TheoremBase

Ambient covers are traded for covers in the subspace topology and back, picking one ambient open set for each of finitely many traces; finite unions and increasing covers then follow from the cover criterion, and the distance sets are handled by the Lipschitz estimate for the distance and the triangle inequality.

Proof

Throughout, the open subsets of XX are the members of the topology Td\mathcal{T}_{d}, a set U\mathcal{U} of sets covers KK when K⊆⋃UK\subseteq\bigcup\mathcal{U}, and finite sets are as in Finite Sets (i); singletons are finite, unions of two finite sets are finite, unions of finite families of finite sets are finite, and images of finite sets under maps are finite, by Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (iii), Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (v), Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (x) and Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (vi), and segments of N\mathbb{N} are finite by Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (vii); set-builder conditions are predicative by the formula trees of Formula Trees with Values: Syntax, Two Readings, and Definedness.

Covers by open sets. Let K⊆XK\subseteq X, and let TK={K∩U:U∈Td}\mathcal{T}_{K}=\{K\cap U:U\in\mathcal{T}_{d}\} be the subspace topology on KK. By Compact Topological Space and Compact Subset (ii) and Compact Topological Space and Compact Subset (i), KK is compact exactly when every subset of TK\mathcal{T}_{K} covering KK has a finite subset covering KK.

Suppose first that KK is compact, and let U\mathcal{U} be a set of open subsets of XX covering KK. The image W\mathcal{W} of U\mathcal{U} under U↦K∩UU\mapsto K\cap U is a subset of TK\mathcal{T}_{K} covering KK, so some finite W0⊆W\mathcal{W}_{0}\subseteq\mathcal{W} covers KK. For each W∈W0W\in\mathcal{W}_{0} the set {U∈U:W=K∩U}\{U\in\mathcal{U}:W=K\cap U\} is nonempty, so Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (xi) gives a map cc on W0\mathcal{W}_{0} with c(W)∈Uc(W)\in\mathcal{U} and W=K∩c(W)W=K\cap c(W) for every W∈W0W\in\mathcal{W}_{0}. The image V=c(W0)\mathcal{V}=c(\mathcal{W}_{0}) is a finite subset of U\mathcal{U}, and it covers KK: each x∈Kx\in K lies in some W∈W0W\in\mathcal{W}_{0}, and W⊆c(W)W\subseteq c(W).

Conversely, suppose that every set of open subsets of XX covering KK has a finite subset covering KK, and let W⊆TK\mathcal{W}\subseteq\mathcal{T}_{K} cover KK. Put U={U∈Td:K∩U∈W}\mathcal{U}=\{U\in\mathcal{T}_{d}:K\cap U\in\mathcal{W}\}. Every W∈WW\in\mathcal{W} has the form K∩UK\cap U with U∈TdU\in\mathcal{T}_{d}, and such a UU lies in U\mathcal{U}; hence U\mathcal{U} covers KK, and some finite V⊆U\mathcal{V}\subseteq\mathcal{U} covers KK. The image of V\mathcal{V} under U↦K∩UU\mapsto K\cap U is then a finite subset of W\mathcal{W} covering KK. So KK is compact.

Finite unions. Let U\mathcal{U} be a set of open subsets of XX covering ⋃K\bigcup\mathcal{K}. For every L∈KL\in\mathcal{K}, U\mathcal{U} covers the compact set LL, so by (i) the set SLS_{L} of finite subsets of U\mathcal{U} covering LL is nonempty. By Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (xi) there is a map cc on K\mathcal{K} with c(L)∈SLc(L)\in S_{L} for every L∈KL\in\mathcal{K}. The set V=⋃L∈Kc(L)\mathcal{V}=\bigcup_{L\in\mathcal{K}}c(L) is finite, it is contained in U\mathcal{U}, and it covers every L∈KL\in\mathcal{K}, hence ⋃K\bigcup\mathcal{K}. By (i), ⋃K\bigcup\mathcal{K} is compact.

Increasing covers. We have Um⊆Um′U_{m}\subseteq U_{m'} whenever m≤m′m\le m' (by induction on m′m'). The image U={Um:m∈N}\mathcal{U}=\{U_{m}:m\in\mathbb{N}\} of N\mathbb{N} covers KK, so by (i) some finite V⊆U\mathcal{V}\subseteq\mathcal{U} covers KK. For V∈VV\in\mathcal{V} let μ(V)\mu(V) be the least m∈Nm\in\mathbb{N} with V=UmV=U_{m}, which exists by Arithmetic and Order of the Natural Numbers (xv). The set μ(V)∪{1}\mu(\mathcal{V})\cup\{1\} is a nonempty finite subset of N\mathbb{N}; the element 11 is added because V\mathcal{V}, and with it μ(V)\mu(\mathcal{V}), may be empty. So it has a greatest element MM by Finite Sets: the Pigeonhole Principle, Uniqueness of the Length, Subsets, Unions, Products, Images, Bounded Sets of Natural Numbers, Extreme Elements, Sets of Maps, Finite Unions and Finite Choice (viii). Every V∈VV\in\mathcal{V} equals Uμ(V)⊆UMU_{\mu(V)}\subseteq U_{M}, and therefore

K⊆⋃V⊆UM.K\subseteq\bigcup\mathcal{V}\subseteq U_{M}.

Sublevel sets of the distance. Put O={x∈X:dist⁡(x,A)<δ}O=\{x\in X:\operatorname{dist}(x,A)<\delta\} and N={x∈X:dist⁡(x,A)≤δ}N=\{x\in X:\operatorname{dist}(x,A)\le\delta\}. We use the Lipschitz estimate ∣dist⁡(x,A)−dist⁡(y,A)∣≤d(x,y)|\operatorname{dist}(x,A)-\operatorname{dist}(y,A)|\le d(x,y) for x,y∈Xx,y\in X.

Let x∈Ox\in O and r=δ−dist⁡(x,A)>0r=\delta-\operatorname{dist}(x,A)>0. For yy in the open ball B(x,r)B(x,r) we have dist⁡(y,A)≤dist⁡(x,A)+d(x,y)<δ\operatorname{dist}(y,A)\le\operatorname{dist}(x,A)+d(x,y)<\delta, so B(x,r)⊆OB(x,r)\subseteq O. Hence OO is open, which is (v).

Let x∈X∖Nx\in X\setminus N and r=dist⁡(x,A)−δ>0r=\operatorname{dist}(x,A)-\delta>0. For y∈B(x,r)y\in B(x,r) we have dist⁡(y,A)≥dist⁡(x,A)−d(x,y)>δ\operatorname{dist}(y,A)\ge\operatorname{dist}(x,A)-d(x,y)>\delta, so B(x,r)⊆X∖NB(x,r)\subseteq X\setminus N. Hence X∖NX\setminus N is open and NN is closed, which is (vi).

Closed balls. Let x∈Xx\in X, r≥0r\ge0 and y∈X∖Bˉ(x,r)y\in X\setminus\bar{B}(x,r), so that s=d(x,y)−r>0s=d(x,y)-r>0. For z∈B(y,s)z\in B(y,s) the triangle inequality gives d(x,z)≥d(x,y)−d(y,z)>rd(x,z)\ge d(x,y)-d(y,z)>r, so z∉Bˉ(x,r)z\notin\bar{B}(x,r). Thus X∖Bˉ(x,r)X\setminus\bar{B}(x,r) is open, and Bˉ(x,r)\bar{B}(x,r) is closed. This proves (iv).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…