OctoInformBeta

OpenAI preprint, September 2026 · §1. The problem and the result

Theorem 1.1 [Smooth Coulomb nonattainment]: A smooth compact density with smooth square root, equal infima and finite-cost Monge approximation

A Coulomb minimum without a Monge optimizer.

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 three-marginal Coulomb problem minimizes the repulsion of three particles in ℝ3\mathbb R^3 while prescribing the same probability measure for each particle. A configuration is a triple (x,y,z)∈(ℝ3)3(x,y,z)\in(\mathbb R^3)^3; write p1,p2,p3p_1,p_2,p_3 for its coordinate projections. The inverse distance of xx and yy is |x−y|−1∈[0,+∞]|x-y|^{-1}\in[0,+\infty], with value +∞+\infty when x=yx=y.✦ The Coulomb cost c:(ℝ3)3→[0,+∞]c:(\mathbb R^3)^3\to[0,+\infty] is given by (1), with value +∞+\infty at every collision: c(x,y,z)=1|x−y|+1|x−z|+1|y−z|.c(x,y,z)=\frac{1}{|x-y|}+\frac{1}{|x-z|}+\frac{1}{|y-z|}.(1)✦

A three-marginal self-coupling π\pi of a Borel measure μ\mu on ℝ3\mathbb R^3 is a probability measure on (ℝ3)3(\mathbb R^3)^3 whose three coordinate marginals are μ\mu.✦ The set Π(μ,μ,μ)\Pi(\mu,\mu,\mu) of self-couplings of a Borel measure μ\mu on ℝ3\mathbb R^3 consists of probability measures on (ℝ3)3(\mathbb R^3)^3 whose three coordinate marginals are μ\mu, as in (2): Π(μ,μ,μ)={π a probability measure on (ℝ3)3:(pi)#π=μfor i=1,2,3}.\Pi(\mu,\mu,\mu)= \left\{\pi\text{ a probability measure on }(\mathbb R^3)^3: (p_i)_\#\pi=\mu\ \text{for }i=1,2,3\right\}.(2)✦ The Kantorovich value E*(μ)E_*(\mu) of a Borel measure μ\mu on ℝ3\mathbb R^3 is the infimum of the Coulomb integral over Π(μ,μ,μ)\Pi(\mu,\mu,\mu), as in (3): E*(μ)=infπ∈Π(μ,μ,μ)∫c(x,y,z)dπ(x,y,z).E_*(\mu)=\inf_{\pi\in\Pi(\mu,\mu,\mu)} \int c(x,y,z)\,d\pi(x,y,z).(3)✦

A measure-preserving map for μ\mu is a Borel map T:ℝ3→ℝ3T:\mathbb R^3\to\mathbb R^3 with T#μ=μT_\#\mu=\mu; let 𝒯μ\mathcal T_\mu be the set of these maps.✦ For a Borel measure μ\mu on ℝ3\mathbb R^3 and maps T2,T3:ℝ3→ℝ3T_2,T_3:\mathbb R^3\to\mathbb R^3, the graph cost 𝒢μ(T2,T3)\mathcal G_\mu(T_2,T_3) is the nonnegative extended Coulomb integral along the graph of (T2,T3)(T_2,T_3)✦, and the Monge value EM(μ)E_{\mathrm M}(\mu) is its infimum over measure-preserving pairs, as in (4): 𝒢μ(T2,T3)=∫c(x,T2(x),T3(x))dμ(x),EM(μ)=infT2,T3∈𝒯μ𝒢μ(T2,T3).\begin{aligned} \mathcal G_\mu(T_2,T_3) &=\int c\bigl(x,T_2(x),T_3(x)\bigr)\,d\mu(x),\\ E_{\mathrm M}(\mu) &=\inf_{T_2,T_3\in\mathcal T_\mu}\mathcal G_\mu(T_2,T_3). \end{aligned}(4)

Densities and Attainment

A smooth compact probability density is a nonnegative function ρ:ℝ3→ℝ\rho:\mathbb R^3\to\mathbb R such that ρ,ρ∈Cc∞(ℝ3)\rho,\sqrt{\rho}\in C_c^\infty(\mathbb R^3) and ∫ρdx=1\int\rho\,dx=1.✦ The density measure μ\mu associated with a real function ρ:ℝ3→ℝ\rho:\mathbb R^3\to\mathbb R is the Borel measure obtained by weighting Lebesgue measure by max⁡{ρ,0}\max\{\rho,0\}.✦ For nonnegative ρ\rho, this is μ=ρdx\mu=\rho\,dx, specified by (5): μ(A)=∫Aρ(x)dxfor every Borel set A⊆ℝ3.\mu(A)=\int_A\rho(x)\,dx \qquad\text{for every Borel set }A\subseteq\mathbb R^3.(5)

Kantorovich attainment for a Borel probability measure μ\mu on ℝ3\mathbb R^3 is the existence of a self-coupling attaining the Kantorovich value E*(μ)E_*(\mu), as in (6): there is π∈Π(μ,μ,μ)with∫c(x,y,z)dπ(x,y,z)=E*(μ).\text{there is }\pi\in\Pi(\mu,\mu,\mu) \quad\text{with}\quad \int c(x,y,z)\,d\pi(x,y,z)=E_*(\mu).(6)✦ Monge nonattainment for a Borel probability measure μ\mu on ℝ3\mathbb R^3 is the strict inequality in (7) for every pair of Borel maps preserving μ\mu: E*(μ)<𝒢μ(T2,T3)for all T2,T3∈𝒯μ.E_*(\mu)<\mathcal G_\mu(T_2,T_3) \qquad\text{for all }T_2,T_3\in\mathcal T_\mu.(7)✦ A Borel probability measure μ\mu on ℝ3\mathbb R^3 admits finite-cost Monge approximation when, for every ϵ>0\epsilon>0, there are maps T2,T3∈𝒯μT_2,T_3\in\mathcal T_\mu satisfying (8): 𝒢μ(T2,T3)<+∞,𝒢μ(T2,T3)≤E*(μ)+ϵ.\mathcal G_\mu(T_2,T_3)<+\infty, \qquad \mathcal G_\mu(T_2,T_3)\le E_*(\mu)+\epsilon.(8)✦

The class 𝒞\mathcal C of Coulomb counterexample densities consists of smooth compact probability densities ρ\rho whose associated measure μ=ρdx\mu=\rho\,dx is a probability measure, has finite Kantorovich value, and satisfies Kantorovich attainment and Monge nonattainment.✦ A density ρ\rho has the full Coulomb conclusion when it belongs to the class ℱ\mathcal F in (9): ℱ={ρ∈𝒞:EM(μ)=E*(μ),μ admits finite-cost Monge approximationwhere μ=ρdx}.\mathcal F= \left\{\rho\in\mathcal C: \begin{array}{l} E_{\mathrm M}(\mu)=E_*(\mu),\\ \mu\text{ admits finite-cost Monge approximation} \end{array} \quad\text{where }\mu=\rho\,dx \right\}.(9)✦

The Claim

Theorem 1 (Smooth Coulomb nonattainment). The class ℱ\mathcal F in (9) is nonempty: ℱ≠⌀.\mathcal F\ne\varnothing.(10)✦ The separated local Coulomb charts and their global certificate supply a density with the full Coulomb conclusion: ∃ρ:ℝ3→ℝ,ρ∈ℱ.\exists\rho:\mathbb R^3\to\mathbb R,\quad \rho\in\mathcal F.(11)✦

In the paper, the result reads: there is a nonnegative density ρ\rho on ℝ3\mathbb R^3, with ∫ρdx=1\int\rho\,dx=1 and ρ,ρ∈Cc∞(ℝ3)\rho,\sqrt{\rho}\in C_c^\infty(\mathbb R^3), such that for μ=ρdx\mu=\rho\,dx the value E*(μ)E_*(\mu) is finite and attained in Π(μ,μ,μ)\Pi(\mu,\mu,\mu), while every pair of Borel maps T2,T3:ℝ3→ℝ3T_2,T_3:\mathbb R^3\to\mathbb R^3 with (T2)#μ=(T3)#μ=μ(T_2)_\#\mu=(T_3)_\#\mu=\mu satisfies ∫c(x,T2(x),T3(x))dμ(x)>E*(μ)\int c(x,T_2(x),T_3(x))\,d\mu(x)>E_*(\mu), allowing an infinite left side.

What Is Checked

The statement (10) concerns real functions ρ\rho on ℝ3\mathbb R^3 and the class ℱ\mathcal F of (9), with extended Coulomb integrals over probability self-couplings and graph costs of Borel maps preserving μ=ρdx\mu=\rho\,dx.