OctoInformBeta

OpenAI preprint, September 2026 · §2. Finite coordinates and normalized increments

Lemma 2.1: An nn-point subset fits in (n2)\binom n2 real coordinates for p≥1p\ge1, including empty and singleton subsets

Finite LpL_p metrics admit exact coordinates.

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

Lemma 2.1 concerns an exact realization of the LpL_p distances between points in finitely many real coordinates.

For a real exponent pp and a nonnegative integer dd, the coordinate distance ρp(u,v)\rho_p(u,v) of vectors u,v∈ℝdu,v\in\mathbb R^d is the real number given by (1). ρp(u,v)=(∑a=1d|ua−va|p)1/p.\rho_p(u,v)=\left(\sum_{a=1}^{d}|u_a-v_a|^p\right)^{1/p}.(1)✦ Write νd\nu_d for counting measure on {1,…,d}\{1,\ldots,d\}, so that Lp(νd)L_p(\nu_d) denotes the real LpL_p space on this finite measure space.

The Claim

Theorem 1. Let p≥1p\ge1, let (Ω,μ)(\Omega,\mu) be a measure space, and let x1,…,xnx_1,\ldots,x_n be distinct points of real Lp(μ)L_p(\mu), where nn is a nonnegative integer. Put m=(n2)m=\binom n2.

For n≥2n\ge2, Ball’s finite isometry gives a map y:{1,…,n}→ℝmy:\{1,\ldots,n\}\to\mathbb R^m satisfying (2): ρp(yi,yj)=∥xi−xj∥Lp(μ)(1≤i,j≤n).\rho_p(y_i,y_j)=\|x_i-x_j\|_{L_p(\mu)} \qquad (1\le i,j\le n).(2)✦ For arbitrary nn, including n=0n=0 and n=1n=1, there is a map y:{1,…,n}→ℝmy:\{1,\ldots,n\}\to\mathbb R^m satisfying (3): ρp(yi,yj)=∥xi−xj∥Lp(μ)(1≤i,j≤n).\rho_p(y_i,y_j)=\|x_i-x_j\|_{L_p(\mu)} \qquad (1\le i,j\le n).(3)✦ For arbitrary nn, there is an injective coordinate map y:{1,…,n}→ℝmy:\{1,\ldots,n\}\to\mathbb R^m satisfying (4): ρp(yi,yj)=∥xi−xj∥Lp(μ)(1≤i,j≤n).\rho_p(y_i,y_j)=\|x_i-x_j\|_{L_p(\mu)} \qquad (1\le i,j\le n).(4)✦ There is also an injective map z:{1,…,n}→Lp(νm)z:\{1,\ldots,n\}\to L_p(\nu_m) satisfying (5): ∥zi−zj∥Lp(νm)=∥xi−xj∥Lp(μ)(1≤i,j≤n).\|z_i-z_j\|_{L_p(\nu_m)}=\|x_i-x_j\|_{L_p(\mu)} \qquad (1\le i,j\le n).(5)✦

What Is Checked

The claims are stated over injectively indexed finite families in real Lp(μ)L_p(\mu) for arbitrary measure spaces and p≥1p\ge1, with coordinate distance (1) and counting measure on the target coordinate set.