10.1 Definitions[2CC]

A metric space is a pair \((X,d)\) where \(X\) is a set (nonempty) with associated distance \(d\).

Definition 275

[0MS] A distance is a function \(d:X× X→ [0,∞)\) that enjoys the following properties:

  • \(d(x,x)=0\);

  • (separation property) if \(d(x,y)=0\) then \(x=y\);

  • (symmetry) \(d(x,y)=d(y,x)\) for each \(x,y∈ X\);

  • (triangle inequality) \(d(x,z)≤ d(x,y)+d(y,z)\) for each \(x,y,z∈ X\).

An example is \(ℝ^ n\) with the Euclidean distance \(d(x,y)=|x-y|\).

Definition 276

[0MT] Given a sequence \((x_ n)_ n⊆ X\) and \(x∈ X\),

  • we will say that ”\((x_ n)_ n\) converges to \(x\)” if \(\lim _ n d(x_ n,x)=0\); we will also write \(x_ n→_ n x\) to indicate that the sequence converges to \(x\).

  • We will say that ”\((x_ n)_ n\) is a Cauchy sequence” if

    \[ ∀ \varepsilon {\gt}0~ ~ ∃ N∈ℕ~ ,~ ∀ n,m≥ N~ ~ d(x_ n,x_ m){\lt}\varepsilon ~ ~ . \]
Example 277

[2C1]To any given set \(X\) we may associate the discrete distance

\[ d(x,y) = \begin{cases} 0 & x=y\\ 1 & x\neq y \end{cases} \]

The induced topology is the discrete topology where every subset of \(X\) is an open set.

[Note. If you are not familiar with the concept of metric space, you can assume that \(X=ℝ^ n\) and \(d(x,y)=|x-y|\) in all exercises.]

E279

[0MV]Prove that a converging sequence \((x_ n)_ n⊆ X\) is Cauchy.

E279

[0MW]Given a sequence \((x_ n)_ n⊆ X\) show that, if it converges to \(x\) and converges to \(y\), then \(x=y\).

This result is known as Theorem of the uniqueness of the limit.

E279

[0MX]We generalize the definition of metric space assuming that \(d:X→[0,∞]\) (the other axioms are the same). Show that the relation \(x∼ y\) defined by

\[ x∼ y\iff d(x,y){\lt}∞ \]

is an equivalence relation, and that equivalence classes are open, and therefore are disconnected from each other.

Hidden solution: [UNACCESSIBLE UUID ’0MY’]

E279

[0MZ] Given \(f,g\) continuous functions on \(ℝ\), we define

\[ d(f,g)=\sup _{x∈ℝ}|f(x)-g(x)|\ . \]

Prove that \(d\) is a distance on \(X=C(ℝ)\), in the extended sense of the exercise 3.

Let \(f∼ g\iff d(f,g){\lt}∞\) as before, show that the family of equivalence classes \(\frac X∼\) has the cardinality of the continuum.

Hidden solution: [UNACCESSIBLE UUID ’0N0’]

E279

[0N1] Prerequisites:15.Note:See also eserc. 2. Suppose \(𝜑:[0,∞)→[0,∞)\) is monotonic weakly increasing and subadditive, i.e. \(𝜑(t)+𝜑(s)≥ 𝜑(t+s)\) for each \(t,s≥ 0\); and suppose that \(𝜑(x)=0\) if and only if \(x=0\).

Then \(𝜑◦ d\) is again a distance. Examples: \(𝜑(t)=\sqrt t\), \(𝜑(t)=t/(1+t)\), \(𝜑(t)=\arctan (t)\), \(𝜑(t)=\min \{ t,1\} \).

Moreover show that if \(𝜑\) is continuous in zero then the associated topology is the same.  1 Hidden solution: [UNACCESSIBLE UUID ’0N2’]

E279

[0N3] If \((x_ n)_ n⊂ X\) is a sequence and \(x∈ X\), show that \(\lim _{n→∞} x_ n=x\) if and only if, for each sub–sequence \(n_ k\) there exists a sub–sub–sequence \(n_{k_ h}\) such that \(\lim _{h→∞} x_{n_{k_ h}}=x\). Hidden solution: [UNACCESSIBLE UUID ’0N4’]

E279

[0N5]A sequence \((x_ n)⊂ X\) is a Cauchy sequence if and only if

\[ \lim _{N→∞}\sup \{ d(x_ n,x_ m) : n≥ N, m≥ N\} =0~ ~ . \]

E279

[0N6] A sequence \((x_ n)⊂ X\) is a Cauchy sequence if and only if there exists a sequence \(\varepsilon _ n\) with \(\varepsilon _ n≥ 0\) and \(\varepsilon _ n→_ n 0\) such that, for every \(n\) and every \(m≥ n\), we have \(d(x_ n,x_ m)≤ \varepsilon _ n\).

Hidden solution: [UNACCESSIBLE UUID ’0N7’]

E279

[0N8] If \((x_ n)⊂ X\) is a Cauchy sequence and there exists \(x\) and a subsequence \(n_ m\) such that \(\lim _{m→∞} x_{n_ m}=x\) then \(\lim _{n→∞} x_{n}=x\).

Hidden solution: [UNACCESSIBLE UUID ’0N9’]

This ”lemma” is used in some important proofs, e.g. to show that a compact metric space is also complete.

E279

[0NC] Let \(\varepsilon _ n{\gt}0\) be an infinitesimal decreasing sequence. If \((x_ n)⊂ X\) is a Cauchy sequence, there exists a subsequence \(n_ k\) such that

\[ ∀ k ∈ℕ,~ ∀ h ∈ℕ,~ h{\gt}k ⇒ d(x_{n_ k},x_{n_ h})≤ \varepsilon _ k~ ~ . \]

Hidden solution: [UNACCESSIBLE UUID ’0ND’] This property is often used by choosing \(\varepsilon _ n=2^{-n}\), or other sequence whose series converges.

E279

[0NF] Let \((x_ n)_ n\) be a sequence such that \(∑_{n=1}^∞ d(x_ n,x_{n+1}) {\lt} ∞\): prove that it is a Cauchy sequence.

Compare this exercise, the previous 10 in case \(∑_ n \varepsilon _ n{\lt}∞\), and exercise 9.

E279

[0NG]If \((x_ n)⊂ X\) is a Cauchy sequence, \((y_ n)⊂ X\) is another sequence, and \(d(x_ n,y_ n)→_ n 0\), then \((y_ n)⊂ X\) is a Cauchy sequence.

E279

[0NH] Given \((X,d)\) a metric space, show that \(d\) is continuous (as a function \(d:X× X→ℝ\)). You can actually show that it is Lipschitz, by associating to \(X× X\) the distance

\[ \hat d (x,y) = d(x_ 1,y_ 1) + d(x_ 2,y_ 2), \text{ for } x=(x_ 1,x_ 2),y=(y_ 1,y_ 2) \in X× X~ . \]

Hidden solution: [UNACCESSIBLE UUID ’0NK’] [UNACCESSIBLE UUID ’0NJ’]

[0NM]Prerequisites:3, 2,8.Difficulty:*.Note:Exercise 2, written exam, 9 July 2011.

Let \(𝛼(x)\) be a continuous function on \(ℝ\), bounded and strictly positive. Given \(f,g\) continuous functions on \(ℝ\), we define

\[ d(f,g)=\sup _{x∈ℝ}\big(\min \{ 𝛼(x),|f(x)-g(x)|\} \big)\ . \]

Prove that \(d\) is a distance on \(C(ℝ)\) and that \(\big(C(ℝ),d\big)\) is complete. Hidden solution: [UNACCESSIBLE UUID ’0NP’] [0NQ] Note:Exercise 2, written exam, 25 March 2017.

Show that the following properties are equivalent for a metric space \(X\):

  • every sequence of elements of \(X\) admits a Cauchy subsequence;

  • The completion \(X^*\) of \(X\) is compact.

Hidden solution: [UNACCESSIBLE UUID ’0NR’]

[UNACCESSIBLE UUID ’0NS’] [UNACCESSIBLE UUID ’0NT’] [UNACCESSIBLE UUID ’0NV’]

  1. See Sec. 10.2 below for a summary of definitions regarding topology in metric spaces: in particular the result 15 will be useful.