23. The Lebesgue Number of a Covering

(1875-1941)

Review:

23.1 Definition:

Given a Metric Space $\QTR{Large}{(M,d)}$ , and a subset MATH we say $\QTR{Large}{x}$ is a limit point of $\QTR{Large}{A\ }$if

          ___________

MATH

That is $\QTR{Large}{x\ }$is in the closure of MATH

Notes:

  1. The definition does not claim that MATH

  2. It is not necessarily the case that the set of limit points of $\QTR{Large}{A\ }$is the closure of $\QTR{Large}{A}$. For example, a singleton set $\QTR{Large}{\{x\}}$ has no limit points but is its own closure.

  3. Remember, $\QTR{Large}{x\ }$is in the closure of MATH means that for any MATH0

    MATH

 

23.2 Definitions:

  1. A sequence in $\QTR{Large}{M}$ is just a map MATH Notationally MATH

  2. Given a one to one order preserving map MATH, the map MATH

    is called a subsequence. We will usually just write " MATH is a subsequence."

  3. A covering MATHof a Set M is a collection of subsets such that MATH.

------------------------------------------------------------------------------------

23.3 Definition:

A Metric Space, $\QTR{Large}{(M,d)}$ is called

  1. limit compact if every infinite subset MATH has a limit point.

  2. sequentially compact if Every sequence in $\QTR{Large}{M}$ has a convergent subsequence. That is given MATH MATH we can find a subsequence MATH

  3. countably compact if every covering MATH by a countable number of open sets , MATH, contains a finite subcover. That is there is some finite subset MATH such that MATH

  4. "totally"(my term) compact if every covering MATH by a open sets contains a finite subcover.

 

23.4 Theorem:

For Metric Spaces these four forms of compactness are equivalent.

In General Topology these four forms of compactness are not equivalent.

Proof:

1. MATH2. This is 21.4.

We will not present counter-examples to the general theorem.

 

The remainder of this Page is devoted to completing the proofs of the various equivalences.

-------------------------------------------------------------------------------------

Review (Continued)

21.4.1 Corollary:

Every sequentially/limit compact Metric Space is complete:

 

21.5 Theorem:

Given a sequentially/limit compact Metric Spaces $\QTR{Large}{(M,d)}$ :

  1. For any MATH0 and some nMATH there exists points MATH MATH

    such that MATH.

  2. In particular, M is bounded in the sense that there exists some bMATH0 such that MATHb for any MATH .

 

21.6 Theorem(Restated):

Every sequentially/limit compact metric space has a countable dense subset.

---------------------------------------------------------------------------------------

23.5 Theorem: (Lebesgue Number)

Given a sequentially/limit compact Metric Space $\QTR{Large}{(M,d)}$ and a covering MATH by open sets

( MATH), there exists a real number MATH such that every open ball of radius MATH is contained in some element of MATH.

The number MATH is called a Lebesgue number for the covering.

 

Proof:

 

Suppose that no Lebesgue number existed. Then there exists an open cover MATH such that for all MATH there exists an MATH such that no MATH contains MATH. In particular for each n we can choose a sequence MATHsuch that

MATHfor any MATH .

The proof continues in the same way as several of the proofs on Page 21:

Since $\QTR{Large}{M}$ is compact choose a convergent subsequence MATH with MATH for some MATH. Since MATH is an open cover, we know there is some MATH and some MATH with MATH. Again, as before, choose n such that MATH and MATH for i $\QTR{Large}{>\ }$n. Check that MATH

 

23.6 Corollary:

Every open covering, MATH of a compact Metric Space $\QTR{Large}{(M,d)}$ has a finite subcovering. In particular 1. MATH2.MATH4.MATH3.

Proof:

Let MATH be a Lebesgue number for MATH. Applying 21.5, we know there is some

nMATH and points MATH MATHsuch that MATH. But since MATH is a Lebesgue number for each i we can find MATH such that MATH and hence

MATH

 

23.4 Theorem:

Let $\QTR{Large}{(M,d)}$ be countabily compact, then every infinite set contains a limit point. In particular, 3.MATH1. Hence, 1. MATH2. MATH3. MATH4.

Proof:

Let $\QTR{Large}{A}$ be an infinite subset of $\QTR{Large}{M}$ without limit points, then . Without loss of generality we can assume that $\QTR{Large}{A}$ is countable. Writing MATH , and MATH $\QTR{Large}{M-A}$ is open since $\QTR{Large}{A}$ has no limit points. More generally MATH is open. One notes that,

  1. MATH

    and

  2. There is no finite subset that covers $\QTR{Large}{M}$ . In particular, MATH