OctoInformBeta

The maximal triangular Hilbert transform at the symmetric point

For complex functions FF and GG on ℝ2\mathbb{R}^2, hard truncation endpoints 0<ε<R<∞0 < \varepsilon < R < \infty, and (x,y)∈ℝ2(x,y) \in \mathbb{R}^2, the truncated triangular Hilbert transform is defined by Bε,R(F,G)(x,y)=∫ε<|t|<RF(x+t,y)G(x,y+t)dttB_{\varepsilon,R}(F,G)(x,y) = \int_{\varepsilon < |t| < R} F(x+t,y)G(x,y+t)\,\frac{dt}{t}. The maximal triangular Hilbert transform is B*(F,G)=sup⁡0<ε<R<∞|Bε,R(F,G)|B_*(F,G) = \sup_{0 < \varepsilon < R < \infty} |B_{\varepsilon,R}(F,G)|. For complex functions G0,G1,G2G_0, G_1, G_2 on ℝ2\mathbb{R}^2 and endpoints 0<ε<R<∞0 < \varepsilon < R < \infty, the scalar triangular form in symmetric coordinates is Λε,R(G0,G1,G2)=∭ε<|a+b+c|<RG0(a,b)G1(b,c)G2(c,a)dadbdca+b+c\Lambda_{\varepsilon,R}(G_0,G_1,G_2) = \iiint_{\varepsilon < |a+b+c| < R} G_0(a,b)G_1(b,c)G_2(c,a)\,\frac{da\,db\,dc}{a+b+c}. For exponents p0,p1,p2p_0, p_1, p_2 satisfying the scalar Hölder relation 1/p0+1/p1+1/p2=11/p_0 + 1/p_1 + 1/p_2 = 1, the choice p0=p1=p2=3p_0 = p_1 = p_2 = 3 is the symmetric point. The question is whether B*B_* maps L3(ℝ2)×L3(ℝ2)L^3(\mathbb{R}^2) \times L^3(\mathbb{R}^2) boundedly into L3/2(ℝ2)L^{3/2}(\mathbb{R}^2) and Λε,R\Lambda_{\varepsilon,R} admits an endpoint-uniform trilinear L3(ℝ2)L^3(\mathbb{R}^2) bound at the symmetric point.

Thiele stated the symmetric triangular Hilbert transform problem in Grafakos, Oliveira e Silva, Pramanik, Seeger, and Stovall. The triangular Hilbert transform appears in the discussion by Demeter and Thiele of bilinear averages for two commuting transformations. Estimates for a Walsh model with one structurally restricted input are due to Kovač, Thiele, and Zorin-Kranich. Sublogarithmic cancellation for one-dimensional multilinear Hilbert transforms was proved by Tao. Zorin-Kranich extended this cancellation to simplex Hilbert transforms. Across finitely many smooth scales, square-function estimates of Durcik, Kovač, Škreb, and Thiele gave square-root cancellation for the scalar triangular form. With hard truncation endpoints 0<ε<R<∞0 < \varepsilon < R < \infty, Durcik, Kovač, and Thiele proved power-type cancellation, giving a bound at the symmetric point that grows like log⁡(R/ε)\sqrt{\log(R/\varepsilon)} and retains dependence on the scale range. Bounds for averages over direction parameters were obtained by Durcik and Roos. For rough angular weights, Lin and Slavíková obtained bounds for directional averages. Christ, Durcik, and Roos proved bounds for curved translations along the curve (t,t2)(t,t^2). Hsu and Lin treated more general curves under geometric hypotheses excluding the flat curve (t,t)(t,t).

The September 24, 2026 preprint proves that there exists an absolute constant CC such that, for every pair of complex functions F,G∈L3(ℝ2)F, G \in L^3(\mathbb{R}^2), ∥B*(F,G)∥L3/2(ℝ2)≤C∥F∥L3(ℝ2)∥G∥L3(ℝ2)\|B_*(F,G)\|_{L^{3/2}(\mathbb{R}^2)} \le C\|F\|_{L^3(\mathbb{R}^2)}\|G\|_{L^3(\mathbb{R}^2)}. From this estimate, for complex functions F,G∈L3(ℝ2)F, G \in L^3(\mathbb{R}^2), the preprint also derives joint almost-everywhere and L3/2(ℝ2)L^{3/2}(\mathbb{R}^2) convergence of Bε,R(F,G)B_{\varepsilon,R}(F,G) as ε↓0\varepsilon \downarrow 0 and R↑∞R \uparrow \infty, together with an endpoint-uniform L3(ℝ2)L^3(\mathbb{R}^2) bound and joint limit as ε↓0\varepsilon \downarrow 0 and R↑∞R \uparrow \infty for Λε,R\Lambda_{\varepsilon,R} on complex L3(ℝ2)L^3(\mathbb{R}^2) triples, none of which the Lean library contains.

The Lean library proves this maximal inequality with a named constant built from the library’s dissipation, jump and kernel-error constants. The results on these pages are the Lean library’s, each stated as the Lean library states it and linked to the Lean declaration behind it. Lean has checked every result of the Lean library, with no unfinished proofs and no external result taken as an unproved hypothesis, so the Lean library is the reference. Where the preprint states a result differently, the statement on the page is the Lean library’s, with the preprint’s given beside it. The preprint is a guide only, giving the results their order and their names. OpenAI catalog entry 082 also lists two October 5, 2026 preprints, one on annular variation of the triangular Hilbert transform at the symmetric point and one on an L3L^3 bound for the dyadic triangular Hilbert form; neither of those preprints is covered, and the Lean library covers only the results of the September 24, 2026 preprint.

The Lean library: TriangularHilbert

TriangularHilbert is a Lean library: 23 items, in seven groups. Each has a page written by OctoInform, listed below.

§1. Introduction 2 items
  1. §1 (setting): the triangular Hilbert transform, maximal operator, and cyclic coordinates Defining the maximal triangular Hilbert operator and cyclic coordinates.
  2. Theorem 1.1 An L3/2L^{3/2} estimate for the maximal triangular Hilbert transform.
§2. A matrix energy and its Hessian 4 items
  1. §2 (setting): matrix energy, support, and Hessian quadratic form Matrix energy and its variational Hessian on complex matrices.
  2. Lemma 2.1 Derivatives and Hessian bounds for the matrix energy.
  3. Lemma 2.2 Sylvester solvability monotonicity and Hessian cost bounds under Loewner order.
  4. Lemma 2.3 Sylvester solvability and Hessian cost bounds for inserted matrices.
§3. A mixed trace inequality 2 items
  1. §3 (setting): cyclic row stars and Gram matrices Cyclic row stars and energy invariance under Gram flip.
  2. Lemma 3.1 Mixed trace inequality for Hermitian insertions and Hessian costs.
§4. Gaussian heat dissipation 6 items
  1. §4 (setting): Gaussian weights, scores, quadratic costs, and the center plane Gaussian weights and variational energies on center planes.
  2. Lemma 4.1 Continuity and differentiability of heat energy at positive variances.
  3. Lemma 4.2 Gaussian heat derivatives and plane dissipation for star energies.
  4. Lemma 4.3 Bounding summed neighbor Hessian costs by the star cost.
  5. Lemma 4.4 Gaussian plane heat integrals admit derivatives and dissipation bounds.
  6. Lemma 4.5 Bounds on Gaussian weight sums and plane heat energy.
§5. Output-dependent truncations and energy jumps 4 items
  1. §5 (setting): centered maximal averages and masked arrays Discrete centered maximal operators and masked heat cost.
  2. Lemma 5.1 Discrete maximal estimates and Gaussian domination on finite sets.
  3. Lemma 5.2 Differentiability and jump bounds for Gram matrix energy.
  4. Proposition 5.3 Cubic bounds on energy jumps and output-dependent masked dissipation.
§6. From masked energies to the maximal transform 4 items
  1. §6 (setting): odd Gaussian windows and the smooth kernel Defining the smooth kernel via odd Gaussian windows.
  2. Proposition 6.1 Bounds for entry-dependent smooth mesh truncations.
  3. Lemma 6.3 Hard-kernel decomposition and Gaussian endpoint-error bounds.
  4. Proposition 6.5 Moment bounds for hard-truncated cyclic kernel operators.
§7. General inputs and principal values 1 item
  1. Lemma 7.1 Finite truncations for L3L^3 inputs are integrable almost everywhere.

No item matches.