TheoremBase

Proof

Throughout, X∖SX\setminus S is the complement relative to XX of S⊆XS\subseteq X; the finite-subcover criterion is condition 2 of Compact Subset Criterion via Open Covers in the Ambient Space, which characterises compactness by open covers in the ambient space; claims of Elementary Order Arithmetic in an Ordered Field (below, the order arithmetic lemma) are cited by number; and Td\mathcal{T}_d contains the empty set and is closed under arbitrary unions, being a topology.

Claim 1. Since CC is closed in (X,Td)(X,\mathcal{T}_d), the set X∖CX\setminus C belongs to Td\mathcal{T}_d. Let (Ui)i∈I(U_i)_{i\in I} be an open cover of K∩CK\cap C in XX; we verify the finite-subcover criterion for K∩CK\cap C.

If II is empty, then the union of the family is empty, so K∩CK\cap C is empty and the subfamily indexed by the empty subset of II is a subcover. Assume therefore that II is nonempty.

For i∈Ii\in I set Wi=Ui∪(X∖C)W_i=U_i\cup(X\setminus C). Each WiW_i is a union of members of Td\mathcal{T}_d and hence belongs to Td\mathcal{T}_d. The family (Wi)i∈I(W_i)_{i\in I} is an open cover of KK: let x∈Kx\in K. If x∈Cx\in C, then x∈K∩Cx\in K\cap C, so x∈Uix\in U_i for some i∈Ii\in I and therefore x∈Wix\in W_i. If x∉Cx\notin C, then x∈X∖Cx\in X\setminus C, and since II is nonempty we may pick any i∈Ii\in I and get x∈Wix\in W_i.

As KK is compact in XX, the finite-subcover criterion gives a finite subset J⊆IJ\subseteq I with K⊆⋃j∈JWjK\subseteq\bigcup_{j\in J}W_j. Let x∈K∩Cx\in K\cap C. Then x∈Wjx\in W_j for some j∈Jj\in J; since x∈Cx\in C we have x∉X∖Cx\notin X\setminus C, and therefore x∈Ujx\in U_j. Hence K∩C⊆⋃j∈JUjK\cap C\subseteq\bigcup_{j\in J}U_j, so (Uj)j∈J(U_j)_{j\in J} is a subcover and K∩CK\cap C is compact in XX.

Claim 2. Write S={x∈K:c≤w(x)}S=\{x\in K: c\le w(x)\} and let (Ui)i∈I(U_i)_{i\in I} be an open cover of SS in XX.

If SS is empty, the subfamily indexed by the empty subset of II is a subcover and we are done. Assume therefore that SS is nonempty; then II is nonempty, since a point of SS lies in some UiU_i.

Let PP be the set of all ordered pairs (z,r)(z,r) such that z∈Kz\in K, r∈Rr\in\mathbb{R}, 0<r0<r, and every y∈Ky\in K with d(z,y)<rd(z,y)<r satisfies w(y)<cw(y)<c. For (z,r)∈P(z,r)\in P let B(z,r)B(z,r) be the open ball of centre zz and radius rr, which belongs to Td\mathcal{T}_d by Open Ball in a Metric Space is Open. Let

G=⋃(z,r)∈PB(z,r),G=\bigcup_{(z,r)\in P}B(z,r),

which belongs to Td\mathcal{T}_d, being a union of members of Td\mathcal{T}_d indexed by PP (and being the empty set, which belongs to Td\mathcal{T}_d, when PP is empty). No choice of a radius at each point is made here: PP is the set of all admissible pairs.

For i∈Ii\in I set Wi=Ui∪GW_i=U_i\cup G, which belongs to Td\mathcal{T}_d. The family (Wi)i∈I(W_i)_{i\in I} is an open cover of KK: let x∈Kx\in K. If c≤w(x)c\le w(x), then x∈Sx\in S, so x∈Ui⊆Wix\in U_i\subseteq W_i for some i∈Ii\in I. Otherwise w(x)<cw(x)<c; adding −w(x)-w(x) to both sides and using claim 1 of the order arithmetic lemma gives 0<c−w(x)0<c-w(x), so upper semicontinuity of ww at xx relative to KK, applied with ε=c−w(x)\varepsilon=c-w(x), provides δ∈R\delta\in\mathbb{R} with 0<δ0<\delta such that every y∈Ky\in K with d(x,y)<δd(x,y)<\delta satisfies

w(y)<w(x)+(c−w(x))=c.w(y)<w(x)+(c-w(x))=c .

Thus (x,δ)∈P(x,\delta)\in P. Since d(x,x)=0<δd(x,x)=0<\delta by the definition of a metric, x∈B(x,δ)⊆Gx\in B(x,\delta)\subseteq G, and hence x∈Wix\in W_i for every i∈Ii\in I; as II is nonempty, xx is covered.

Since KK is compact in XX, there is a finite subset J⊆IJ\subseteq I with K⊆⋃j∈JWjK\subseteq\bigcup_{j\in J}W_j. The sets SS and GG are disjoint: if x∈S∩Gx\in S\cap G, then x∈B(z,r)x\in B(z,r) for some (z,r)∈P(z,r)\in P, so d(z,x)<rd(z,x)<r, and x∈Kx\in K because x∈Sx\in S; the defining property of PP then gives w(x)<cw(x)<c, contradicting c≤w(x)c\le w(x).

Now let x∈Sx\in S. Then x∈Kx\in K, so x∈Wj=Uj∪Gx\in W_j=U_j\cup G for some j∈Jj\in J; since x∉Gx\notin G, we get x∈Ujx\in U_j. Hence S⊆⋃j∈JUjS\subseteq\bigcup_{j\in J}U_j, so (Uj)j∈J(U_j)_{j\in J} is a subcover and SS is compact in XX.

Claim 3. Let −w:K→R-w:K\to\mathbb{R} be the function whose value at x∈Kx\in K is −w(x)-w(x). By claim 1 of Negation, Restriction, and Separated Differences of Semicontinuous Functions, applied at every point of KK, the function −w-w is upper semicontinuous on KK. By claim 2 of the present lemma, applied to −w-w and to the real number −c-c, the set

{x∈K: −c≤−w(x)}\{x\in K:\ -c\le -w(x)\}

is compact in XX. Finally, for x∈Kx\in K the inequality −c≤−w(x)-c\le -w(x) holds if and only if w(x)≤cw(x)\le c, by claim 4 of the order arithmetic lemma; hence this set is {x∈K:w(x)≤c}\{x\in K: w(x)\le c\}, which is therefore compact in XX.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…