0

I want to apply the proof strategies and overall scratch work diagram based framework introduced in the book "How to Prove It" in order to rewrite the following theorem:

Theorem Any positive rational number has an expression as a fraction in lowest form.

Proof. First write a given positive rational number as a quotient of positive integers $\frac{m}{n}$. We know that 1 is a common divisor of $m$ and $n$. Furthermore, any common divisor is at most equal to $m$ or $n$. Thus among all common divisors there is a greatest one, which we denote by $d$. Thus we can write $$ m = dr \qquad \text{and} \qquad n = ds $$ with positive integers $r$ and $s$. Our rational number is equal to $$ \frac{m}{n} = \frac{dr}{ds} = \frac{r}{s} $$

All we have to do now is to show that the only common divisor of $r$ and $s$ is $1$. Suppose that $e$ is a common divisor which is greater than 1. Then we can write $$ r = ex \qquad \text{and} \qquad s = ey $$ with positive integers $x$ and $y$. Hence $$ m = dr = dex \qquad \text{and} \qquad n = ds = dey $$ Therefore $de$ is a common divisor for $m$ and $n$, and is greater than $d$ since $e$ is greater than $1$. This is impossible because we assumed that $d$ was the greatest common divisor of $m$ and $n$. Therefore $1$ is the only common divisor of $r$ and $s$, and our theorem is proved. $\blacksquare$

The first thing that I did was to rewrite the statement using logical connectives and quantifiers $$ \forall a: (a \in \mathbb{Q}_+) \rightarrow \exists \ r,s : (r,s \in \mathbb{Z}_+) \ \land (a = r/s) \ \land (\operatorname{gcd}(r,s) = 1). $$ making it clear that we are trying to prove a conditional statement.

So we should assume that $(a \in \mathbb{Q}_+)$ and our scratch work diagram becomes \begin{array}{c c c c c} \text{Givens} & & \qquad \qquad \qquad \qquad \qquad & & \text{Goal} \\ \hline (a \in \mathbb{Q}_+) & & & & \exists \ r,s : (r,s \in \mathbb{Z}_+) \ \land (a = r/s) \ \land (\operatorname{gcd}(r,s) = 1) \\ \end{array}

How should I proceed from here? This seems like an existence proof using a proof by contradiction at some point, and while the original proof makes sense to me, I'm having trouble factoring it in small steps using this framework.

fire-bee
  • 322
  • This basic result has been proved here many times in many ways, e,g, see the linked dupe. It is not clear what you are asking, but the question is not answerable unless you precisely define "proof strategy ...framework". Rewriting everything in symbolic logic is the wrong way to go if what you ultimately seek is a better understanding of the key idea(s) of the proof. – Bill Dubuque May 23 '22 at 12:25
  • @BillDubuque, I didn't say I wanted a proof with just symbols. I'm not asking for a formal proof and nowhere in what I wrote that seems to be implied. What I'm asking for is for a proof rewriting using the techniques to approach proofs that appear in a book for beginners that I mentioned in text. The argument that the basic result has been proved here doesn't even make sense, since I even gave a proof of it myself. – fire-bee May 23 '22 at 12:35
  • But you have not defined "proof rewriting using the techniques to approach proofs that appear in a book that I mentioned", so most readers will have no idea what that means. Why do you seek such a rewrite? – Bill Dubuque May 23 '22 at 12:41
  • 2
    I said: scratch work diagram based framework introduced in the book "How to Prove It". And in the title: using Velleman's 'given–goal diagrams' This is what I asked for, there isn't any other question here that asks for that. It doesn't seem like a duplicate to me. And it is contextualized for people that actually read the book. I'm appealing to those people, and even the title seems enough for that. Only those who know the name will click on it. – fire-bee May 23 '22 at 12:46
  • The reason for the rewrite is for self-study. I'm not studying this specific proof, I'm studying the book, and its techniques. And I want to see how far can I apply it outside of it. Not just on the examples presented in it. – fire-bee May 23 '22 at 12:48
  • @BillDubuque If I define such "framework" is the question valid then? – fire-bee May 23 '22 at 12:53
  • I reopened it and removed some tags (the question is really about various proof methods that have to do with logic - they are not specific to those tags). – Bill Dubuque May 23 '22 at 13:06
  • @BillDubuque, thank you for your understanding. – fire-bee May 23 '22 at 13:07
  • 3
    I don't understand why you think rewriting is necessary. The proof already follows the strategies in How To Prove It. The given $a\in\mathbb{Q}+$ means $\exists m\exists n(m,n \in \mathbb{Z}+ \wedge a = m/n)$, so the proof begins by introducing $m$ and $n$. The proof then produces positive integers r and s such that $r/s=a$ and $\text{gcd}(r,s) = 1$. The proof that $\text{gcd}(r,s)=1$ uses proof by contradiction. All of this is following the strategies in How To Prove It. – Dan Velleman May 26 '22 at 13:16
  • @DanVelleman, I don't think that it is necessary to rewrite. I just wanted to try and reproduce the thought process of the person that wrote the proof and write that process by iteratively updating givens–goal tables in the scratch work. What I'm interested in here is the scratch work. Something like examples 3.1.3 or 3.2.1 in the book (regardless of the forms in those examples) where every step is explained, and stuff is added to or removed from the givens and goal columns.

    Are you the author of How to Prove it?

    – fire-bee May 27 '22 at 11:49
  • 2
    Yes, I wrote How To Prove It. Theorem 6.4.5 in How To Prove It involves similar reasoning. – Dan Velleman May 29 '22 at 22:17
  • @DanVelleman, thanks, I hadn't reached that far into the book. The book is awesome btw! – fire-bee May 30 '22 at 10:20
  • @DanVelleman I ended up trying to answer my own question, i.e., write the scratch work of the proof. Would you be kind to take a look? – fire-bee Jul 21 '23 at 16:41

1 Answers1

0

Our goal is the universal conditional statement

$$ \forall a \; \biggl[\; a \in \mathbb{Q}_+ \; \rightarrow \; \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr)\; \biggr]$$ with no additional assumptions.

Scratch work

The starting givens-goals diagram is \begin{array}{c c c} \text{Givens} & \qquad \qquad & \text{Goal} \\ \hline & & \forall a \; \biggl[\; a \in \mathbb{Q}_+ \; \rightarrow \; \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr)\; \biggr] \end{array}

$\\[15pt]$

We introduce the new variable $a$ into the proof to stand for an arbitrary object

\begin{array}{c c c} \text{Givens} & \qquad \quad \qquad \qquad & \text{Goal} \\ \hline & & a \in \mathbb{Q}_+ \; \rightarrow \; \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr) \end{array}

$\\[15pt]$

Since our goal is a conditional statement we begin by assuming that the antecedent is true

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr) \end{array}

$\\[15pt]$

Assuming that $a \in \mathbb{Q}_+$ allows us to use the definition of positive rational number

\begin{array}{c c c} \text{Givens} & & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr) \\ \exists \ m \ \exists \ n \; \Bigl( \; m,n \in \mathbb{Z}_+ \; \land \; a = m/n \;\Bigr) & & \end{array}

$\\[15pt]$

We let $m_0, n_0$ stand for positive integers such that $a = m_0 /n_0$ (existential instantiation)

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr) \\ m_0, n_0 \in \mathbb{Z}_+ & &\\ a = m_0/n_0 & & \\ \end{array}

$\\[15pt]$

We let $d$ stand for $\operatorname{gcd}(m_0,n_0)$

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr) \\ m_0, n_0 \in \mathbb{Z}_+ & &\\ a = m_0/n_0 & & \\ d = \operatorname{gcd}(m_0,n_0) && \end{array}

$\\[15pt]$

Let produce positive integers $r,s$

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \\ m_0, n_0 \in \mathbb{Z}_+ & &\\ a = m_0/n_0 & & \\ d = \operatorname{gcd}(m_0,n_0) && \\ r = m_0/d && \\ s = n_0/d && \end{array}

$\\[15pt]$

We treat the conjunction as separate goals

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & r,s \in \mathbb{Z}_+\\ m_0, n_0 \in \mathbb{Z}_+ & & a = r/s\\ a = m_0/n_0 & & \operatorname{gcd}(r,s) = 1\\ d = \operatorname{gcd}(m_0,n_0) && \\ r = m_0/d && \\ s = n_0/d && \end{array}

$\\[15pt]$

The try to prove $\operatorname{gcd}(r,s) = 1$ by contradiction

\begin{array}{c c c} \text{Givens} & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad & \text{Goal} \\ \hline a \in \mathbb{Q}_+ & & r,s \in \mathbb{Z}_+\\ m_0, n_0 \in \mathbb{Z}_+ & & a = r/s\\ a = m_0/n_0 & & \textrm{contradiction}\\ d = \operatorname{gcd}(m_0,n_0) && \\ r = m_0/d && \\ s = n_0/d && \\ \operatorname{gcd}(r,s) \neq 1 && \end{array}

$\\[15pt]$

Form of final proof:

Let $a$ be arbitrary

$\quad$ Suppose $a \in \mathbb{Q}_+$. Let $m_0, n_0 \in \mathbb{Z}_+$ and $a = m_0 / n_0$. Let $d = \operatorname{gcd}(m_0,n_0)$ .

$\quad$$\quad$ Let $r = m_0/d$ and $s = n_0/d$.

$\quad$$\quad$$\quad$ [Proof of $r,s \in \mathbb{Z}_+$ goes here.]

$\quad$$\quad$$\quad$ [Proof of $a = r/s$ goes here.]

$\quad$$\quad$$\quad$ Suppose $\operatorname{gcd}(r,s) \neq 1$

$\quad$$\quad$$\quad$$\quad$ [Proof of contradiction goes here.]

$\quad$$\quad$$\quad$ Therefore $\operatorname{gcd}(r,s) = 1$

$\quad$$\quad$ Thus, $\exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr)$.

$\quad$ Therefore $a \in \mathbb{Q}_+ \; \rightarrow \; \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr)$.

Since $a$ was arbitrary, we can conclude that $\forall a \; \biggl[\; a \in \mathbb{Q}_+ \; \rightarrow \; \exists \ r \ \exists \ s \; \Bigl( \; r,s \in \mathbb{Z}_+ \; \land \; a = r/s \; \land \; \operatorname{gcd}(r,s) = 1 \;\Bigr)\; \biggr].$

fire-bee
  • 322