
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% exercise set 1 - solutions
%%% due september 4, 2009
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\documentclass{amsart}
\usepackage{graphicx}
\usepackage{amssymb}
\usepackage{epstopdf}
\usepackage{nicefrac}
\DeclareGraphicsRule{.tif}{png}{.png}{`convert #1 `dirname #1`/`basename #1 .tif`.png}


% \input hw.sty

\textwidth = 6 in
\textheight = 8.5 in
\oddsidemargin = 0.0 in
\evensidemargin = 0.0 in
\topmargin = 0.0 in
\headheight = 0.0 in
\headsep = 0.0 in
\parskip = 6pt
\parindent = 0.0in

% \hoffset=-.7in
% \voffset=-.7in


\newcommand{\mN}{{\mathbb N}}
\newcommand{\mR}{{\mathbb R}}


\begin{document}

\begin{center}
Math 445,  Fall 2009 \hfill  Exercise Set \#1 \hfill Solutions
\end{center}
 
 \bigskip
   
 {\bf 1.}   [\#10, page 9] ~ Call a subset $B$ of a set $A$ {\it cofinite} if the complement of $B$ in $A$ is finite.
 If $B$ and $C$ are cofinite subsets of $A$, prove that $B \cap C$ is cofinite.
 
 \proof
 We are given that $(A-B)$ and $(A-C)$ are both finite sets. Then by de~Morgan's Laws,
 $$A - (B \cap C) = (A-B) \cup (A-C) $$
 As the union of finite sets is finite, $A - (B \cap C)$ is finite, hence $B \cap C$ is cofinite.
  \endproof
  
  
  \medskip
  
  
 {\bf 2.} [\#2(*), page 13]   Let $L$ be a partially ordered set in which {\it every} subset has a top and bottom element. Prove that $L$ is a finite chain.
  
\proof
Suppose that $L$ is not finite, then there exists a countably infinite subset of distinct points 
$X_0 = \{x_k \mid k = 1,2, \ldots\} \subset L$. We will use induction to construct an infinite subset $B_{\infty}$ with no top element, which is  a contradiction. Hence, $L$ cannot be infinite.

By assumption, there exists both a top and a bottom element for $X_0$. Let $a_1 = x_{k_1}\in X_0$ denote the top element, and $b_1 = x_{\ell_1} \in X_0$ denote the bottom element  of $X_0$. Set $S_1 = \{a_1 , b_1\}$. Note that $b_1 < a_1$ and that  the complement  $X_1 = X_0 - S_1$ is again infinite.

Now assume that we have chosen a subset 
$S_n = \{a_1, a_2, \ldots , a_n, b_1, b_2, \ldots , b_n\} \subset  X_0$
 where each $a_i = x_{k_i} \in X_0$ and $b_i =x_{\ell_i} \in X_0$ are distinct elements, and they satisfy the order relation
 $$ b_1 < b_2 < \cdots b_n < a_n < \cdots < a_1$$
Set  $X_n = X_0 - S_n$. Since $X_0$ is infinite, the set $X_n$ is again infinite. 

By assumption, there exists a   top and a bottom element for $X_n$. Let $a_{n+1} = x_{k_{n+1}}\in X_n$ denote the top element, and $b_{n+1} = x_{k_{m+1}} \in X_n$ denote the bottom element  of $X_n$. Then $b_{n+1} < a_{n+1}$. 

Since $a_n$ is the top element of $X_n$ and $a_{n+1} \in X_n - \{a_n, b_n\}$ so $a_{n+1} \ne a_n$ we have that $a_{n+1} < a_n$.

Since $b_n$ is the bottom element of $X_n$ and $b_{n+1} \in X_n - \{a_n, b_n\}$ we have that $b_n < b_{n+1}$.

 Then let $S_{n+1} = S_n \cup \{a_{n+1} , b_{n+1}\}$. We are given that $a_{n+1} < a_n$ and $a_n < \cdots < a_1$ hence 
 $a_{n+1} < \cdots < a_1$. Likewise,  we are given that $b_n < b_{n+1}$ and $ b_1  < \cdots b_n$ so 
 $ b_1  < \cdots b_{n+1}$. Combining these inequalities we obtain
  $$ b_1 <  \cdots < b_n < b_{n+1} < a_{n+1} < a_n < \cdots < a_1$$
Set  $X_{n+1} = X_0 - S_{n+1}$. Since $X_0$ is infinite, the set $X_{n+1}$ is again infinite. 

 Now let $B_{\infty} = \{b_n \mid n = 1,2, \ldots \}$. Since the elements are increasing, there can be no top element.
\endproof

 \medskip
  
 {\bf 3.} [\#3, page 13]   Let $\mN$ be the chain of positive integers, in its usual order. Is $\mN$ complete? Is $\mN$ complete if $\omega$ = ``$\infty$'' is placed on top? 
 
\proof
A set $A$ is \emph{complete} if for every   $B \subset A$  there exists a least upper bound $u \in A$ and greatest lower bound $l \in A$.    If $A = \mN$ then we can let $B = \mN$ for example, and while $1 \in \mN$   is a greatest lower bound,   there is no top element. That is, there is no integer upper bound  for the set of integers.

Let  $A = \mN \cup \{\infty\}$, with the order relation $n < m$ for $n, m \in \mN$. Define $n < \infty$ for all $n \in \mN$. Given any subset $B \subset A$, there always exists a least element in $B$ which is the greatest lower bound for $B$. If $B$ is finite, then it has a top element, which is the least upper bound. If $B$ is infinite, then $\infty \in A$ is a least upper bound. So $A$ is complete.


  \medskip
 
 {\bf 4.} [\#4, page 13]   Let $\mN$ be the set of positive integers and define ``$m \leq n$'' to mean that $m$ divides $n$. Is $\mN$ a lattice? Is it complete? If not, how could we make it complete?
 
 \proof
 First we show this defines a partial order on $\mN$.   
 Clearly, for all $m\in \mN$, $m$ divides $m$ so $m \leq m$.
 
 If $m \leq n$ then there exists an integer $p \in \mN$ such that $n = p \cdot m$. If $n \leq m$ then there exists an integer $q \in \mN$ such that $m = q \cdot n$. Thus, $n = p \cdot m = p \cdot q \cdot n$ which implies that $1 = p \cdot q$. As both $p, q$ are positive integers, this implies $ p = q = 1$ and hence $m = n$.
 
 Let $m \leq n$ and $n \leq t$. Then there exists integers $p, q \in \mN$ such that $n = p \cdot m$ and $t = q \cdot n$. Thus, $t = p \cdot (q \cdot m) = (pq) \cdot m$ so $m \leq t$.
 Let $\mN_*$ denote the set $\mN$ with this partial order. 
 
 Next we show that $\mN_*$ is a lattice. Given $m,n \in \mN_*$ let $\gcd\{m,n\}$ be the least common denominator of $\{m,n\}$ and let ${\rm lcm}\{m,n\}$ denote the least common multiple of $\{m,n\}$. These exists due   either to some ancient Greek, or to some ancient Chinese mathematicians. Then $\gcd\{m,n\}$ is the greatest lower bound of the set $\{m,n\}$ - essentially by definition. Similarly, ${\rm lcm}\{m,n\}$ is the least upper bound of $\{m,n\}$. Thus, $\mN_*$ is a lattice for this partial order.
 
 The lattice is not complete. Let $p > 1$ be a prime. Define $C_p = \{a_n = p^n \mid n = 1,2, \ldots\} \subset \mN_*$. Since $a_n \leq a_{n+1}$ for this partial order,  there is no upper bound for the chain $C_p$.
 
How can  we make $\mN_*$ complete? First, we must add to $\mN_*$ an upper bound for each chain $C_p$. Let $p^{\infty}$ denote the upper bound for $C_p$ (This is much like adding $\infty$ to $\mN$ in the previous exercise.) 

Next,  $m \in \mN_*$ with a prime factorization $m = p_1^{\ell_1} \cdots p_k^{\ell_k}$ where each $p_i > 1$ and $\ell_i \geq 1$.
 If $k > 1$ then no prime power $p^m$ is comparable to $m$ hence we must add a point at infinity $m^{\infty}$ which is an upper bound for the chain $C_m$ formed by the powers of $m$.  
 
 But this also does not suffice. Can you find a chain for which none of these additional ``infinities'' is an upper bound? 
  In fact, it is necessary to add an uncountable number of upper bounds, or infinities to $\mN_*$ to compactify this set. 
 So we have not even come close yet to finding all the upper bounds we need to add.   
 \endproof
   
 
 \medskip
 
 {\bf 5.} [\#9, page 13]   Let $L$ be a distributive lattice with a top element ``1'' and a bottom element ``0''. 
 (Recall this means that $0 \leq a \leq 1$ for all $a \in L$.)
 Prove that if an element $a \in L$ has a complement, then the complement $a'$ is unique.   
 \proof
 Recall that $b \in L$ is a \emph{complement} for $a$, if $b$ satisfies $a \cup b = 1$ and $a \cap b = 0$. 

Let $b, c \in L$ be complements for $a$. We will show that $c \leq b$ and $b \leq c$, hence $b =  c$. Thus, one can define $a' =b$ or $a' = c$, unambiguously. 

First, $a \cup c = 1$ and $b \cap a = 0$ implies 
$$ b   = b \cap 1 =  b \cap (a \cup c) = (b \cap a) \cup (b \cup c) = 0 \cup (b \cap c) = b \cup c$$
from which we conclude by the definitions that  $c \leq b$. But also, $a \cup b = 1$ and $c \cap a = 0$ implies 
$$ c   = c \cap 1 =  c \cap (a \cup b) = (c \cap a) \cup (c \cup b) = 0 \cup (c \cap b) = c \cup b$$
Hence $b \leq c$, as was to be shown.
 \endproof
 
 
  \vfill
  \eject
  
 
 {\bf 6.} [\#7, page 17]    Given a function $f \colon A \to A$, we write $f^n$ for the function on $A$ obtained by taking the composite of $f$ with itself $n$ times. Suppose that $f^n$ equals the identity function for some $n$ (one then says that $f$ is \emph{periodic}.) Prove that such $f$ is one-to-one  and onto. 

\proof
Suppose that $f^n = id$. Set $g = f^{n-1}$. Then $f \circ g = g \circ f = Id$,  so the map $f$ has an inverse $g$. Therefore, $f$ is one-to-one and onto.
\endproof


 \medskip
 
 {\bf 7.} [\#8, page 17]    As a generalization of periodic functions, 
 we say that $f \colon A \to A$ is \emph{locally periodic} if for every $x \in A$ 
 there exists an integer $n(x) \geq 1$, depending on $x$, such that $f^{n(x)}(x) = x$.
Prove that a locally periodic function is one-to-one and onto. 
 
 \proof
 This problem is like the above, except we cannot just write down the inverse function. We must show that $f$ is one-to-one, and then show it is onto.
 
We first show that $f$ is 1-1.  Let $y,z \in A$ and suppose that $f(y) = f(z)$. Let $m = n(y)$ and $n = n(z)$.

Observe that $y = f^m(y)$, so $y = f^m(f^m(y)) = f^{2m}(y)$ and we can iterate this as many times as needed. 
In particular, $y = f^{nm}(y)$. 

Similarly,     $z = f^n(z)$, so $z = f^n(f^n(z)) = f^{2n}(z)$ and we can iterate this as many times as needed. 
In particular, $z = f^{mn}(z)$. 

Thus, $y = f^{nm}(z) = f^{nm-1}(f(z)) = f^{nm-1}(x)$ and $z = f^{mn}(z) = f^{mn-1}(f(z)) = f^{mn-1}(x)$.

So $y=  f^{nm-1}(x) =  f^{mn-1}(x) = z$.



Next, we show that $f$ is onto, which is simpler. 
Let $x \in A$, then $x = f^{n(x)}(x) = f(f^{n(x)-1}(x))$.  
Set $w = f^{n(x) -1}(x) \in A$, then $x = f(w)$.
 \endproof
 
 
 
 
 \medskip
 
 {\bf 8.} [\#14, page 18]    Fix a set $A$. For a subset $S \subset A$, the characteristic function $\phi_S$ of $S$ is the function from $A$ to the set $\{0,1\}$ which takes the value $1$ on every element of $S$, and the value $0$ on every element of the complement $S' = A - S$. Prove, for subsets $S, T \subset A$:
 
 \begin{enumerate}
\item $\phi_{S \cap T} = \phi_S \cdot \phi_T$ ~ (the product of the two functions)
\item $\phi_{S'} = 1 - \phi_{S}$
\item $\phi_S + \phi_T = \phi_{S \cup T} + \phi_{S \cup T}$ (note typo!)
\end{enumerate}
  
 \proof Rule (1) is typical: 
$$\phi_{S \cap T}(x) =1 \Longleftrightarrow  x \in S \cap T \Longleftrightarrow x \in S  ~ \& ~  x \in T \Longleftrightarrow  \phi_S(x) =1 ~ \& ~ \phi_T(x) = 1  \Longleftrightarrow \phi_S(x) \cdot \phi_T(x) = 1$$
 
 Rules (2) and (3) follow the same pattern.
 \endproof
 
   
 \vfill
 
\end{document}


