100만년 만에 지옥에서 돌아온 주인장이다!
결국 기말고사 기간이 겹치기 시작하면서 라텍스 연습과 블로그 글쓰기를 유기해버렸었다.
기말이 얼추 끝난 지금 블로그를 어떻게 운영해갈까 고민을 하다 보니, 그냥 내가 보는 책들의 연습문제 풀이를 하는 용도로 사용하는 것이 좋을 것 같다는 생각이 들었다.
책 소개를 조금 하자면, 이 책은 행렬과 행렬곱으로 이루어진 각종 군들을 분석하는 책이다. 선대 수업에서 이 책을 5장까지 했었는데, 책이 뒤에 내용이 흥미로운 부분이 많은 것 같아서 독학으로 공부하기로 했다.
바로 들어가 보자!
** 다만, 이 책의 경우 표기법이 매우 비통상적인 경우가 많아서, 잘 감안하도록 하자. 예컨대 통상적으로 path-connected라고 표시하는 성질을 이 책은 connected라고 표시한다.
1. Show that the definition of continuity reduces to the usual $\epsilon-\delta$ definition for $f:(a,b) \to \mathbb{R}$.
proof.
Recall that a function defined on metric spaces $f: (M,d) \to (N,d')$ is continuous at $x \in M$ if
$$\forall \epsilon>0, \exists \delta > 0 \space\space s.t. \space\space z \in B(x,\delta) \Rightarrow f(z) \in B(f(x),\epsilon)$$
now if $M = (a,b), N= \mathbb{R}$ and the distance functions are given by the absolute value,
$$z \in B(x,\delta) \Leftrightarrow |z-x|<\delta, f(z) \in B(f(x),\delta) \Leftrightarrow |f(z)-f(x)|<\epsilon$$
which implies
$$\forall \epsilon > 0, \exists \delta >0 s.t. |z-x|<\delta \Rightarrow |f(z)-f(x)| < \epsilon$$
Which is what we wished to show.
2. Suppose we have $A\subset \mathbb{R}^{n}$ and have functions
$$ A \xrightarrow{f} \mathbb{R}^{m}, f(A) \xrightarrow{g} \mathbb{R}^{p}$$.
We have seen that f and g continuous implies that $g \circ f$ is continuous. Give examples to show:
i) f contiuous and $g \circ f$ continuous does not imply g is continuous.
ii) g continuous and $g \circ f$ continuous does not imply f is continuous.
Sol of i).
Consider the simplest cases, $m=p=1$
We may have $$f: \mathbb{R} \to \mathbb{R}, f(x)\mapsto sin(x)$$
and $$g: \mathbb{R} \to \mathbb{R}, g(x)\mapsto \begin{cases} 0 \space\space x \leq 0 \\ 2\pi \space\space x > 0 \end{cases}$$
Where clearly g is discontinuous at $x=0$, but $g\circ f$ is continuous.
Sol of ii).
We may have $$f: \mathbb{R} \to \mathbb{R}, f(x)\mapsto \begin{cases} 0 \space\space x \leq 0 \\ 2\pi \space\space x>0 \end{cases}$$, $$g: \mathbb{R} \to \mathbb{R}, g(x)\mapsto sin(x)$$
then f is discontinuous at $x=0$ but $f\circ g$ is continuous.
3. Show that for $A\subset \mathbb{R}^{n}, f: A \to \mathbb{R}^{m}$, if f is continuous for each open set $U \subset \mahtbb{R}^{m}$, then $f^{-1}(U)$ is an open set in A.
proof.
It suffices to show that $f$ is continuous at every point of $\mathbb{R}^{m}$. But since $\mathbb{R}^{m}$ is clearly an open set in $\mathbb{R}^{m}$, it follows that $f$ is continuous at every point, which proves the claim.
(제대로 푼 건지 잘 모르겠다.)
4. Show that if A,B are connected sets in $\mathbb{R}^{n}$ and $A\cap B \neq \emptyset$, then $A\cup B$ is connected.
proof.
As $A\cap B \neq \emptyset, \exists x \in A\cap B$.
Take two elements $y,z\in A\cup B$. Say that $y \in A, z \in B$ (however in all four cases the argument remains identical).
As $x \in A\cap B$, there exists a path $\gamma_{1} : [0,\frac{1}{2}] \to A, \gamma(0) = y, \gamma(\frac{1}{2}) = x$.
Further, there also exists a path $\gamma_{2}: [\frac{1}{2},1] \to B, \gamma(\frac{1}{2}) = x, \gamma(1) = z$
appending the two paths together by $$\gamma : [0,1] \to A\cup B, \gamma(t) = \begin{cases} \gamma_{1}(t) \space\space 0 \leq t \leq \frac{1}{2}, \gamma_{2}(t) \space\space \frac{1}{2} < t \leq 1 \end{cases}$$ results in a (continuous) path in $A\cup B$ connecting y and z, as desired. Since y and z are arbitrary this shows connectedness of $A\cup B$.
5. Let H be any connected subgroup of a matrix group G. Show that
$$ S = \cup_{x \in G} xHx^{-1}$$
is connected.
proof.
It suffices to show the following:
Claim.
If $z \in xHx^{-1}, \exists \gamma: [0,1] \to xHx^{-1}, \gamma(0) = z, \gamma(1) = I$
or that any element of $xHx^{-1}$ is connected to the identity matrix I.
proof of the claim.
If $z \in xHx^{-1}, \exists h\in H, z = xhx^{-1}$.
Since $I \in H, \exists \xi : [0,1] \to H, \xi(0) = h, \xi(1) = I$.
We now define $\gamma : [0,1] \to xHx^{-1}, \gamma(t) = x*\xi(t)*x^{-1}$.
Then clearly $\gamma(0) = xhx^{-1} = z, \gamma(1) = I, \gamma(t) \in xHx^{-1}$.
Continuity follows from the fact that the left product $L_{x}: G \to G, L_{x}(y) \mapsto xy$ and right product $R_{x}: G \to G, R_{x}(y) \mapsto yx$ is continuous in a matrix group G, and $H \leq G$. (this is shown in problem 6.)
Therefore $z \in xHx^{-1}$ and I are connected. Therefore if we have $y,z \in \cup_{x\in G} xHx^{-1}$, we first think of a curve from y to I, then append it with a curve from I to z.
6. Show that matrix multiplication is continuous (with one matrix fixed); that is, $A \in M_{n}(k), L_{A}: M_{n}(K) \to M_{n}(K)$ given by $L_{A}(B) = AB$ is continuous.
proof.
Since all matrices of size n are isomorphic to real vector spaces of dimension $n^{2} (K = \mathbb{R}), 2n^{2} (K=\mathbb{C}), 4n^{2} (K = \mathbb{H})$, it suffices to show that the function $L_{A,ij}: M_{n}(K) \to K, L_{A,ij}(B) = (AB)_{ij}$ is continuous.
But $(AB)_{ij} = \Sigma_{k=1}^{n} a_{ik}b_{kj}$; as the function taking the ij-th component of a matrix $ij : M_{n}(K) \to K, ij(A) \mapsto a_{ij}$ is continuous, multiplying a field (or division ring) element by another to the right is continuous, and adding continuous functions is continuous, it follows that $L_{A,ij}$ is also continuous for all i,j, implying $L_{A}$ is continuous.
7. Show that an arbitrary union of open sets is open. (with the usual toplogy in a metric space)
proof.
Consider $\{U_{\alpha}\}$, a collection of open sets in some metric space (M,d).
If $x\in \cup_{\alpha} U_{\alpha}$, $\exists U' \in \{U_{\alpha}\}, x\in U'$
Since $U'$ is open, $\exists \epsilon > 0, B(x,\epsilon) \subset U' \subset \cup_{\alpha} U_{\alpha}$
but this implies (since x was arbitrary) that $\cup_{\alpha} U_{\alpha}$ is open.
8. Let $A \subset \mathbb{R}^{n}, x\in \mathbb{R}^{n}$. We say that x is a limit point of A if
$$B(x,r)\cap A$$
is an infinite set for all $r>0$.
Show that $C \subset \mathbb{R}^{n}$ is closed $\Leftrightarrow (x \space lp \space C \Rightarrow x \in C)$.
proof.
$(\Rightarrow)$
We will show that $\mathbb{R}^{n} - C$ is open.
Assume $x \notin C$; then x is not a limit point of C, so $\exists r>0, B(x,r) \cap C = \{x_{1},...,x_{n}\}$ is finite.
That means by taking $r < min\{d(x,x_{1}), ... , d(x,x_{n})\}$ we have $B(x,r) \cap C = \emptyset$ or $B(x,r) \subset \mathbb{R}^{n} - C$ implying that this set is open.
$(\Leftarrow)$
Suppose $C \subset \mathbb{R}^{n}$ is closed and $x \space lp \space C$.
We will show that $x \in C$. Suppose it is not the case and $x \notin C$. Then since $\mathbb{R}^{n} - C$ is open,
$\exists r>0, B(x,r) \subset \mathbb{R}^{n} - C$. But this implies $B(x,r)\cap C = \emptyset$ which is not infinite; this is a contradiction, which arises from our assumption that $x \notin C, x \space lp \space C$. This concludes the proof.
9. Let $D \subset \mathbb{R}^{n}$ be open and closed. Show that if D is not empty, then $D = \mathbb{R}^{n}$.
proof.
Suppose $D \neq \emptyset, D \neq \mathbb{R}^{n}$. Then $\exists x \in \mathbb{R}^{n} - D$. Take $x \notin D, y \in D$. Since $\mathbb{R}^{n}$ is connected, we can always have some continuous $\gamma : [0,1] \to \mathbb{R}^{n}, \gamma(0) = x, \gamma(1) = y$. Now $D \cap \gamma ([0,1])$ must be an open and closed set in $\gamma([0,1])$. Since it is bounded as well and has a smallest element. The smallest element must be zero, for it it were greater we can always take a ball around it that has a smaller element. But then this implies $\gamma(0) \in D$, a contradiction to $x \notin D$.