OctoInformBeta

OpenAI preprint, September 2026 · §2. The finite-interval action inequality · §2

§2 (setting): finite-interval action and lower triangular coefficients: The Sobolev space H01H^1_0 on bounded intervals and lower-triangular matrices, with the pointwise action density

Finite-interval action functional and the extended-real action claim.

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

Fix a bounded open interval (l,r)(l, r) with l<rl < r and a spectral exponent γ∈(1/2,3/2)\gamma \in (1/2, 3/2) with σ=γ−1/2∈(0,1)\sigma = \gamma - 1/2 \in (0, 1). For any nonnegative integer NN and positive parameters k1,…,kN>0k_1, \dots, k_N > 0, transforming a family of NN functions in the Dirichlet Sobolev space on (l,r)(l, r) that is orthonormal in L2((l,r))L^2((l, r)) by a lower-triangular N×NN \times N real matrix yields a test input for a finite-interval action functional. The question is how the supremum of this action functional over all lower-triangular coefficient matrices compares with the one-dimensional spectral integrals attached to σ\sigma and k1,…,kNk_1, \dots, k_N.

Sobolev Space and Lower Triangular Matrices

For real numbers ll and rr, the Sobolev space H01(l,r)H^1_0(l, r) consists of all pairs (v,v′)(v, v') of functions ℝ→ℝ\mathbb{R} \to \mathbb{R} satisfying (1) for every smooth compactly supported function φ:ℝ→ℝ\varphi : \mathbb{R} \to \mathbb{R} vanishing outside (l,r)(l, r): v,v′∈L2((l,r)),v is continuous on [l,r],v(l)=v(r)=0,∫(l,r)v(x)φ′(x)dx=−∫(l,r)v′(x)φ(x)dx.v, v' \in L^2((l, r)), \quad v \text{ is continuous on } [l, r], \quad v(l) = v(r) = 0, \quad \int_{(l, r)} v(x)\varphi'(x)\,dx = -\int_{(l, r)} v'(x)\varphi(x)\,dx.(1)✦

Given a nonnegative integer NN, a real N×NN \times N matrix C=(Cij)1≤i,j≤N∈ℝN×NC = (C_{ij})_{1 \le i, j \le N} \in \mathbb{R}^{N \times N} is lower triangular if it satisfies (2): Cij=0for all 1≤i<j≤N.C_{ij} = 0 \qquad \text{for all } 1 \le i < j \le N.(2)✦

The Definition

Definition 1. The finite-interval action ℰk,σ(u,C)\mathcal{E}_{k,\sigma}(u, C) of an NN-tuple u=((u1,u1′),…,(uN,uN′))∈H01(l,r)Nu = ((u_1, u_1'), \dots, (u_N, u_N')) \in H^1_0(l, r)^N as in (1) and a matrix C∈ℝN×NC \in \mathbb{R}^{N \times N}, for a nonnegative integer NN, real numbers l,r,σl, r, \sigma, and a vector k=(k1,…,kN)∈ℝNk = (k_1, \dots, k_N) \in \mathbb{R}^N, is the real number defined by (3), writing u(x)=(u1(x),…,uN(x))T∈ℝNu(x) = (u_1(x), \dots, u_N(x))^T \in \mathbb{R}^N and u′(x)=(u1′(x),…,uN′(x))T∈ℝNu'(x) = (u_1'(x), \dots, u_N'(x))^T \in \mathbb{R}^N: ℰk,σ(u,C)=∫(l,r)(∑i=1N(Cu′(x))i2+∑i=1Nki2(Cu(x))i2−(∑i=1N(Cu(x))i2)1+1/σ)dx.\mathcal{E}_{k,\sigma}(u, C) = \int_{(l, r)} \left( \sum_{i=1}^N (C u'(x))_i^2 + \sum_{i=1}^N k_i^2 (C u(x))_i^2 - \left(\sum_{i=1}^N (C u(x))_i^2\right)^{1 + 1/\sigma} \right) dx.(3)✦

Let NN be a nonnegative integer, l<rl < r real numbers, σ∈(0,1)\sigma \in (0, 1), k=(k1,…,kN)∈ℝNk = (k_1, \dots, k_N) \in \mathbb{R}^N a vector with ki>0k_i > 0 for all 1≤i≤N1 \le i \le N, and u=((u1,u1′),…,(uN,uN′))∈H01(l,r)Nu = ((u_1, u_1'), \dots, (u_N, u_N')) \in H^1_0(l, r)^N an NN-tuple satisfying ∫(l,r)ui(x)uj(x)dx=1\int_{(l, r)} u_i(x) u_j(x)\,dx = 1 for i=ji = j and 00 for i≠ji \neq j. Then the extended-real inequality (4) holds in ℝ¯=[−∞,+∞]\overline{\mathbb{R}} = [-\infty, +\infty], where lower triangularity is defined by (2) and ℰk,σ(u,C)\mathcal{E}_{k,\sigma}(u, C) by (3): ∑i=1N∫−kiki(ki2−s2)σds≤supC∈ℝN×NC lower triangularℰk,σ(u,C).\sum_{i=1}^N \int_{-k_i}^{k_i} (k_i^2 - s^2)^\sigma\,ds \le \sup_{\substack{C \in \mathbb{R}^{N \times N} \\ C \text{ lower triangular}}} \mathcal{E}_{k,\sigma}(u, C).(4)✦

The pointwise action density ℒk,σ(a,v)∈ℝ\mathcal{L}_{k,\sigma}(a, v) \in \mathbb{R} for a nonnegative integer NN, σ∈ℝ\sigma \in \mathbb{R}, and vectors k,a,v∈ℝNk, a, v \in \mathbb{R}^N is given by (5): ℒk,σ(a,v)=∑i=1Nvi2+∑i=1Nki2ai2−(∑i=1Nai2)1+1/σ.\mathcal{L}_{k,\sigma}(a, v) = \sum_{i=1}^N v_i^2 + \sum_{i=1}^N k_i^2 a_i^2 - \left(\sum_{i=1}^N a_i^2\right)^{1 + 1/\sigma}.(5)✦

In the paper, the result reads: for σ=γ−1/2∈(0,1)\sigma = \gamma - 1/2 \in (0, 1), p=1+σp = 1 + \sigma, q=1+1/σq = 1 + 1/\sigma, a bounded interval I=(l,r)I = (l, r), a positive diagonal matrix K=diag⁡(k1,…,kN)>0K = \operatorname{diag}(k_1, \dots, k_N) > 0, and a∈H01(I;ℝN)a \in H^1_0(I; \mathbb{R}^N), ℰK(a)=∫I(|a′|2+aTK2a−|a|2q)dx\mathcal{E}_K(a) = \int_I (|a'|^2 + a^T K^2 a - |a|^{2q})\,dx, where (l,r)(l, r) stands for the paper’s (x−,x+)(x_-, x_+) and k=(k1,…,kN)k = (k_1, \dots, k_N) gives the diagonal of KK.

What Is Checked

The action claim (4) is stated over nonnegative integers NN, bounded open intervals (l,r)(l, r) with l<rl < r, parameters σ∈(0,1)\sigma \in (0, 1), positive vectors k∈ℝNk \in \mathbb{R}^N, NN-tuples in H01(l,r)NH^1_0(l, r)^N orthonormal in L2((l,r))L^2((l, r)) as in (1), and the extended-real supremum of the finite-interval action (3) over lower-triangular matrices (2), while the pointwise action density (5) is defined for real σ\sigma and vectors in ℝN\mathbb{R}^N.