0
$\begingroup$

Most texts I've read have been using one of the following versions (or a mix of them) as order axioms for the real numbers.

Version 1: $(\mathbb{R},\leq)$ is a totally ordered set, such that for all $a,b,c\in\mathbb{R}$ we have that \begin{align*} (0\leq a)\wedge (0\leq b)&\Rightarrow 0\leq a+b \\ (0\leq a)\wedge (0\leq b)&\Rightarrow 0\leq a\cdot b. \end{align*}

Version 2: $(\mathbb{R},\leq)$ is a totally ordered set, such that for all $a,b,c\in\mathbb{R}$ we have that \begin{align*} a\leq b&\Rightarrow a+c\leq b+c \\ (a\leq b)\wedge (0\leq c)&\Rightarrow a\cdot c\leq b\cdot c. \end{align*}

Since both of them can be used as axioms, they should be equivalent. However, so far I didn't succeed in proving that. Would be great if anyone could help me with that.

$\endgroup$
3
  • $\begingroup$ Version 2 should be sufficient, but Version 1 definitely isn't sufficient: you can't say anything about $a\leq b$ when $a,b\neq 0$ unequal. $\endgroup$ Commented Oct 14, 2021 at 16:17
  • $\begingroup$ Forgot to mention that in all versions it is stated that $(\mathbb{R},\leq)$ is a totally ordered set. $\endgroup$
    – PAT
    Commented Oct 14, 2021 at 16:25
  • $\begingroup$ @DonThousand The reverse ordering satisfies neither set of axioms. For we have $0 \geq -1$, but we do not have $0 \geq (-1) \cdot (-1)$. $\endgroup$ Commented Oct 14, 2021 at 16:53

1 Answer 1

0
$\begingroup$

Version 1 does not work. For consider the order $\leq'$ defined by

$$a \leq' b :\equiv \begin{cases} b \leq a & a, b < 0 \\ a \leq b & otherwise \end{cases}$$

This is a total order on $\mathbb{R}$. Furthermore, one can show that $(\mathbb{R}, \leq')$ and $(\mathbb{R}, \leq)$ are order-isomorphic, so $(\mathbb{R}, \leq')$ also satisfies any other properties of the order of $\mathbb{R}$ which are phrased purely in terms of the order relation (and not in reference to any other functions or operations defined on $\mathbb{R}$).

$(\mathbb{R}, \leq')$ satisfies Version 1 of the axioms, but it clearly fails to satisfy Version 2.

In fact, consider any total order $\leq''$ on $\mathbb{R}^2$ which satisfies Version 2. Note that version 2 implies that $0 \leq'' c^2$ for any $c$. First, note that this follows immediately for $0 \leq c$, since we would have $0 \leq c \cdot c$. And for $c \leq 0$, we see that $-c + c \leq -c + 0$, so $0 \leq -c$, so $c^2 = (-c)^2 \leq 0$.

Now suppose that $a \leq b$. Then there exists $c$ such that $a + c^2 = b$. Then we see that $a + c^2 + 0 \leq'' b + c^2$. Then $a + c^2 \leq'' b + c^2$. Adding $-(c^2)$ to both sides gives us $a \leq'' b$.

So therefore, we see that $\leq''$ agrees with $\leq$.

Therefore, version 2 is enough to completely specify the order relation on $\mathbb{R}$.

In fact, consider any ordered ring $(R, \leq)$ in which all nonnegative elements have square roots. Then version 2 uniquely specifies the order on $R$ by the above argument.

Note that in rings where not every nonnegative element has a square root, there can be more than one total order satisfying (2). For consider $R = \{a + b \sqrt{2} \mid a, b \in \mathbb{Q}\}$, which is an ordered field under the normal $\leq$ order inherited from $\mathbb{R}$. But there is a nontrivial ring isomorphism $R \to R$, given by $a + b\sqrt{2} \mapsto a - b\sqrt{2}$. This indicates there must be another order relation $\leq'$ on $R$, defined by

$$a + b \sqrt{2} \leq' c + d \sqrt{2} :\equiv a - b \sqrt{2} \leq c - d \sqrt{2}$$

which also satisfies the ordered ring requirements. This suffices to prove that $R$ is not closed under the taking of square roots of nonnegative elements.

$\endgroup$
1
  • $\begingroup$ Thanks for your help. $\endgroup$
    – PAT
    Commented Oct 21, 2021 at 14:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .