9.2 Topology in metric spaces[2C2]
Let \((X,d)\) be a metric space.
[0NW] Let \(x∈ X, r{\gt}0\) be given; we will indicate with \(B(x,r)\) the ball,
that is also indicated with \(B_ r(x)\); and with
the disk, that is also indicated with \({\overline B}_ r(x)\).
[0NX] For the following exercises we define that
- a set \(E\) is open if \begin{equation} \label{eq:defaperti} ∀ x_ 0 ∈ E , ∃ r>0: B(x_ 0,r)⊆ E\quad . \end{equation}283- It is easily seen that \(∅,X\) are open sets; that the intersection of a finite number of open sets is an open set; that the union of an arbitrary number of open set is an open set. So these open sets form a topology. 
- The interior \({{E}^\circ }\) of a set \(E\) is \begin{equation} \label{eq:internooperativo} {{E}^\circ }=\bigl\{ x∈ E: \exists r>0, B_ r(x)⊆ E \bigr\} \ ; \end{equation}284- It is easy to verify that \({{E}^\circ }⊆ E\), and that \(E\) is open if and only if \({{E}^\circ }=E\) (exercise 7). 
- A point \(x_ 0∈ X\) is adherent to \(E\) if \[ ∀\, r{\gt}0\ ,\quad E∩ B_ r(x_ 0)≠∅ \quad . \]
- The closure \(\overline E\) of \(E\) is the set of adherent points; it is easy to verify that \({E}⊆ \overline{E}\); It is shown that \(\overline E=E\) if and only if \(E\) is closed (exercise 11). 
- \(A\) is dense in \(B\) if \(\overline A ⊇ B\), that is, if for every \(x ∈ B\) and for every \(r {\gt} 0\) the intersection \(B_ r (x) ∩ A\) is not empty. 
Note that, having the operational definition 283 of ”open set”, then the axioms (in the definition 244) in this case become theorems.
- E284
- [0NZ] Topics:balls. - Prove that \begin{equation} \label{eq:inclpalle} B_{𝜌}(x)⊆ B_ r(x_ 0) \end{equation}285- for every \(x∈ B_𝜌(x_ 0)\) and for every \(0{\lt}𝜌≤ r-d(x,x_ 0)\). Hidden solution: [UNACCESSIBLE UUID ’0P0’] 
- E284
- [0P1] Topics:balls, disks. Let \(x_ 1,x_ 2∈ X\), \(r_ 1,r_ 2{\gt}0\), if \(d(x_ 1,x_ 2)≥ r_ 1+r_ 2\) then \begin{equation} \label{eq:disg_ palle} B_{r_ 1}(x_ 1)∩ D_{r_ 2}(x_ 2)=∅\quad . \end{equation}286- Hidden solution: [UNACCESSIBLE UUID ’0P2’] 
- E284
- [0P3] Topics:interior. Prerequisites:1.Show that \(B_ r(x)\) is an open set using the definition 283. Hidden solution: [UNACCESSIBLE UUID ’0P4’] 
- E284
- [0P5] Prove that a metric space is \(T_ 2\) i.e. Hausdorff (see definition in 246). 
- E284
- [0P6] If \(A=B^ c\) then show that \((\overline B)^ c={{A}^\circ }\) (using the definitions in this section). - Hidden solution: [UNACCESSIBLE UUID ’0P7’] 
- E284
- [0P8]Prerequisites:5.Show that the notions of interior and closure seen above are equivalent to those presented in the definition 244. 
- E284
- [0PB] Topics:interior. Show that \(E\) is open if and only if \({{E}^\circ }=E\). Hidden solution: [UNACCESSIBLE UUID ’0PC’] 
- E284
- [0PD] Topics:interior. Show that if \(A⊆ B⊆ X\) and \(A\) is open then \(A⊆ {{B}^\circ }\) using the above definitions. - Hidden solution: [UNACCESSIBLE UUID ’0PF’] 
- E284
- [0PG] Topics:interior. Show that if \(A⊆ B⊆ X\) then \({{A}^\circ }⊆ {{B}^\circ }\). Hidden solution: [UNACCESSIBLE UUID ’0PH’] 
- E284
- [0PJ] Topics:interior.Prerequisites:3,8. - Given \(X\) metric space and \(A⊆ X\), show that \[ {{A}^\circ } = {{\left({{A}^\circ }\right)}^\circ }~ ~ , \]- using the above definitions. - For what has been said in 7, this is equivalent to saying that \({{A}^\circ }\) is an open set. - (For the case of \(X\) topological space, see the 5) - Hidden solution: [UNACCESSIBLE UUID ’0PK’] 
- E284
- [0PM] Topics:interior. - Show that \(\overline E=E\) if and only if \(E\) is closed. 
- E284
- [0PP] Topics:closure. Prerequisites:9,5.(Replaces 0PN) - Show that if \(B⊆ A⊆ X\) then \( \overline B⊆ \overline A\); using the above definitions, or by switching to complement set and using 9. 
- E284
- [0PQ] Topics:closure.Prerequisites:5, 12. - Given a metric space \(X\) and a set \(A⊆ X\), show that \[ \overline A= \overline{\left(\overline A\right)} \]- either by transitioning to the complement set and using 10, or by using the definition of \(\overline A\) as ”set of adherent points”. - As discussed in 11, this is equivalent to saying that \(\overline A\) is a closed set. 
- E284
- [0PR] Let \(E⊆ X\), then \(E\) is a metric space with the restricted distance \(\tilde d=d|_{E× E}\). - Show that \(A⊆ E\) is open in \((E,\tilde d)\) (as defined at the beginning of this section) if and only there exists a set \(V⊆ X\) open in \((X,d)\) such that \(V∩ E = A\). - (The second way of defining ”open” is used in topology.) - Hidden solution: [UNACCESSIBLE UUID ’2GD’] 
- E284
- [0PS] Prerequisites:2.Let \(X\) be a set with two distances \(d_ 1,d_ 2\); let’s call \(𝜏_ 1,𝜏_ 2\) respectively the induced topologies. We have that \(𝜏_ 1⊆ 𝜏_ 2\) if and only if \[ ∀ x∈ X~ ∀ r_ 1{\gt}0~ ∃ r_ 2{\gt}0 ~ :~ B^ 2(x,r_ 2)⊆ B^ 1(x,r_ 1) \]- where \[ B^ 2(x,r_ 2)=\{ y∈ X:d^ 2(x,y){\lt}r_ 2\} \quad ,\quad B^ 1(x,r_ 1)=\{ y∈ X:d^ 1(x,y){\lt}r_ 1\} \quad . \]- Note that this exercise is the analogue in metric spaces of the principle 2 for the bases of topologies. 
- E284
- [0PT] Prerequisites:2, 15, 4, 2,3. - Having fixed \((X_ 1,d_ 1),\ldots ,(X_ n,d_ n)\) metric spaces, let \(X=X_ 1× \cdots × X_ n\). - Let \(𝜑\) be one of the norms defined in eqn. 11.1 in Sec. 11.1. Two possible examples are \(𝜑(x)=|x_ 1|+\cdots + |x_ n|\) or \(𝜑(x)=\max _{i=1\ldots n} |x_ i|\). - Finally, let’s define for \(x,y∈ X\) \begin{equation} d(x,y)=𝜑\big( d_ 1(x_ 1,y_ 1), \ldots , d_ n(x_ n,y_ n)\big)\quad . \label{eq:dist_ prodotto} \end{equation}287- Show that \(d\) is a distance; show that the topology in \((X,d)\) coincides with the product topology (see 2). - Note that this approach generalizes the definition of the Euclidean distance between points in \(ℝ^ n\) (taking \(X_ i=ℝ\) and \(𝜑(z)=\sqrt{∑_ i |z_ i|^ 2}\)). We deduce that the topology of \(ℝ^ n\) is the product of the topologies of \(ℝ\). - Hidden solution: [UNACCESSIBLE UUID ’0PX’] - See also the exercise 2, which reformulates the above using the concept of bases of topologies. 
Let \(D(x,r){\stackrel{.}{=}}\{ y\in X : d(x,y)\le r\} \) be the disk, show that it is closed.
Let \(S(x,r){\stackrel{.}{=}}\{ y\in X : d(x,y)= r\} \) be the sphere, show that it is closed.
Hidden solution: [UNACCESSIBLE UUID ’0PZ’] [0Q0]Prerequisites:287,3,287, 12.Let \(r{\gt}0\).
Let \(D(x,r){\stackrel{.}{=}}\{ y∈ X : d(x,y)≤ r\} \) be the disk; show that \(\overline{B(x,r)}⊆ D(x,r)\) and that \(B(x,r)⊆ {{D(x,r)}^\circ }\).
Let \(S(x,r){\stackrel{.}{=}}\{ y∈ X : d(x,y)= r\} \) be the sphere; show that \(∂{B(x,r)}⊆ S(x,r)\).
Find examples of metric spaces in which the above equalities (one, or both) do not hold.
Find an example of a metric space where there is a disk that is open 1 .
(See also 1 for the case of space \(ℝ^ n\)). Hidden solution: [UNACCESSIBLE UUID ’0Q1’][UNACCESSIBLE UUID ’0Q2’] [0Q3] Prerequisites:6.Let \(A⊆ X\) where \((X,d)\) is a metric space, we have that \(x∈ ∂ A\) if and only if there exists \((y_ n)⊆ A\) and \((z_ n)⊆ A^ c\) sequences such that \(y_ n→ x\) and \(z_ n→ x\). Hidden solution: [UNACCESSIBLE UUID ’0Q4’] [0Q5] Prerequisites:Section 7.9. Find an example of a metric space \((M,d)\) that does not satisfy the second axiom of countability, i.e. such that there is no countable base for the topology associated with \((M,d)\).
Hidden solution: [UNACCESSIBLE UUID ’0Q6’] [0Q7] Prerequisites:Section 7.9.Let \((M,d)\) be a metric space and suppose that there exists \(D⊆ M\) that is countable and dense. Such \((M,d)\) is called separable. Show that \((M,d)\) satisfies the second axiom of countability.
The converse is true in any topological space, see 2. [0Q8] Prerequisites:8,12, 7, 5.Difficulty:*.
Let \(X\) be a metric space, and \(A⊆ X\). We want to study the ”open-close” operation \(\overline{({{ A}^\circ })}\) (which is the closure of the interior of \(A\)).
- Show a simple example where \(\overline{({{ A}^\circ })}\) is not contained \(A\). 
- Then write a characterization of \(\overline{({{ A}^\circ })}\) using sequences and balls. 
- Use it to show that the ”open-close” operation is idempotent, that is, if \(D=\overline{({{ A}^\circ })}\) and then \(E=\overline{({{D}^\circ })}\) then \(E=D\). 
Hidden solution: [UNACCESSIBLE UUID ’0Q9’][UNACCESSIBLE UUID ’0QB’] [0QC] Prerequisites:1.Show that, for every closed set \(C⊆ X\) there exist countably many open sets \(A_ n\) such that \(⋂_ n A_ n=C\).
Hidden solution: [UNACCESSIBLE UUID ’0QD’]
A set obtained as an intersection of countably many open sets is known as ”a \(G_𝛿\) set”. The previous exercise shows that in a metric space every closed is a \(G_𝛿\).
Passing to the complement set, one obtains this statement. A set that is union of countably many closed sets is known as ”an \(F_𝜎\) set”. The previous exercise shows that in a metric space every open set is an \(F_𝜎\) set.
See also the section 13.4. [0QF] Difficulty:**.Find an example of a metric space where for every \(x∈ X,r{\gt}0\), \(B_ r(x)\) is a closed set, but the associated topology is not discrete. 2
Hidden solution: [UNACCESSIBLE UUID ’0QG’]
We note that such a space must be totally disconnected as shown in 1.
Bases composed of balls
To face these exercises it is necessary to know the concepts seen in Sec. 7.8.
- E287
- [0QJ]Prerequisites:2, 2.Show that the intersection of two balls is an open set (according to the definition 282). Hence the family of all balls meets the requirements (a) and (b) in exercise 2; so (as shown in 2), the family of balls is a base for the topology that it generates (which is the topology associated with metric space). - Hidden solution: [UNACCESSIBLE UUID ’0QK’] 
- E287
- Let’s review the exercise 16. - Having fixed \((X_ 1,d_ 1),\ldots ,(X_ n,d_ n)\) metric spaces, let \(X=X_ 1× X_ 1× \cdots × X_ n\). - Let \(d\) be the distance \[ d(x,y)=\max _{i=1,\ldots n} d_ i(x_ i,y_ i)~ ~ . \]- This is the same \(d\) defined as in eqn. ?? inside 16, setting \(𝜑(x)=\max _{i=1\ldots n}|x_ i|\). We indicate with \(B^ d(x,r)\) the ball in \((X,d)\) of center \(x∈ X\) and radius \(r{\gt}0\). - We want to show that \(d\) induces the product topology on \(X\), using the results seen in Sec. 7.8. - Taken \(t∈ X_ i, r{\gt}0\) we indicate with \(B^{d_ i}(t,r)\) the ball in metric space \((X_ i,d_ i)\). Let \({\mathcal B}_ i\) be the family of all balls in \((X_ i,d_ i)\). - Let \(\mathcal B\) be defined as \[ {\mathcal B}=\left\{ ∏_{i=1}^ n B^{d_ i}(x_ i,r_ i) : ∀ i,x_ i∈ X_ i,r_ i{\gt}0\right\} \]- This is the same \({\mathcal B}\) defined in 2. - Show that every ball \(B^ d(x,r)\) in \((X,d)\) is the Cartesian product of balls \(B^{d_ i}(x_ i,r)\) in \((X_ i,d_ i)\). So let \(\mathcal P\) be the family of balls \(B^ d(x,r)\) in \((X,d)\). - From 1 we know that \(\mathcal P\) is a base for the standard topology in the metric space \((X,d)\). - Use 2 to show that \(\mathcal P\) and \({\mathcal B}\) generate the same topology \(𝜏\). - Use 2 to prove that \(𝜏\) is the product topology. - We conclude that the distance \(d\) generates the product topology. 
Accumulation points, limit points
Let’s redefine this notion (a special case of what we saw in 251)
[0QN] Given \(A⊆ X\), a point \(x∈ X\) is an accumulation point for \(A\) if, for every \(r{\gt}0\), \(B(x,r)∩ A⧵\{ x\} \) is not empty.
- E288
- [0QP] Topics:adherent point, accumulation point. - Check that - Each accumulation point is also an adherent point, in symbols \(D(A)⊆ \overline A\); 
- if a point adhering to \(A\) is not in \(A\) then it is an accumulation point; 
 - So we obtain that \(\overline A=A∪ D(A)\). [UNACCESSIBLE UUID ’0QQ’] 
Let’s add this definition (a special case of 263).
[0QX] Given a sequence \((x_ n)_ n⊆ X\), a point \(x∈ X\) is said to be a limit point for \((x_ n)_ n\) if there is a subsequence \(n_ k\) such that \(\lim _{k→∞} x_{n_ k}=x\).
- E289
- [0QY] Find an example of a metric space \((X,d)\) and a bounded sequence \((x_ k)_ k⊆ X\) that has a single limit point \(x\) but that does not converge. - See also 2. 
- E289
- 
- If a sequence \((a_ k)_ k⊆ X\) converges to \(x\) then it has an unique limit point, which is \(x\). 
- If a Cauchy sequence \((a_ k)_ k⊆ X\) has a limit point then there is only one limit point \(x\) and \(\lim _ k a_ k=x\). 
 Hidden solution: [UNACCESSIBLE UUID ’0R0’] 
- E289
- [2F3]Topics:perfect set.Prerequisites:1,118,267.Difficulty:**. - Suppose \((X,d)\) is a complete metric space. A closed set \(E⊆ X\) without isolated points, i.e. consisting only of accumulation points, is called a perfect set. - Let \(C\) be the Cantor set. Assume that \(E\) is perfect and non-empty. Show that there exists a continuous function \(𝜑:C→ E\) that is an homeomorphism with its image. This implies that \(|E|≥ |ℝ|\). - So, in a sense, any non-empty perfect set contains a copy of the Cantor set. - This can be proven without relying on continuum hypothesis 118. Cf. 4. - Due to 4, it is enough to show that there exists a \(𝜑:C→ E\) continuous and injective. - Hidden solution: [UNACCESSIBLE UUID ’2F4’]