One point compactification
Theorem (One point compactification) Any locally compact space \(X\) can be embedded in another compact space \(Y\), which just has one more point than \(X\), such that the relative topology of \(X\) with respect to \(Y\) is the same as the original topology of \(X\). The space \(Y\) thus constructed is called one point compactification of \(X\).
Proof The key step of the proof is to construct the topology of \(Y\) by adopting the open sets from \(X\) whose complements are compact in \(X\).
Construction of the topology \(\mathcal{T}_Y\) for \(Y\)
Let \(\{U_i\}_{i \in I}\) be a collection of open sets in \(X\) such that \(X - U_i\) is compact in \(X\) for all \(i \in I\). Figuratively speaking, these open sets are the spaces \(X\) punctured with holes having bounded dimension. We also note that because the space \(X\) is locally compact, hence for all \(x \in X\), there is a compact neighborhood containing \(x\). This ensures the above selected collection of open sets is not empty.
Let \(Y\) be the space by appending one point \(y_0\) to \(X\). Hence, \(Y = X \cup \{y_0\}\). Let \(\mathcal{T}_Y\) be the topology of \(Y\) which is defined via the following two rules.
For all \(V \in \mathcal{T}_Y\),
- if \(V\) does not contain \(y_0\), \(V\) is an open set of \(X\) in \(X\)'s original topology \(\mathcal{T}_X\).
- if \(V\) contains \(y_0\), then \(V \cap X \in \{U_i\}_{i \in I}\), where \(X - U_i\) is compact in \(X\).
Prove \(\mathcal{T}_Y\) is a topology of \(Y\)
We'll then prove that such constructed \(\mathcal{T}_Y\) really defines a topology for \(Y\).
- The empty set \(\Phi\) does not contain \(y_0\) and it belongs to \(\mathcal{T}_X\), so \(\Phi \in \mathcal{T}_Y\) according to rule 1.
- The whole space \(Y\) contains the point \(y_0\) and \(Y \cap X = X\). Because \(X^{c} = \Phi\), which is compact in \(X\) as a trivial case, \(Y\) belongs to \(\mathcal{T}_Y\).
We should check that: a) the union of any number of sets in \(\mathcal{T}_Y\) belongs to \(\mathcal{T}_Y\); b) the intersection of any finite number of sets in \(\mathcal{T}_Y\) belongs to \(\mathcal{T}_Y\).
- If the selected sets from \(\mathcal{T}_Y\) all satisfy rule 1, i.e., they are selected from \(X\)'s original topology \(\mathcal{T}_X\), it is obvious that the union and finite intersection of these sets are still open in \(X\). Because they do not contain \(y_0\), they satisfy rule 1.
If the selected sets from \(\mathcal{T}_Y\) all satisfy rule 2, they all contain point \(y_0\). Let such selected collection of sets be \(\{V_j\}_{j \in J}\). Let \(\{U_j\}_{j \in J}\) be the associated open sets such that \(U_j = V_j \cap X\) for all \(j \in J\). Then we have
\[
\left( \left( \bigcup_{j \in J} V_j \right) \cap X \right) ^c = \left( \bigcup_{i \in J} U_j \right)^c = \bigcap_{j \in J} U_j^c.
\]Because all \(U_j^c\) are compact and hence closed in \(X\), their intersection is also closed. Because a closed subset of a compact set is still compact, the intersection of all \(U_j^c\) is compact. Thus, rule 2 is satisfied.
Similarly, the intersection of a finite number of sets selected from \(\mathcal{T}_Y\) satisfying rule 2 can be proved to belong to \(\mathcal{T}_Y\).
If there are two subgroups in the selected sets from \(\mathcal{T}_Y\), such that the sets in one subgroup satisfy rule 1 and those in the other subgroup satisfy rule 2, the union or intersection of all the selected sets can also be divided into two groups corresponding to the two rules. Then we only need to verify:
\(\forall V_1, V_2 \in \mathcal{T}_Y\), where \(V_1\) satisfies rule 1 and \(V_2\) satisfies rule 2, then both \(V_1 \cup V_2\) and \(V_1 \cap V_2\) are in \(\mathcal{T}_Y\).
For the union of \(V_1\) and \(V_2\), it contains \(y_0\). Then we check if the complement of \((V_1 \cup V_2) \cap X\) is compact in \(X\).
\[
\left( V_1 \cup V_2 \right) \cap X = (V_1 \cap X) \cup (V_2 \cap X) = V_1 \cup U_2.
\]Then
\[
(V_1 \cup U_2)^c = V_1^c \cap U_2^c,
\]which is a closed subset of the compact set \(U_2^c\). Therefore, the complement of \((V_1 \cup V_2) \cap X\) is compact.
For the intersection of \(V_1\) and \(V_2\), it does not contain \(y_0\). We have
\[
V_1 \cap V_2 = V_1 \cap U_2.
\]Because both \(V_1\) and \(U_2\) are open sets in \(X\), their intersection is an open set in X, so is \(V_1 \cap V_2\).
Summarizing the above, we've proved \(\mathcal{T}_Y\) is really a topology for \(Y\). It is also obvious to see from the above proof that the relative topology of \(X\) with respect to \(Y\) is the same as its original topology.
Prove \(Y\) is compact
Let \(\{V_i\}_{i \in I}\) be an open covering of \(Y\). Then \(\{ V_i \cap X \}_{i \in I}\) is an open covering of \(X\). Meanwhile, there exists an index \(i_0 \in I\) such that \(y_0 \in V_{i_0}\). Let \(U_{i_0} = V_{i_0} \cap X\), so \(U_{i_0}^c\) is compact in \(X\). Because \(\{ V_i \cap X \}_{i \in I}\) covers \(U_{i_0}^c\), there exists a finite subcovering
\[
U_{i_0}^c \subset \bigcup_{k=1}^n \{ V_{i_k} \cap X \}.
\]
Because \(V_{i_0}\) contains both \(y_0\) and \(U_{i_0}\), the collection \(\{ V_{i_k}\}_{k=1}^n\) appended with \(V_{i_0}\) forms an finite subcovering of \(Y\). Hence \(Y\) is compact.
Examples of one point compactification
- The real line \(\mathbb{R}\) is not compact. By adding an infinity point \(\infty\), the real line can be transformed to a circle with \(\infty\) as the paste point.
- The plane \(\mathbb{R}^2\) is not compact. By adding an infinity point \(\infty\), the plane can be transformed to a sphere with \(\infty\) as the paste point.
One point compactification的更多相关文章
- 普林斯顿数学指南(第一卷) (Timothy Gowers 著)
第I部分 引论 I.1 数学是做什么的 I.2 数学的语言和语法 I.3 一些基本的数学定义 I.4 数学研究的一般目的 第II部分 现代数学的起源 II.1 从数到数系 II.2 几何学 II.3 ...
- Discrete cosine transform(离散余弦转换)
A discrete cosine transform (DCT) expresses a finite sequence of data points in terms of a sum of co ...
随机推荐
- EntityFramework之事务
一.EF事务 引用程序集 using System.Transactions; 用法 var writer = new System.IO.StringWriter(); try { using (v ...
- 题解 SP26045 【GCDMAT2 - GCD OF MATRIX (hard)】
承接一下洛咕上的题解,这里基本就是谈谈优化,放个代码的 我们发现这里的常数主要来自于除法,那么我们优化除法次数,把所有的 \(n/1...n/s\) (\(s=\sqrt n\))存下来,然后归并排( ...
- OpenStack实践系列⑦深入理解neutron和虚拟机
OpenStack实践系列⑦深入理解neutron和虚拟机 五.深入理解Neutron 5.1 虚拟机网卡和网桥 [root@node1 ~]# ifconfig brq65c11cc3-8e: fl ...
- Fragment与Acitvity通信
Fragment与Activity通信的方式如下: 一.通过初始化函数提供 1.在动态添加Fragment的过程中,我们在Activity中通过Fragment.setArguments()的方法为F ...
- python2x 和 python 3x的区别
1,大环境: python2x: 大神贡献源码,这些源码有自己语言的特色,源码不规范,源码重复代码太多. python崇尚的是优美清晰简单. python3x: 龟叔重新整理,将源码规范化,简单化,统 ...
- [C]gcc编译器的一些常用语法
简单的GCC语法: 如果你只有一个文件(或者只有几个文件),那么就可以不写Makefile文件(当然有Makefile更加方便),用gcc直接编译就行了.在这里我们只介绍几个我经常用的几个参数,第一是 ...
- Python traceback的优雅处理
刚接触Python的时候,简单的异常处理已经可以帮助我们解决大多数问题,但是随着逐渐地深入,我们会发现有很多情况下简单的异常处理已经无法解决问题了,如下代码,单纯的打印异常所能提供的信息会非常有限. ...
- ES的Query、Filter、Metric、Bucketing使用详解
由于笔者在实际项目仅仅将ES用作索引数据库,并没有深入研究过ES的搜索功能.而且鉴于笔者的搜索引擎知识有限,本文将仅仅介绍ES简单(非全文)的查询API. 笔者原本打算在本文中介绍聚合API的内容,但 ...
- Codeforces 280D k-Maximum Subsequence Sum [模拟费用流,线段树]
洛谷 Codeforces bzoj1,bzoj2 这可真是一道n倍经验题呢-- 思路 我首先想到了DP,然后矩阵,然后线段树,然后T飞-- 搜了题解之后发现是模拟费用流. 直接维护选k个子段时的最优 ...
- LuoGu P2002 消息扩散
题目传送门 这个题其实就是tarjan缩点的板子题对吧....至少我是这么想的 首先这是个有向图,对于一个有向图,我们肯定要考虑环的存在与否,恰好这个题又是让我们找出最少的点,使得这几个点能够走遍全图 ...