8.1 Metrics

Definition 8.1.1 (Metric Space).label Let $X$ be a set and $d: X \times X \to [0, \infty]$, then $d$ is a metric if:

  1. (PM1)

    For any $x \in X$, $d(x, x) = 0$.

  2. (M)

    For any $x, y \in X$ with $x \ne y$, $d(x, y) > 0$.

  3. (PM2)

    For any $x, y \in X$, $d(x, y) = d(y, x)$.

  4. (PM3)

    For any $x, y, z \in X$, $d(x, z) \le d(x, y) + d(y, z)$.

The pair $(X, d)$ is a metric space, which comes with the metric uniformity induced by $d$, and the corresponding topology.

Proposition 8.1.2.label Let $(X, d)$ be a metric space, then the following are equivalent:

  1. (1)

    $X$ is second countable.

  2. (2)

    $X$ is Lindelöf.

  3. (3)

    $X$ is separable.

In particular, since second countability is hereditary, separability and the Lindelöf property are both hereditary in metric spaces.

Proof. (1) $\Rightarrow$ (2): Let $\seqi{U}\subset 2^{X}$ be an open cover of $X$. Let $\seq{V_n}\subset 2^{X}$ be a countable base for the topology on $X$, and

\[K = \bracs{n \in \natp| \exists i \in I: V_n \subset U_i}\]

For any $x \in X$, there exists $i \in I$ such that $x \in U_{i}$, and $n \in \natp$ such that $x \in V_{n} \in U_{i}$. Therefore $\bigcup_{k \in K}V_{k} = X$.

Let $J \subset I$ be countable such that for each $k \in K$, there exists $j \in J$ with $V_{n} \subset U_{j}$, then $X = \bigcup_{k \in K}V_{k} = \bigcup_{j \in J}V_{j}$.

(2) $\Rightarrow$ (3): Let $n \in \nat$, then $\bracs{B(x, 1/n)|x \in X}$ is an open cover of $X$, so there exists $\seq{x_{n, k}}\subset X$ such that $X = \bigcup_{k \in \natp}B(x_{k}, 1/n)$. Let $D = \bracs{x_{n, k}| n, k \in \natp}$, then for any $x \in X$ and $n \in \natp$, there exists $k \in \natp$ such that $x \in B(x_{n, k}, 1/n)$, so $x_{n, k}\in B(x, 1/n)$. Therefore $D$ is dense in $X$, and $X$ is separable.

(3) $\Rightarrow$ (1): Let $\seq{x_n}\subset X$ be a countable dense subset. Let $x \in X$ and $k \in \natp$, then there exists $x_{n} \in \natp$ such that $d(x, x_{n}) < 1/(2k)$. In which case, $x \in B(x_{n}, 1/(2k)) \subset B(x_{n}, 1/k)$. Therefore $\bracs{B(x_n, 1/k)|n, k \in \natp}$ forms a countable basis for $X$.$\square$

Proposition 8.1.3.label Let $\seq{(X_n, d_n)}$ be metrisable spaces, then $\prod_{n \in \natp}X_{n}$ is also metrisable.

Proof. For each $n \in \natp$, let

\[d_{n}': \braks{\prod_{n \in \natp}X_n}^{2} \to [0, \infty] \quad (x, y) \mapsto d_{n}(\pi_{n}(x), \pi_{n}(y))\]

then $d_{n}'$ is a pseudometric on $X$, and $\bracsn{d_n'}_{1}^{\infty}$ induces the product uniformity on $\prod_{n \in \natp}X_{n}$. By Theorem 6.3.10, $\prod_{n \in \natp}X_{n}$ is also metrisable.$\square$

Theorem 8.1.4 (Banach’s Fixed Point Theorem).label Let $(X, d)$ be a complete metric space and $f: X \to X$. If there exists $C \in (0, 1)$ such that

\[d(f(x), f(y)) \le Cd(x, y) \quad \forall x, y \in X\]

then:

  1. (1)

    There exists a unique $x \in X$ such that $f(x) = x$.

  2. (2)

    For any $y \in X$, $\limv{n}f^{n}(y) = x$.

Proof. Let $x_{0} \in X$ be arbitrary, and $x_{n} = f^{n}(x_{0})$, then for ecah $n \in \natp$,

\[d(x_{n}, x_{n+1}) \le C d(x_{n-1}, x_{n}) \le C^{n} d(x_{0}, x_{1})\]

Thus $\seq{x_n}\subset X$ is Cauchy, and converges to a point $x \in X$.

(2): For any $y_{0} \in X$, let $y_{n} = f^{n}(y_{0})$, then $d(x_{n}, y_{n}) \to 0$ as $n \to \infty$, so $\limv{n}f^{n}(y_{0}) = x$.

(1): Since $f$ is Lipschitz continuous,

\[f(x) = f\braks{\limv{n}f^n(x)}= \limv{n}f^{n+1}(x) = x\]

$\square$

Post a Comment

Name:Email:
Please enter the tag of the current page (7A) to post the comment.
Tag: