $C(\\pi_j\\times\\tilde\\pi_j)^{1/2}$ along a sequence of non-dihedral forms with bounded ramification, then $|\\varphi_j|^2$ converges to Haar measure on [PGL₂].","Theorem 11: the standard subconvex bound in the form (12) yields $|\\omega_\\varphi(\\Psi)-\\mu(\\Psi)|\\le B\\lambda^{-\\delta}$ for every smooth test function $\\Psi$ on phase space.","The K-type dependence of the local bound is $(\\dim_K v_3)^{\\Theta+\\varepsilon}$ with $\\Theta<1/2$, which is polynomial rather than exponential in the weight; this is precisely what allows smooth observables rather than only K-finite ones.","The same local estimate holds uniformly at archimedean and non-archimedean places, so the eigenvalue, weight, level, and local-field aspects are all handled in one statement.","For square-integrable representations the proof simplifies and gives the bound with $\\Theta=\\varepsilon=0$, recovering a formal-degree calculation."],"supporting_citations":[{"why":"Gives the general triple product formula that relates the global period to a product of local factors and an L-value; the local factors are exactly what Theorem 1 bounds.","marker":"[Ich08]"},{"why":"Establishes the triple product formula for the central critical value, the background identity that Ichino's formula generalizes.","marker":"[HK91]"},{"why":"Supplies the linearization identity (Lemma 3.4.2) that converts the matrix-coefficient integral into a product of Rankin–Selberg integrals, reproduced as Lemma 16.","marker":"[MV10]"},{"why":"Provides the convexity principle for local Rankin–Selberg integrals, generalized to all local fields, that powers the key estimate (21).","marker":"[NPS14]"},{"why":"Yields the almost L² matrix coefficient bounds used to justify the temperedness assumptions and the reduction to complementary series.","marker":"[CHH88]"},{"why":"Lemma 9.1 supplies the diagonal matrix coefficient decay used in Lemma 14 to interpolate the K-type dependence.","marker":"[Ven10]"},{"why":"Gives the 7/64 temperedness bound for automorphic representations, used to verify the hypotheses of Theorem 1 in global applications.","marker":"[BB11]"},{"why":"Computes the archimedean local factor in the classical case that the paper's uniform estimate recovers and generalizes.","marker":"[Wat02]"},{"why":"Proved the K-finite case of the smooth-observable QUE theorem; the K-type uniformity here is what upgrades that result to all smooth observables.","marker":"[BHM+24]"}],"fun_headline_variants":["Soft local bound turns subconvexity into effective QUE for GL2","Uniform local triple bound yields effective QUE from subconvexity","Soft proof: subconvexity forces equidistribution on GL2","No exact local factors: subconvexity alone implies effective QUE"],"cache_read_input_tokens":18048,"weakest_assumption_plain":"The proof leans on a known identity, cited rather than proved, that decomposes a certain integral of matrix coefficients into a product of two period integrals; if that identity has hidden hypotheses or fails for some local field or vector, the central bound collapses.","fun_headline_variants_meta":{"raw":{"variants":["Soft local bound turns subconvexity into effective QUE for GL2","Uniform local triple bound yields effective QUE from subconvexity","Soft proof: subconvexity forces equidistribution on GL2","No exact local factors: subconvexity alone implies effective QUE"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000539,"raw_usage":{"total_tokens":2565,"prompt_tokens":904,"completion_tokens":1661,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":520,"completion_tokens_details":{"reasoning_tokens":1586}},"tokens_in":520,"tokens_out":1661,"duration_ms":12619,"temperature":1.0,"reasoning_tokens":1586,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:17:57.067325+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a ramified principal series representation over a non-archimedean local field, take a K-finite unit vector and compare both sides of the identity in Lemma 16 numerically or symbolically for several s in the stated range; a single mismatch, or a proof that the identity needs a hypothesis not stated in the paper, would falsify the main theorem.","supporting_citations":[],"review_version":1}