OctoInformBeta

OpenAI preprint, September 2026 · §1. Introduction · §1

Proposition 1.2 [Signed finite-band estimate]: For smooth compactly supported gg on the plane, ∫|∇Sa,2mag|≤C∫|∇g|\int|\nabla S_{a,2^m a} g| \le C\int|\nabla g| with one constant CC for all a>0a > 0 and mm

A uniform L1L^1 gradient bound for dyadic bands of disk averages.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Let gg be a smooth real function with compact support on the plane. Averaging |g||g| over the disk of radius tt centered at a point xx, and taking the supremum of these averages over all t>0t > 0, gives the centered Hardy-Littlewood maximal function of gg. At the endpoint p=1p = 1 of the Sobolev exponents W1,pW^{1,p}, the question is whether the maximal function again has an integrable gradient, of total mass at most a fixed multiple of ∫|∇g|\int|\nabla g|.

Proposition 1.2 is a signed form of this bound on a band of radii. The averages are those of gg itself, not of |g||g|, and for real numbers 0<a≤b0 < a \le b with b/ab/a a power of two the supremum runs only over radii tt with a≤t≤ba \le t \le b; the multiple may not depend on aa, bb or the number of dyadic scales between them.

Disk Averages and Band Envelopes

The Euclidean plane ℝ2\mathbb{R}^2 is the two-dimensional real vector space with the Euclidean norm |v|=v12+v22|v| = \sqrt{v_1^2 + v_2^2}.✦ The complex plane ℂ\mathbb{C} is identified with ℝ2\mathbb{R}^2 by a fixed isometry, so that gradients on ℂ\mathbb{C} are complex numbers and |⋅||\cdot| is the modulus; the averages and envelopes below are defined on both by the same formulas.

The disk average Atg(x)A_t g(x) of a real function gg on the plane, at a point xx and a radius tt, is the integral of gg over the open disk B(x,t)B(x,t) of center xx and radius tt, divided by the area of that disk: Atg(x)=1πt2∫B(x,t)g(y)dy.A_t g(x) = \frac{1}{\pi t^2} \int_{B(x,t)} g(y)\,dy.(1)✦ We write ut=Atgu_t = A_t g.

The radius envelope of a family (ut)(u_t) of real functions on a space EE, indexed by real radii tt, over real numbers aa and bb, is the function on EE whose value at xx is the supremum of ut(x)u_t(x) over the closed interval a≤t≤ba \le t \le b.✦ The signed band envelope Sa,bgS_{a,b} g of a real function gg on the plane is the radius envelope of its disk averages: Sa,bg(x)=supa≤t≤bAtg(x).S_{a,b} g(x) = \sup_{a \le t \le b} A_t g(x).(2)✦

Weak Gradients

A vector field GG on ℝ2\mathbb{R}^2 is a weak gradient of a real function FF on ℝ2\mathbb{R}^2 if, for every smooth real function φ\varphi with compact support on ℝ2\mathbb{R}^2 and each i∈{1,2}i \in \{1, 2\}, ∫ℝ2F∂iφdx=−∫ℝ2Giφdx,\int_{\mathbb{R}^2} F\,\partial_i \varphi\,dx = -\int_{\mathbb{R}^2} G_i\,\varphi\,dx,(3)✦ where ∂iφ\partial_i \varphi is the partial derivative of φ\varphi along the ii-th coordinate axis.

For a vector field GG on ℝ2\mathbb{R}^2, the L1L^1 norm of GG is ∫ℝ2|G(x)|dx\int_{\mathbb{R}^2} |G(x)|\,dx, with |⋅||\cdot| the Euclidean length.✦

Dyadic Radii

Dyadic levels are counted from the finest: above a finest scale rr, level kk has scale 2kr2^k r.✦ The geometric radius grid of base a>0a > 0 and depth m∈ℕm \in \mathbb{N} is the descending sequence that starts at the scale of level mm above aa and halves at each step, rk=2ma2k,k∈ℕ,r_k = \frac{2^m a}{2^k}, \qquad k \in \mathbb{N},(4)✦ so that r0=2mar_0 = 2^m a and rm=ar_m = a.

For a real function gg on ℂ\mathbb{C} and a sequence of radii (rk)(r_k), the dyadic run envelope of gg from level ii to level jj is the band envelope from the radius of level jj to twice the radius of level ii: Ri,jg=Srj,2rig.R_{i,j}\, g = S_{r_j,\,2 r_i}\, g.(5)✦

The Claim

Theorem 1 (Signed finite-band estimate).

  1. Let gg be a smooth real function with compact support on ℝ2\mathbb{R}^2, let a>0a > 0 and let m∈ℕm \in \mathbb{N}. Then the envelope Sa,2magS_{a,2^m a} g is locally integrable and has an integrable weak gradient GG satisfying (6), with a constant C≥0C \ge 0 independent of gg, aa and mm. ∫ℝ2|G(x)|dx≤C∫ℝ2|∇g(x)|dx.\int_{\mathbb{R}^2} |G(x)|\,dx \le C \int_{\mathbb{R}^2} |\nabla g(x)|\,dx.(6)✦

In (ii) to (v), gg is a real function of class C1C^1 with compact support on ℂ\mathbb{C}, a>0a > 0, m∈ℕm \in \mathbb{N}, (rk)(r_k) is the geometric radius grid of base aa and depth mm, and ∇\nabla is the pointwise gradient.

  1. There is a constant C≥0C \ge 0, independent of gg, aa and mm, with ∫ℂ|∇Sa,2mag(x)|dx≤C∫ℂ|∇g(x)|dx.\int_{\mathbb{C}} |\nabla S_{a,2^m a} g(x)|\,dx \le C \int_{\mathbb{C}} |\nabla g(x)|\,dx.(7)✦

  2. There is a constant C>0C > 0, independent of gg, aa and mm, with ∫ℂ|∇Sa,2m+1ag(x)|dx≤C∫ℂ|∇g(x)|dx.\int_{\mathbb{C}} |\nabla S_{a,2^{m+1} a} g(x)|\,dx \le C \int_{\mathbb{C}} |\nabla g(x)|\,dx.(8)✦

  3. For the single radius b=ab = a the constant is one: ∫ℂ|∇Sa,ag(x)|dx≤∫ℂ|∇g(x)|dx.\int_{\mathbb{C}} |\nabla S_{a,a} g(x)|\,dx \le \int_{\mathbb{C}} |\nabla g(x)|\,dx.(9)✦

  4. The run envelope of the grid from level 00 to level mm obeys, with a constant C>0C > 0 independent of gg, aa and mm, ∫ℂ|∇R0,mg(x)|dx≤C∫ℂ|∇g(x)|dx.\int_{\mathbb{C}} |\nabla R_{0,m}\, g(x)|\,dx \le C \int_{\mathbb{C}} |\nabla g(x)|\,dx.(10)✦

In the paper, the result reads: for every real g∈Cc∞(ℝ2)g \in C_c^\infty(\mathbb{R}^2) and all 0<a≤b0 < a \le b with b/ab/a a power of two, one included, ∫|∇Sa,bg|≤C∫|∇g|\int|\nabla S_{a,b} g| \le C \int |\nabla g| with an absolute constant CC, independent of aa, bb and the number of dyadic radius scales. Part (i) answers it with b=2mab = 2^m a, the gradient of Sa,bgS_{a,b} g taken in the weak sense.

What Is Checked

Part (i) is the library’s signed finite-band bound, stated over the Euclidean plane ℝ2\mathbb{R}^2, the disk averages (1), the signed band envelope (2), weak gradients (3) and the L1L^1 norm of a vector field; parts (ii) to (v) are stated over the complex plane, the same averages and envelopes, the geometric radius grid (4) and the dyadic run envelope (5).