OctoInformBeta

OpenAI preprint, September 2026 · §3. Dyadic vector masses and common prices · §3

§3 (setting): tents, refinement and vector masses: Bilinear hats at each scale sum to one; scalar masses of an integrable hh at finitely many vertices total at most ∫|h|\int|h|

Dyadic hats and vector masses refine exactly between scales.

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

The vector measure behind the endpoint gradient bound for the centered disk maximal operator is hdxh\,dx, with h=∇gh = \nabla g the gradient of a real C1C^1 function gg of compact support. With the plane identified with ℂ\mathbb{C}, the gradient is an integrable function h:ℂ→ℂh : \mathbb{C} \to \mathbb{C}.

The argument discretizes hh at dyadic scales. At a scale rr the plane carries the square grid rℤ2r\mathbb{Z}^2, and hh is replaced by one complex number per vertex, its integral against the bilinear hat at that vertex. The questions are how the hats and these vertex masses at scale 2r2r are assembled from those at scale rr, and how much the moduli of the masses at finitely many vertices can add up to.

Index Sets and the Unit Cell

The vertices of a grid are indexed by p=(p1,p2)∈ℤ2p = (p_1, p_2) \in \mathbb{Z}^2. For N∈ℕN \in \mathbb{N}, the box grid GNG_N is the set of indices p∈ℤ2p \in \mathbb{Z}^2 with |p1|≤N|p_1| \le N and |p2|≤N|p_2| \le N.✦ The children of an index p∈ℤ2p \in \mathbb{Z}^2 form the set C(p)C(p) of the nine indices q∈ℤ2q \in \mathbb{Z}^2 with |q1−2p1|≤1|q_1 - 2p_1| \le 1 and |q2−2p2|≤1|q_2 - 2p_2| \le 1.✦

The unit cell [0,1)2[0,1)^2 is the set of points X∈ℂX \in \mathbb{C} whose real and imaginary parts both lie in [0,1)[0,1).✦ For a point z∈ℂz \in \mathbb{C} and a scale r>0r > 0, the rescaling at zz is the map y↦r−1(y−z)y \mapsto r^{-1}(y - z) of ℂ\mathbb{C} to itself.✦

The Claim

Definition 1 (Dyadic hats, refinement and masses). The tent is the function Λ:ℝ→ℝ\Lambda : \mathbb{R} \to \mathbb{R} given by Λ(t)=max⁡{1−|t|,0}.\Lambda(t) = \max\{1 - |t|,\, 0\}.(1)✦ For a scale r>0r > 0 and an index p∈ℤ2p \in \mathbb{Z}^2, the grid point of index pp is rp=rp1+irp2∈ℂrp = rp_1 + i\,rp_2 \in \mathbb{C}✦, and the bilinear hat at that vertex is the function φr,p:ℂ→ℝ\varphi_{r,p} : \mathbb{C} \to \mathbb{R} given by φr,p(x)=Λ(Re⁡xr−p1)Λ(Im⁡xr−p2).\varphi_{r,p}(x) = \Lambda\Big(\frac{\operatorname{Re} x}{r} - p_1\Big)\, \Lambda\Big(\frac{\operatorname{Im} x}{r} - p_2\Big).(2)✦ At every scale r>0r > 0 the hats form a partition of unity: for every x∈ℂx \in \mathbb{C}, ∑p∈ℤ2φr,p(x)=1.\sum_{p \in \mathbb{Z}^2} \varphi_{r,p}(x) = 1.(3)✦

The refinement weights λpq\lambda_{pq} from an index pp of the coarse grid, of scale 2r2r, to an index qq of the fine grid, of scale rr, are the numbers λpq=Λ(q12−p1)Λ(q22−p2),p,q∈ℤ2.\lambda_{pq} = \Lambda\Big(\frac{q_1}{2} - p_1\Big)\,\Lambda\Big(\frac{q_2}{2} - p_2\Big), \qquad p, q \in \mathbb{Z}^2.(4) They are normalized by columns: for every q∈ℤ2q \in \mathbb{Z}^2, ∑p∈ℤ2λpq=1.\sum_{p \in \mathbb{Z}^2} \lambda_{pq} = 1.(5)✦ With them the hats refine exactly: for every scale r>0r > 0, every p∈ℤ2p \in \mathbb{Z}^2 and every x∈ℂx \in \mathbb{C}, φ2r,p(x)=∑q∈ℤ2λpqφr,q(x).\varphi_{2r,p}(x) = \sum_{q \in \mathbb{Z}^2} \lambda_{pq}\,\varphi_{r,q}(x).(6)✦

Over a finest scale r>0r > 0, level k∈ℕk \in \mathbb{N} has the dyadic scale 2kr2^k r.✦ For N∈ℕN \in \mathbb{N}, the composed refinement coefficients λpq(k)\lambda^{(k)}_{pq}, for p,q∈GNp, q \in G_N, are the entries of the kkth power of the matrix (λpq)p,q∈GN(\lambda_{pq})_{p,q \in G_N}, where δpq\delta_{pq} is the Kronecker delta, equal to 11 if p=qp = q and to 00 otherwise: λpq(0)=δpq,λpq(k+1)=∑a∈GNλpaλaq(k).\lambda^{(0)}_{pq} = \delta_{pq}, \qquad \lambda^{(k+1)}_{pq} = \sum_{a \in G_N} \lambda_{pa}\,\lambda^{(k)}_{aq}.(7) They are values of the hats of level kk at the grid points of level 00: for every r>0r > 0, every k∈ℕk \in \mathbb{N} and all p,q∈GNp, q \in G_N, λpq(k)=φ2kr,p(rq).\lambda^{(k)}_{pq} = \varphi_{2^k r,\,p}(rq).(8)✦

Let h:ℂ→ℂh : \mathbb{C} \to \mathbb{C} be integrable, of total mass V=∫ℂ|h|V = \int_{\mathbb{C}} |h|. Its vector mass at the index pp on the scale r>0r > 0 is the complex number hr,ph_{r,p}✦, and its scalar mass there is the modulus ar,pa_{r,p}: hr,p=∫ℂφr,p(y)h(y)dy,ar,p=|hr,p|.h_{r,p} = \int_{\mathbb{C}} \varphi_{r,p}(y)\,h(y)\,dy, \qquad a_{r,p} = |h_{r,p}|.(9)✦ The vector masses refine over the children: for every r>0r > 0 and p∈ℤ2p \in \mathbb{Z}^2, h2r,p=∑q∈C(p)λpqhr,q.h_{2r,p} = \sum_{q \in C(p)} \lambda_{pq}\,h_{r,q}.(10)✦ For every r>0r > 0 and every finite set S⊂ℤ2S \subset \mathbb{Z}^2 of indices, ∑p∈Sar,p≤V.\sum_{p \in S} a_{r,p} \le V.(11)✦

For a point z∈ℂz \in \mathbb{C} and a scale r>0r > 0, the half-open cell Q(z,r)Q(z,r) with corner zz and side rr is the set of points that the rescaling at zz sends into the unit cell: Q(z,r)={y∈ℂ:r−1(y−z)∈[0,1)2}.Q(z,r) = \{\, y \in \mathbb{C} : r^{-1}(y - z) \in [0,1)^2 \,\}.(12)✦

In the paper, the result reads: at the scale rk=2−kr_k = 2^{-k} the vertices form Vk=rkℤ2V_k = r_k\mathbb{Z}^2, the vertex p∈Vkp \in V_k carries the half-open cell Qp=p+[0,rk)2Q_p = p + [0, r_k)^2 and the hat φp\varphi_p, a product of two one-dimensional tents, the refinement weights λpq\lambda_{pq} satisfy the identities (3.1), and the masses hp=∫φpdhh_p = \int \varphi_p\,dh and ap=|hp|a_p = |h_p|, with hp=apθph_p = a_p\theta_p for a direction θp\theta_p, obey ∑p∈Viap≤∑q∈Vkaq≤V\sum_{p \in V_i} a_p \le \sum_{q \in V_k} a_q \le V for i≤ki \le k, which is (3.3).

What Is Checked

The claim is stated over the tent Λ\Lambda, the bilinear hats φr,p\varphi_{r,p} at the grid points rprp and the dyadic scales 2kr2^k r, the refinement weights λpq\lambda_{pq} with the children C(p)C(p), the composed coefficients λpq(k)\lambda^{(k)}_{pq} on the box grid GNG_N, the vector and scalar masses hr,ph_{r,p} and ar,pa_{r,p} of an integrable h:ℂ→ℂh : \mathbb{C} \to \mathbb{C}, and the half-open cells Q(z,r)Q(z,r) cut out from the plane by the rescaling and the unit cell [0,1)2[0,1)^2.