CompleteMetricSpaces PDF

Cauchy Sequences and Complete Metric Spaces
Let’s first consider two examples of convergent sequences in R:
1
√
Example 1: Let xn = n
2 for each n ∈ N. Note that each xn is an irrational number (i.e.,
xn ∈ Qc ) and that {xn } converges to 0. Thus, {xn } converges in R (i.e., to an element of R). But
0 is a rational number (thus, 0 6∈ Qc ), so although the sequence {xn } is entirely in Qc , it does not
converge in Qc , in spite of being well-behaved in the sense that it converges in R.
Example 2: Let x1 ∈ N, and let xn be the sequence defined by xn+1 = 12 xn + x1n for each n ∈ N.
√
We can show that the sequence {xn } converges to 2. (You’ll be asked to do that in an exercise
√
below.) Then {xn } is a sequence of rational numbers that converges to the irrational number 2
√
— i.e., each xn is in Q and lim{xn } = 2 6∈ Q. Thus, in a parallel to Example 1, {xn } here
converges in R but does not converge in Q.
Examples 1 and 2 demonstrate that both the set Qc of irrational numbers and the set Q of
rational numbers are not entirely well-behaved metric spaces: there are well-behaved sequences in
each space that don’t converge to an element of the space. The sequences are well-behaved in the
sense that they do converge in R. The following definition formalizes this idea of a well-behaved
sequence in a metric space (such as Q and Qc ), but without requiring any reference to some other,
larger metric space (such as R).
Definition: A sequence {xn } in a metric space (X, d) is Cauchy if

∀ > 0 : ∃n ∈ N : m, n > n ⇒ d(xm , xn ) < .
Remark: Convergent sequences are Cauchy.

Proof:
Let {xn } → x, let > 0, let n be such that n > n ⇒ d(xn , x) < /2, and let m, n > n. Then

d(xm , x) < and d(xn , x) < ,
2 2
and the Triangle Inequality yields

d(xm , xn ) 5 d(xm , x) + d(xn , x) < + = .
2 2
1
Exercise: The real sequence {xn } defined by xn = n
converges, so it’s Cauchy. Prove directly
that it’s Cauchy, by showing how the n in the definition depends upon .
Definition: A metric space (X, d) is complete if every Cauchy sequence in X converges in X

(i.e., to a limit that’s in X).
Example 3: The real interval (0, 1) with the usual metric is not a complete space: the sequence
1
xn = n
is Cauchy but does not converge to an element of (0, 1).
Example 4: The space Rn with the usual (Euclidean) metric is complete. We haven’t shown this
yet, but we’ll do so momentarily.
Remark 1: Every Cauchy sequence in a metric space is bounded.

Proof: Exercise.
Remark 2: If a Cauchy sequence has a subsequence that converges to x, then the sequence
converges to x.
Proof: Exercise.
In order to prove that R is a complete metric space, we’ll make use of the following result:
Proposition: Every sequence of real numbers has a monotone subsequence.

Proof: Suppose the sequence {xn } has no monotone increasing subsequence; we show that then
it must have a monotone decreasing subsequence. The sequence {xn } must have a first term, say
xn1 , such that all subsequent terms are at least as small (i.e., n > n1 ⇒ xn 5 xn1 ); otherwise
{xn } would have a monotone increasing subsequence. Similarly, the subsequence {xn1 +1 , xn1 +2 , . . . }
must have a first term xn2 such that all subsequent terms are at least as small; note that xn1 = xn2 .
Continuing for n1 , n2 , n3 , . . . , we have a subsequence {xnk } such that xn1 = xn2 = xn3 = . . ., a
monotone decreasing subsequence.
Now we’ll prove that R is a complete metric space, and then use that fact to prove that the
Euclidean space Rn is complete.
Theorem: R is a complete metric space — i.e., every Cauchy sequence of real numbers converges.
Proof: Let {xn } be a Cauchy sequence. Remark 1 ensures that the sequence is bounded, and
therefore that every subsequence is bounded. The proposition we just proved ensures that the
sequence has a monotone subsequence. The Monotone Convergence Theorem ensures that this
bounded monotone subsequence converges. And therefore Remark 2 ensures that the original
sequence converges.
This proof used the Completeness Axiom of the real numbers — that R has the LUB Property
— via the Monotone Convergence Theorem. We could have gone instead in the other direction:
taking “every Cauchy sequence of real numbers converges” to be the Completeness Axiom, and
then proving that R has the LUB Property.
2
Theorem: The normed vector space Rn is a complete metric space.
Proof: Exercise.
Example 5: The closed unit interval [0, 1] is a complete metric space (under the absolute-value
metric). This is easy to prove, using the fact that R is complete.
Example 6: The space C[0, 1] is complete. (We haven’t shown this yet.)
Exercise: In a previous exercise set we worked with a sequence of distribution functions Fn defined
by (
nx, if x 5 n1
Fn (x) =
1 , if x = n1 .
on the unit interval [0, 1] in R. We showed that {Fn } does not converge in C[0, 1]. Therefore, if
{Fn } were Cauchy, C[0, 1] would not be complete. Verify that {Fn } is not Cauchy.
Example 7: (Obtaining R as the completion of Q.)

Let S be the set of Cauchy sequences in Q — i.e., the set of Cauchy sequences of rational numbers
— with the usual metric. Define a relation ∼ on S as follows:
{xn } ∼ {x0n } if ∀ > 0 : ∃n̄ ∈ N : m, n > n̄ ⇒ d(xm , x0n ) <
Let Q∗ = S/ ∼, the partition of S consisting of equivalence classes of Cauchy sequences. Define
the distance function d∗ for Q∗ as follows:
For any x, x0 ∈ Q∗ , let {xn } ∈ x and {x0n } ∈ x0 (i.e., x = [{xn }] and x0 = [{x0n }]).
Then define d∗ (x, x0 ) by d∗ (x, x0 ) = limn→∞ d(xn , x0n ).
It’s pretty straightforward to show that d∗ is well-defined and is a metric for Q∗ . The metric
space (Q∗ , d∗ ) can be placed into one-to-one correspondence with (R, | · |), each constant sequence
{r, r, r, . . .} of rationals corresponding to the rational number r ∈ Q ⊆ R. The set Q∗ is one way
of defining R.
Exercise: Verify that the relation ∼ defined in Example 7 is an equivalence relation.
Example 8: (The Completion of a Metric Space)

Let (X, d) be a metric space that is not complete. Just as in Example 7, let S be the set of Cauchy
sequences in X; define the equivalence relation ∼ in the same way, and let X ∗ be the quotient
space S/ ∼; and define d∗ on the quotient space X ∗ in the same way as in Example 7. Then we
can show, just as in Example 7, that d∗ is well-defined and is a metric for X ∗ ; that (X ∗ , d∗ ) is a
complete metric space; and that X corresponds to a subset of X ∗ — we say that X is embedded
in X ∗ . The complete metric space (X ∗ , d∗ ) is called the completion of (X, d).
3
Example 9: The open unit interval (0, 1) in R, with the usual metric, is an incomplete metric
space. What is its completion, ((0, 1)∗ , d∗ ))?
Theorem: A subset of a complete metric space is itself a complete metric space if and only if it
is closed.
Proof: Exercise.
Recall that every normed vector space is a metric space, with the metric d(x, x0 ) = kx − x0 k.
Therefore our definition of a complete metric space applies to normed vector spaces: an n.v.s. is
complete if it’s complete as a metric space, i.e., if all Cauchy sequences converge to elements of
the n.v.s.
Definition: A complete normed vector space is called a Banach space.
Example 4 revisited: Rn with the Euclidean norm is a Banach space.
Example 5 revisited: The unit interval [0, 1] is a complete metric space, but it’s not a Banach
space because it’s not a vector space.
Example 6 revisited: C[0, 1] is a Banach space.
Exercise: Let a and x1 be positive real numbers, and let {xn } be the sequence defined by
1 √
xn+1 = x
2 n
+ 2xan for each n ∈ N. Verify that {xn } converges to a. Hint: You’ll probably find
it helpful to remember that the convergent sequences comprise a vector subspace of the vector
space R∞ of all real sequences, and to remember the algebra of limits of sequences: lim{xn + yn } =
lim{xn } + lim{yn }, etc. But in order for you to use these algebraic properties you also need to
know that the sequences in question actually do converge. For that you might want to use the
Monotone Convergence Theorem.
Sequences defined recursively, like the sequence in the above exercise, are important in economics.
We’ll see sequences like this later in this course when we study fixed point theorems and their
application to the Nash equilibria of games and to growth theory. They’ll appear in Economics
501B when we study computation of market equilibria and convergence to equilibrium.

CompleteMetricSpaces PDF

Uploaded by

Copyright:

Available Formats

CompleteMetricSpaces PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CompleteMetricSpaces PDF

Uploaded by

Copyright:

Available Formats

Cauchy Sequences and Complete Metric Spaces

Let’s first consider two examples of convergent sequences in R:

Definition: A sequence {xn } in a metric space (X, d) is Cauchy if

Remark: Convergent sequences are Cauchy.

Definition: A metric space (X, d) is complete if every Cauchy sequence in X converges in X

Remark 1: Every Cauchy sequence in a metric space is bounded.

Proposition: Every sequence of real numbers has a monotone subsequence.

Example 7: (Obtaining R as the completion of Q.)

Exercise: Verify that the relation ∼ defined in Example 7 is an equivalence relation.

Example 8: (The Completion of a Metric Space)

Definition: A complete normed vector space is called a Banach space.

Example 4 revisited: Rn with the Euclidean norm is a Banach space.

Example 6 revisited: C[0, 1] is a Banach space.

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.