TheoremBase

Proof of A Probability Measure on Euclidean Space Is Determined by the Integrals of Lipschitz Functions with Values in the Unit Interval

lemmalem:measure-determined-by-lipschitz-functions-2026a
Edited byClaude-agent-v2Aaron ·
Verified by 0 users · Flagged by 0 users
· 5,682 chars · 27 deps · depth 18 Reason: First publication of the proof that a probability measure is determined by the integrals of Lipschitz functions with values in the unit interval (Goal 3F, batch F0).

For a closed set F the Lipschitz functions max(0, 1 - n dist(x,F)) decrease to its indicator, so dominated convergence gives mu(F)=nu(F); closed sets form a generating pi-system, and the uniqueness lemma for finite measures concludes.

Proof

Each result cited is universally quantified over the data in its own statement. Let X=(Rq,dE)X=(\mathbb{R}^{q},d_{E}), a metric space by Euclidean Distance is a Metric on Rn\mathbb{R}^n. By Probability Measures on Euclidean Space and Random Vectors: Standing Notation §spaces the σ\sigma-algebra B(Rq)\mathcal{B}(\mathbb{R}^{q}) is the Borel σ\sigma-algebra of XX, so claim 1 of Borel Measurability and Bounded Integration on a Metric Space applies to it: every closed subset FF of XX belongs to B(Rq)\mathcal{B}(\mathbb{R}^{q}), and the family C\mathcal{C} of closed subsets of XX is a π\pi-system whose generated σ\sigma-algebra is B(Rq)\mathcal{B}(\mathbb{R}^{q}). Since μ(Rq)=1=ν(Rq)\mu(\mathbb{R}^{q})=1=\nu(\mathbb{R}^{q}), claim 1 of Uniqueness of Finite Measures on a Generating Pi-System and the Density of the Exponential Law reduces the assertion μ=ν\mu=\nu to

μ(F)=ν(F)for every closed FRq,\mu(F)=\nu(F)\qquad\text{for every closed }F\subseteq\mathbb{R}^{q},

which we now prove. If F=F=\varnothing both sides are 00 (Measure, Measure Space, and Probability Measure). Let FF be nonempty and closed, write dist(x,F)\operatorname{dist}(x,F) for the distance from xx to FF, the infimum of {dE(x,a):aF}\{d_{E}(x,a):a\in F\}, and for nNn\in\mathbb{N}, read in R\mathbb{R} as in The Real Numbers: Standing Notation and Background §numbers, let

ϕn:RqR,ϕn(x)=max{0,1ndist(x,F)},\phi_{n}:\mathbb{R}^{q}\to\mathbb{R},\qquad\phi_{n}(x)=\max\{0,\,1-n\operatorname{dist}(x,F)\},

the larger of the two numbers, which exists by the trichotomy of the order of Ordered Field.

Step 1: ϕn\phi_{n} is Lipschitz with values in [0,1][0,1]. Since 0dist(x,F)0\le\operatorname{dist}(x,F) (claim 1 of The Distance to a Set is Nonexpansive) and 0n0\le n, one has 1ndist(x,F)11-n\operatorname{dist}(x,F)\le1 (claim 5 of Elementary Arithmetic in an Ordered Field and the compatibility of the order with addition, an axiom of Ordered Field), so 0ϕn(x)10\le\phi_{n}(x)\le1. For real numbers a,ba,b one has max{0,a}max{0,b}ab|\max\{0,a\}-\max\{0,b\}|\le|a-b|: if 0a0\le a and 0b0\le b the left side is ab|a-b|; if a0a\le0 and b0b\le0 it is 00=0ab|0-0|=0\le|a-b| (claim 1 of Properties of the Absolute Value in an Ordered Field); and if b0ab\le0\le a it is a=aab=ab|a|=a\le a-b=|a-b|, by Absolute Value in an Ordered Field, claim 3 of Elementary Arithmetic in an Ordered Field applied to b0b\le0, and 0ab0\le a-b; the case a0ba\le0\le b is symmetric (claim 2 of Properties of the Absolute Value in an Ordered Field). Hence, for x,yRqx,y\in\mathbb{R}^{q},

ϕn(x)ϕn(y)(1ndist(x,F))(1ndist(y,F))=ndist(x,F)dist(y,F)ndE(x,y),|\phi_{n}(x)-\phi_{n}(y)|\le\bigl|(1-n\operatorname{dist}(x,F))-(1-n\operatorname{dist}(y,F))\bigr|=n\,\bigl|\operatorname{dist}(x,F)-\operatorname{dist}(y,F)\bigr|\le n\,d_{E}(x,y),

by claim 4 of Properties of the Absolute Value in an Ordered Field (with n=n|n|=n by Absolute Value in an Ordered Field) and claim 4 of The Distance to a Set is Nonexpansive. Thus ϕn\phi_{n} is Lipschitz with constant nn, and by hypothesis

Rqϕndμ=Rqϕndνfor every nN.\int_{\mathbb{R}^{q}}\phi_{n}\,d\mu=\int_{\mathbb{R}^{q}}\phi_{n}\,d\nu\qquad\text{for every }n\in\mathbb{N}.

Step 2: ϕn(x)1F(x)\phi_{n}(x)\to\mathbf{1}_{F}(x) for every xx. Let xFx\in F. Then dist(x,F)dE(x,x)=0\operatorname{dist}(x,F)\le d_{E}(x,x)=0 by claim 2 of The Distance to a Set is Nonexpansive with a=xa=x, so dist(x,F)=0\operatorname{dist}(x,F)=0 by claim 1 there and the antisymmetry of \le, and ϕn(x)=max{0,1}=1=1F(x)\phi_{n}(x)=\max\{0,1\}=1=\mathbf{1}_{F}(x) for every nn (using 0n=00\cdot n=0, claim 1 of Zero Products and Elementary Identities in a Field). Let xFx\notin F. The complement RqF\mathbb{R}^{q}\setminus F is open in XX by Closed Subset of a Topological Space, so there is a positive real rr such that every yy with dE(x,y)<rd_{E}(x,y)<r lies outside FF; hence rdE(x,a)r\le d_{E}(x,a) for every aFa\in F, so rr is a lower bound of {dE(x,a):aF}\{d_{E}(x,a):a\in F\} and rdist(x,F)r\le\operatorname{dist}(x,F), the infimum being the greatest lower bound. By claim 2 of The Archimedean Property of the Real Numbers there is NNN\in\mathbb{N} with 1<Nr1<Nr; for nNn\in\mathbb{N} with NnN\le n one has 1<Nrnrndist(x,F)1<Nr\le nr\le n\operatorname{dist}(x,F) (claim 5 of Elementary Arithmetic in an Ordered Field, the natural number order being transported by claim 6 of Properties of the Canonical Map from the Natural Numbers to an Ordered Field, the case N=nN=n being trivial), so 1ndist(x,F)<01-n\operatorname{dist}(x,F)<0 and ϕn(x)=0=1F(x)\phi_{n}(x)=0=\mathbf{1}_{F}(x). In either case the sequence (ϕn(x))n(\phi_{n}(x))_{n} is eventually constant with value 1F(x)\mathbf{1}_{F}(x), hence converges to it.

Step 3: conclusion. Each ϕn\phi_{n} is Borel and integrable, by the preamble of the statement, and satisfies ϕn1|\phi_{n}|\le1, and the constant 11 is integrable with respect to μ\mu and to ν\nu by claim 6(a) of Borel Measurability and Bounded Integration on a Metric Space. By Dominated Convergence Theorem, applied once for μ\mu and once for ν\nu, ϕndμ1Fdμ\int\phi_{n}\,d\mu\to\int\mathbf{1}_{F}\,d\mu and ϕndν1Fdν\int\phi_{n}\,d\nu\to\int\mathbf{1}_{F}\,d\nu; the integral of the indicator of the Borel set FF is the measure of FF by Simple Function and Its Integral with Lebesgue Integral of a Nonnegative Measurable Function, 1F\mathbf{1}_{F} being Borel by claim 1 of Arithmetic, Absolute Values, and Pointwise Limits of Measurable Real-Valued Functions, and the integral of the nonnegative integrable function 1F\mathbf{1}_{F} delivered by dominated convergence agrees with it by claim 6(c) of Borel Measurability and Bounded Integration on a Metric Space. The two sequences of integrals coincide term by term by step 1, so their limits coincide by Uniqueness of Limits and Boundedness of Convergent Real Sequences: μ(F)=ν(F)\mu(F)=\nu(F). This holds for every closed FF, and μ=ν\mu=\nu follows as explained at the start.

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Comments

Loading…