OctoInformBeta

OpenAI preprint, September 2026 · §5. The marginal and exclusion of every Monge optimizer

§5 (setting): transported square amplitudes and the smooth marginal: The five-component probability density and its square root have compact support

Disjoint smooth charts yield a smooth marginal.

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

A square density on a ball in ℝ3\mathbb R^3 supplies a central mass component and four transported components. The desired common marginal has a smooth square root as well as a smooth density.

Throughout, dxdx denotes Lebesgue measure on ℝ3\mathbb R^3, and supp⁡f\operatorname{supp}f denotes the closed support of a function ff. The density measure associated with a real function ff is the measure μf\mu_f defined by (1), with negative values of ff replaced by zero: μf=f+dx,f+(x)=max⁡{f(x),0}.\mu_f=f_+\,dx,\qquad f_+(x)=\max\{f(x),0\}.(1)✦

A normalized smooth amplitude gg on ℝ3\mathbb R^3 is a nonnegative smooth function with compact support and unit squared integral, as in (2): g∈Cc∞(ℝ3),g≥0,∫ℝ3g(x)2dx=1.g\in C_c^\infty(\mathbb R^3),\qquad g\geq0,\qquad \int_{\mathbb R^3}g(x)^2\,dx=1.(2)✦ A smooth compact probability density ρ\rho on ℝ3\mathbb R^3 is a nonnegative real function satisfying (3): ρ,ρ∈Cc∞(ℝ3),∫ℝ3ρ(x)dx=1.\rho,\sqrt{\rho}\in C_c^\infty(\mathbb R^3),\qquad \int_{\mathbb R^3}\rho(x)\,dx=1.(3)✦

A local chart H:U→VH:U\to V is a homeomorphism between open subsets of ℝ3\mathbb R^3. The extension H̃\widetilde H of a local chart H:U→VH:U\to V is the function equal to HH on UU and to 00 outside UU.✦

Definition 1. For real functions g,Jg,J on ℝ3\mathbb R^3, the local transported amplitude AH(g,J):ℝ3→ℝA_H(g,J):\mathbb R^3\to\mathbb R is the function in (4): AH(g,J)(y)={g(H−1(y))J(y),y∈V,0,y∉V.A_H(g,J)(y)= \begin{cases} g(H^{-1}(y))J(y),&y\in V,\\ 0,&y\notin V. \end{cases}(4)✦

Fix a ball B=B(b,r)B=B(b,r) and four local charts Hj:Uj→VjH_j:U_j\to V_j, for 1≤j≤41\leq j\leq4. The index order j=0,…,4j=0,\ldots,4 corresponds to the component order 0,e1,−e1,e2,−e20,e_1,-e_1,e_2,-e_2, where e1,e2e_1,e_2 are coordinate unit vectors. For a central amplitude gg and four real functions qjq_j, the component amplitudes aj:ℝ3→ℝa_j:\mathbb R^3\to\mathbb R, regions Rj⊂ℝ3R_j\subset\mathbb R^3 and positive weights wjw_j are defined by (5): a0=g,R0=B,w0=13,aj=qj,Rj=Hj(B),wj=16(1≤j≤4).\begin{aligned} a_0&=g,&R_0&=B,&w_0&=\frac13,\\ a_j&=q_j,&R_j&=H_j(B),&w_j&=\frac16 \qquad(1\leq j\leq4). \end{aligned}(5)

Definition 2. The five-component marginal density associated with g,q1,…,q4g,q_1,\ldots,q_4 is the real function ρ\rho on ℝ3\mathbb R^3 given by (6): ρ(x)=∑j=04wjaj(x)2=13g(x)2+16∑j=14qj(x)2.\rho(x)=\sum_{j=0}^4w_j a_j(x)^2 =\frac13g(x)^2+\frac16\sum_{j=1}^4q_j(x)^2.(6)✦

The Claim

Theorem 1. Let B=B(b,r)⊂ℝ3B=B(b,r)\subset\mathbb R^3 with r>0r>0. There is a normalized smooth amplitude gg satisfying (7): supp⁡g⊂B.\operatorname{supp}g\subset B.(7)✦

Let H:U→VH:U\to V be a local chart and gg a normalized smooth amplitude with supp⁡g⊂U\operatorname{supp}g\subset U. Suppose H−1H^{-1} is smooth on VV, and let D(y):ℝ3→ℝ3D(y):\mathbb R^3\to\mathbb R^3 be its derivative at y∈Vy\in V. Assume that y↦|det⁡D(y)|y\mapsto\sqrt{|\det D(y)|} is smooth on VV. There is a normalized smooth amplitude qq satisfying (8): supp⁡q⊂H(supp⁡g),H̃#(g2dx)=q2dy.\operatorname{supp}q\subset H(\operatorname{supp}g),\qquad \widetilde H_{\#}(g^2\,dx)=q^2\,dy.(8)✦

Let Hj:Uj→VjH_j:U_j\to V_j, 1≤j≤41\leq j\leq4, be local charts with B⊂UjB\subset U_j. Suppose each HjH_j and Hj−1H_j^{-1} is smooth on its respective domain, and the five regions R0,…,R4R_0,\ldots,R_4 in (5) are pairwise disjoint. There are normalized smooth amplitudes g,q1,…,q4g,q_1,\ldots,q_4 for which (9) holds, with ν=g2dx\nu=g^2\,dx and μ=ρdx\mu=\rho\,dx for the density in (6): supp⁡g⊂B,supp⁡qj⊂Hj(B)(1≤j≤4),ν(ℝ3)=1,ρ is a smooth compact probability density,μ=13ν+16∑j=14(H̃j)#ν.\begin{gathered} \operatorname{supp}g\subset B,\qquad \operatorname{supp}q_j\subset H_j(B)\quad(1\leq j\leq4),\\ \nu(\mathbb R^3)=1,\qquad \rho\text{ is a smooth compact probability density},\\ \mu=\frac13\nu+\frac16\sum_{j=1}^4(\widetilde H_j)_{\#}\nu. \end{gathered}(9)✦

In the paper, the result reads: choose a nonnegative g∈Cc∞(B)g\in C_c^\infty(B) with ∫Bg2dx=1\int_B g^2\,dx=1 and put ν=g2dx\nu=g^2\,dx; for the four charts H∈{Y1,W1,Y2,W2}H\in\{Y_1,W_1,Y_2,W_2\}, extend gH(y)=g(H−1(y))|det⁡DH−1(y)|1/2g_H(y)=g(H^{-1}(y))|\det DH^{-1}(y)|^{1/2} from H(B)H(B) by zero, so that H#ν=gH2dyH_{\#}\nu=g_H^2\,dy and μ=ρdx\mu=\rho\,dx for ρ=g2/3+(1/6)∑HgH2\rho=g^2/3+(1/6)\sum_Hg_H^2; disjoint supports give ρ=g/3+(1/6)∑HgH\sqrt\rho=g/\sqrt3+(1/\sqrt6)\sum_Hg_H, with both ρ\rho and ρ\sqrt\rho smooth and compactly supported and ∫ρdx=1\int\rho\,dx=1.

What Is Checked

The marginal construction is stated over a positive-radius ball contained in the source of each of four local charts smooth in both directions, with the central ball and its four images pairwise disjoint.