Skip to content

Commit

Permalink
Done with 8.4. Solutions set done
Browse files Browse the repository at this point in the history
  • Loading branch information
mikinty committed Jun 22, 2020
1 parent 505f901 commit 1717a9c
Show file tree
Hide file tree
Showing 3 changed files with 370 additions and 1 deletion.
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -36,4 +36,4 @@ There are also `TODO` comments if there are more minor parts that need work, e.g
- Chapter 5: :white_check_mark:
- Chapter 6: :white_check_mark:
- Chapter 7: :white_check_mark:
- Chapter 8: :hourglass: 8.1
- Chapter 8: :white_check_mark:
Binary file modified build/main.pdf
Binary file not shown.
369 changes: 369 additions & 0 deletions chapters/chapter8/chapter8-4.tex
Original file line number Diff line number Diff line change
Expand Up @@ -2,3 +2,372 @@

\section{A Construction of $\mathbb{R}$ From $\mathbb{Q}$}
\setcounter{exercise}{0}

This chapter has some confusing ordering for the exercises,
where you prove seemingly out of place properties.
I think this chapter could have the exercises ordered better
so that you prove all the properties you can before introducing
new properties.

\bx{
\ea{
\item We just need to verify the properties,
\begin{enumerate}[label=(c\arabic*)]
\item we know $r/2 < r \Rightarrow r/2 \in C_r$, so $C_r \neq \emptyset$. In addition, $C_r$ only contains
rational numbers less than $r$, so $r \not\in C_r \Rightarrow C_r \neq \mathbb{Q}$

\item If $x \in C_r$, then all $y \in \mathbb{Q}, y < x < r$ so therefore $y \in C_r$ as well.

\item $\forall x \in C_r, s=\frac{x + r}{2} \in C_r$, since $s < r$. But we see $s > x$ so
$C_r$ has no maximum.
\end{enumerate}

\item $2$ is a maximum, so this is not a valid cut.
\item This is a cut.
\item This is also a cut.
}
}

\bx{
We have 2 cases,
\begin{itemize}
\item $s \in \mathbb{Q}$, so since $A$ is a cut, we must have $r \leq s$, since otherwise
$r > s$ would be a contradiction since $A$ contains all rationals $< r$.
Then, since $s \not\in A$ but $r \in A$, $r \neq s$, so we have $r < s$.
\end{itemize}
}

\bx{
We examine:
\begin{itemize}
\item $\mathbb{N}$:
\begin{enumerate}[label=(f\arabic*)]
\item Commutativity holds
\item Associativity holds
\item there is no additive identity $0 \not\in\mathbb{N}$, but a mult. identity ($1$) exists
\item Inverses do not exist in general for multiplication and addition
\item Distributive property holds
\end{enumerate}

\item $\mathbb{Z}$:
\begin{enumerate}[label=(f\arabic*)]
\item Commutativity holds
\item Associativity holds
\item There is an additive identity $0$ and a mult. identity ($1$)
\item Inverses do not exist in general for multiplication, but addition does, $x, -x \in \mathbb{Z}$
\item Distributive property holds
\end{enumerate}

\item $\mathbb{Q}$:
\begin{enumerate}[label=(f\arabic*)]
\item Commutativity holds
\item Associativity holds
\item There is an additive identity $0$ and a mult. identity ($1$)
\item Inverses exist in multiplication $r, 1/r \in \mathbb{Q}$ and addition, $x, -x \in \mathbb{Q}$
\item Distributive property holds
\end{enumerate}
\end{itemize}
So we see that $\mathbb{Q}$ is the only field out of these three sets.

The text has an error where for $(f3)$, the additive identity is written as $x+0 = 0$, instead of
\begin{equation*}
x+0 = x.
\end{equation*}
I checked the official solutions manual and the author acknowledges this.
}

\bx{
Verifying the ordering properties,
\begin{enumerate}[label=(o\arabic*)]
\item For $A, B$ one must be a subset of the other, since they are either equal,
or they are not, in which case WLOG $\exists s \in A, s \not\in B$,
then it must be the case that $s > b, \forall b \in B$, and therefore $b \in A$.
We will then conclude $B \subseteq A \Rightarrow B \leq A$.

\item If we have $A \subseteq B, B \subseteq A$, by double containment, $A = B$.

\item If we have $A \subseteq B$ and $B \subseteq C$, then for any element $a \in A$,
we know $a \in B$ by $A \subseteq B$, and since $B \subseteq C$, this means $a \in C$.
THerefore, we conclude $A \subseteq C \Rightarrow A \leq C$.
\end{enumerate}
\label{chap8:ex:ordering_prop}
}

\bx{
\ea{
\item \begin{enumerate}[label=(c\arabic*)]
\item Since $A, B \neq \emptyset$, we know $\exists a+b \in A+B$ where $a \in A, b \in B$, so $A+B \neq \emptyset$.
$A+B \neq \mathbb{Q}$, because WLOG $A \subseteq B$,
then choose an arbitrary $s \in \mathbb{Q}, s < a+b, s \in A+B$, which we know
exists since $A+B$ has no maximum.
Since $s -a < b, s-a \in B$.
Now, $a \in B$ since $A \subseteq B$, so we
can conclude that $(s-a) + a = s \in B$. However, $s$ was arbitrary from $\mathbb{Q}$,
so we conclude that $B = \mathbb{Q}$, which is a contradiction. Therefore,
we must conclude that $A+B \neq \mathbb{Q}$.
\setcounter{enumii}{2}
\item We know $\exists r_a > a$ for $A$, and $\exists r_b > b$ for $B$,
which means for any $a+b \in A+B$, we know that $\exists r_a + r_b > a+b$, where $r_a+r_b \in A+B$
since $r_a \in A, r_b \in B$.
\end{enumerate}
We can therefore conclude that $A+B$ is a cut.

\item $A+B = B+A$ since $a+b = b+a$ over $\mathbb{Q}$. We also have $(A+B)+C = A+(B+C)$,
since this is equivalent to asking if $(a+b)+c = a+(b+c)$ for $a, b, c \in \mathbb{Q}$,
which we know is true since $\mathbb{Q}$ has associativity.

\item Take any $A$. We want to show that $A+O = A$.\footnote{
Once again, this is an error in the text where the author says to show $A+O = O$.
This is a pretty big error and repeated multiple times...
}
$A \subseteq A+O$, since given $r \in A$, we can always find $p \in A$ such that $p > r$.
Then let $q = r-p < 0$. We know $q \in O$, so we have $p + q = p+(r-p) = r \in A+O$.

$A+O \subseteq A$, since given any $s = a+o \in A+O$, $s < a \in A$,
and $A$ must contain all rationals less than $a \in A$, so therefore $s \in A$.
}
}

\bx{
\ea{
\item Verifying the cut properties,
\begin{enumerate}[label=(c\arabic*)]
\item We know $A \neq \emptyset$, so $\exists t \not\in A$.
Now, $-r \in \mathbb{Q}$, and we can always find some rational
less than $-r$, for example $-r/2$ if $-r$ is positive, $-r-1$
if $-r \leq 0$. Therefore $-A \neq\emptyset$.

The complement of $-A$ is
\begin{equation*}
\pbrac{
r \in \mathbb{Q}: \forall t \in A, t \geq -r
}
\end{equation*}
this set is not empty because if we assume it is, then
there is no $-r > t, t \in A$, which means $A$ must contain every rational,
since if $\exists q \not\in A$, then $q > t, t \in A$.
But this is a contradiction, since $A \neq \mathbb{Q}$,
therefore we conclude this set is not empty.
Hence $-A$ is a complement of a nonempty set, so it cannot be
$\mathbb{Q}$.

\item Take some $r \in -A$. If we consider some $q < r \Rightarrow -q > -r$,
we have
\begin{equation*}
t < -r < -q
\end{equation*}
so therefore $q \in -A$ as well.

\item Take some $r \in -A$. We know $\exists t < -r$.
Take $r' = (t-r)/2$, then $r' < -r \Rightarrow -r' \in -A$,
but we also have $-r' > r$.
\end{enumerate}

\item If $A = (-\infty, 2)$, then $-A = (-\infty, -2]$,
which has a maximum.

\item We need to show $a+r < 0$. \AFSOC $a+r \geq 0$,
then we have $a \geq -r$, but we know that $\exists t \not\in A, t < -r$,
so $a \geq -r > t$. But we know if $t \not \in A$, that $t > a$,
so this is a contradiction. We conclude $a+r < 0$, and $A+(-A) \subseteq O$.

We know want to show $O \subseteq A+(-A)$.
Given any element $o \in O$, we know that $o < 0$.
The strategy here is that $-A$ is essentially
``shifting'' $A$ to be $O$, so if $A$
is bounded by some $N$ that is $< 0$,
then $-A$ will shift $A$ towards the positive,
and if $N > 0$, then $-A$ will shift $A$
towards the negative.

Now, what this means, is the ``upper boundary''\footnote{
There is no $\sup$ definition, so we just have to use some intuition here.
} of $-A$
added with the ``upper boundary'' of $A$ should be $< 0$
and close to it. Everywhere else, any element of $A$
can just be added with the ``upper boundary'' of $-A$
to be $< 0$.
Therefore, we need to focus our efforts on this ``upper boundary''.

Let $\epsilon = -\abs{o}/2$, then
we can find a $t \not\in A$ such that $t - \epsilon \in A$,
because otherwise, $A$ is empty, because we can keep on proving that
$t - N\epsilon \not\in A$,
or $A = \mathbb{Q}$ since no such $t$ exists in the first place.
Therefore, from the definition of $-A$, we know that $t + \epsilon > t$,
so $t + \epsilon \in -A$.

Now, if we conclude $(t-\epsilon) + (t+\epsilon) = 2\epsilon < 0$,
we see we can make negative numbers arbitrarily close to $0$,
which means we can find $o < r < 0$, and since $o < r$, and $r \in A+(-A)$,
we conclude $o \in A+(-A)$.

\item Suppose we have $A \subseteq B, B \subseteq C$,
then $x \in A \Rightarrow x \in C$, so $A \subseteq C \Rightarrow A \leq C$
as desired.

Notice we have already proved this in Exercise
\ref{chap8:ex:ordering_prop}.
}
}

\bx{
\ea{
\item We have to show
\begin{enumerate}[label=(c\arabic*)]
\item We have $O \neq \emptyset$, so $AB \neq \emptyset$.
$\exists a, b \not\in A, B$ respectively, so
$a > a' \in A, b > b' \in B$, so $ab > a'b' \in AB$,
and therefore $ab \not\in AB$ and $AB \neq \mathbb{Q}$.

\item Consider $x \in AB$, and $q < x$.
Then we know $q < x = ab, a \in A, b \in B$, so $q/a < b$,
which means $q/a \in B$.
Then $a \cdot q/a = q \in AB$.

\item For any $x \in AB$, we know $x = ab, where a \in A, b \in B$.
Take the $a < a'$, $b < b'$, where $a' \in A, b' \in B$,
then $ab < a'b'$ where $ab, a'b' \in AB$.
\end{enumerate}

\item Suppose $A \leq B \Rightarrow A \subseteq B$. Then we want to show that
\begin{equation*}
C + A \leq C + B \Rightarrow C + A \subseteq C + B.
\end{equation*}
Consider any $x \in C+A$, then we know $x = c + a$.
Since $A \subseteq B, a \in B$ as well so $x$ is the sum of an element
from $C$ and $B$, hence $x \in C+B$.

\item Try the candidate $I = \pbrac{r \in \mathbb{Q}:r <1}$.
Then we can show that
\begin{itemize}
\item $AI \subseteq A$. For any $ai \in AI, a \in A, x \in I$, we have
$ax < a$, so therefore $ax \in A$.

\item $A \subseteq AI$. For any $a \in A$, choose $r_1 = a+\epsilon \in A$,
and $r_2 = 1-\epsilon/N \in I$.
Then we have
\begin{equation*}
(a+\epsilon)(1-\frac{\epsilon}{N}) =
a + \epsilon - \frac{
a\epsilon + \epsilon^2
}{
N
}
\end{equation*}
Since we can make $N$ arbitrarily large, this expression will be $> a$,
and also $\in AI$, so hence $a \in AI$.
\end{itemize}

\item $AO$ is defined as
\begin{equation*}
\pbrac{
ao: a \in A, o \in O, a, o \geq 0
} \cup O
= \emptyset \cup O = O
\end{equation*}
since $\not\exists o \geq 0$.
}
}

\bx{
\ea{
\item Proving cut properies for $S$,
\begin{enumerate}[label=(c\arabic*)]
\item If $\mathcal{A}$ is nonempty, then it must contain at least
one nonempty sets, so $S$, the union of these nonempty sets,
must be nonempty.

Since $\exists B \geq A, A \in \mathcal{A}$, we know
$\exists b \in B, b \not\in A$, so therefore $b \not\in S$,
and $S \neq \mathbb{Q}$.

\item For any $r \in S$, take any $q < r$. Since $r \in A \subseteq \mathcal{A}$,
we know $q \in A$, so therefore $q \in S$.

\item For any $r \in S$, since $r \in A \subseteq \mathcal{A}$,
we know $\exists q > r, q \in A$, so therefore $q \in S$
and $S$ has no maximum.
\end{enumerate}

\item Proving $S$ is the least upper bound for $\mathcal{A}$,
\begin{enumerate}[label=(\roman*)]
\item
Since $S$ is the union of all the way, for any $A \in \mathcal{A}$,
we have $A \subseteq S \Rightarrow A \leq S$.

\item \AFSOC we have some $B$ upper bound such that $S \not\leq B$,
meaning $S \not\subseteq B$. Then $\exists s \in S, s \not\in B$.

Since this $s$ belongs to some $A \in \mathcal{A}$, we know that
this $A \not\subseteq B \Rightarrow A \not\leq B$, and therefore
$B$ is not an upper bound, which is a contradiction.
\end{enumerate}
}
}

\bx{
\ea{
\item We need to show
\begin{itemize}
\item $C_r + C_s \subseteq C_{r+s}$.
We have
\begin{align*}
r' \in C_r, r' &< r\\
s' \in C_s, s' &< s\\
\Rightarrow r'+s' &< r+s
\end{align*}
so we conclude $r'+s' \in C_{r+s}$.

\item $C_{r+s} \subseteq C_r + C_s$. WLOG $C_r \subseteq C_s$.
Take any $x \in C_{r+s}$, then we know $x < r+s \Rightarrow x -r < s$.
Then $x -r-\epsilon \in C_s$. This means that
\begin{equation*}
(x-r-\epsilon) + (r-\epsilon) = x \in C_r + C_s
\end{equation*}
\end{itemize}

For multiplication,
\begin{itemize}
\item $C_rC_s \subseteq C_{rs}$.
We have
\begin{align*}
r' \in C_r, r' &< r\\
s' \in C_s, s' &< s\\
\Rightarrow r's' &< rs
\end{align*}
so we conclude $r's' \in C_{r+s}$.

\item $C_{rs} \subseteq C_{r}C_s$. WLOG $C_r \subseteq C_s$.
Take any $x \in C_{rs}$, then $x < rs \Rightarrow x/r < s$,
so $x/r \in C_s$.
Let $y \in C_s$ where $y > x/r$. Call $\epsilon = y-x/r$,
then we have $y = x/r+\epsilon$.
Now, we also know that $r - \epsilon/N \in C_r$, for sufficiently
large $N$.
Therefore, we can say
\begin{equation*}
\pa{
\frac{x}{r} + \epsilon
}
\pa{
r - \frac{\epsilon}{N}
} = x + \epsilon r - \frac{1}{N}\pa{
\frac{x}{r} + \epsilon
} > x \tag{Sufficiently large $N$}
\end{equation*}
so therefore this product $\in C_rC_s$ and therefore $x \in C_rC_s$
as well since $x$ is less than this product.
\end{itemize}

\item ($\Rightarrow$)
Suppose $C_r \leq C_s$. \AFSOC $r > s$.
Then we can find $r > q > s$, for $q \in C_r$. Since $q > s$,
$q \not\in C_s$, but this means $C_r \not\subseteq C_s$
which is a contradiction.

($\Leftarrow$) Suppose $r \leq s$.
Take any element $x \in C_r$. We know that $x < r \leq s$
so therefore $x \in C_s$, and we have $C_r \subseteq C_s \Rightarrow C_r \leq C_s$.
}
}

0 comments on commit 1717a9c

Please sign in to comment.