2
$\begingroup$

I am trying to understand the definition of what a Dedekind Cut is and I get that these are subsets of the rationals that are non-empty not not the whole of the rationals. I also sort of understand what it meant by downwards closed, but I do not understand the third part of the definition that states there is no largest element. Any suggestions?

A Dedekind Cut is a subset $x\subset\mathbb{Q}$ such that:

$\emptyset\neq\ x\neq\mathbb{Q}$

$x$ is downwards closed, i.e. if $q\in\ x$ and $r<q$, then $r\in\ x$

$x$ has no largest element.

$\endgroup$

5 Answers 5

3
$\begingroup$

If we restrict our space to only rational numbers (no irrationals) then the open interval $(-\infty, 1)$ is an example of a dedekind cut.

1) It isn't empty and it is not all of $\mathbb Q$.

2) If $r < q$ and $q \in (-\infty, 1)$ then $r\in (\infty, 1)$

3) If $q \in (-\infty, 1)$ then there is an $r > q$ so that $r \in (\infty, 1)$.

To find such an $r$ in property 3): We know that $q < 1$ so $1 - q > 0$. And $1-q > \frac {1-q}2 > 0$. So let $r = q + \frac {1-q}2 < q + (1-q) = 1$. As $r$ is rational, $r \in (-\infty, 1)$ and $r >q$. So $q$ is not a maximal element.

No element in $(-\infty, 1)$ is maximal because for any rational number less than $1$, one can always find a larger rational number between it and $1$.

Reading the comments you seem be confusing "maximal" elements with "numbers that are as large or larger than all elements in the set". The difference being that a maximal element has to actually be an element in the set. In the comments you asked "Why is $1$ not the maximal element of $(-\infty, 1)$". The answer to that is because $1$ isn't IN $(-\infty, 1)$.

All $(-\infty, q)$ for rational $q$ are dedikind cuts. In fact the only difference between an open interval $(-\infty, q)$ and a dedekind cut is that a dedekind cut might not have a well defined endpoint. A dedekind cut will be something like $(-\infty, ????)$ but that $????$ might not be describable.

For example:. $D = \{q \in \mathbb Q|q < 0$ or if $q\ge 0$ then $q^2< 2\}$ is of the form $(-\infty, ???)$ but $???$ would have to be some sort of square root of $2$ and there's no such thing in the rational numbers so there is no way to write that.

Which gives a hint to the punchline of Dedekind's joke. Once we define the real numbers, the dedekind cuts are nothing more or less than $(-\infty, x)\cap \mathbb Q$ where $x$ is a real number. But have to define the dedekind cuts before we define the real numbers in order to define the real numbers in the first place.

$\endgroup$
1
$\begingroup$

For example, $x = \mathbb{Q} \cap (-\infty,1]$ satisfies all the requirements in the definition except that it has a largest element $1$, so this $x$ is not a Dedekind cut.

But $x = \mathbb{Q} \cap (-\infty,1)$ has no largest element. Notice that this $x$ does have a least upper bound, namely the number $1$, but $1$ is not an element of $x$. And since $x$ is downward closed and $\emptyset \ne x \ne \mathbb{Q}$, it follows that this $x$ is a Dedekind cut.

$\endgroup$
5
  • $\begingroup$ But how is 1 not the largest element in this case? I mean, isn't there going to be some point where we can no longer find a largest element before getting to 1? $\endgroup$
    – Sam
    Commented Oct 7, 2017 at 4:12
  • $\begingroup$ 1 isn't the largest number because $1$ is NOT in the set. The largest number in a set has to be ... in the set. You are confusing largest number in a set, with the "largest number that still has members of the set 'right below' it". A largest element in a set means... just that. the largest element in the set. ANd $(-\infty, 1)$ clearly does not have one (either real or rational). $\endgroup$
    – fleablood
    Commented Oct 7, 2017 at 4:42
  • 1
    $\begingroup$ " I mean, isn't there going to be some point where we can no longer find a largest element before getting to 1?" How could that possibly ever happen? Consider $1 - \frac 12 < 1- \frac 13 < 1 \frac 14 < 1 - \frac 15 <........ < 1 - \frac 1{5,234} <-1- \frac 1{5,235} < ..... < 1- \frac 1{1,365,789,435,789,203,456,943} < ....$ are all less than $1$. Are we ever going to going to get to some point where we can't go any further? $\endgroup$
    – fleablood
    Commented Oct 7, 2017 at 15:11
  • $\begingroup$ @fleablood Okay, I get it, we can find infinitely many rationals before 1, so we can always find a larger number than the previous? $\endgroup$
    – Sam
    Commented Oct 8, 2017 at 17:11
  • $\begingroup$ Well, yeah. Basically. Not merely infinite but infinitely many within a finite space. But yes. No largest element is not strange. $\endgroup$
    – fleablood
    Commented Oct 8, 2017 at 18:56
1
$\begingroup$

So informally the Dedekind Cut of the real number $x$ is $\{r \in \Bbb Q: r < x\}$. Note the sign of $<$ instead of $\le$, such that the set has no maximum element (but does have a $\sup$remum, namely $x$).

$\endgroup$
5
  • $\begingroup$ OP said $x$ is a subset of rational numbers and then wrote that $r\in x$, so $r$ is a rational number, not a real number. $\endgroup$
    – user26486
    Commented Oct 6, 2017 at 21:14
  • $\begingroup$ @user26486 sorry, edited $\endgroup$
    – Kenny Lau
    Commented Oct 6, 2017 at 21:14
  • $\begingroup$ @KennyLau Okay, but why does the < sign mean no largest element? Aren't we going to eventually get to a point where we are just under the <? $\endgroup$
    – Sam
    Commented Oct 7, 2017 at 4:13
  • 1
    $\begingroup$ @Sam, no. There is never a "just under". There never has been and you've known that your entire life. What is the smallest positive number? Not zero, that isn't positive. For any possible number that could be the smallest there is always a smaller one. It's impossible for there to be a smallest number and it's kind of ridiculous to think there could be a smallest. What would happen if you divided that smallest number in half? $\endgroup$
    – fleablood
    Commented Oct 7, 2017 at 4:20
  • 1
    $\begingroup$ Suppose $q$ were then largest element in the set. $x > q$. That would mean there are no rational numbers between $q$ and $x$. So there are measurable gaps where there are no rational numbers at all. Image you took a rational right before the gap and a rational number right after the gap. Then you took the average. That'd be a rational number right in the gap. That's a contradiction. The idea that there isn't a highest element shouldnt be hard to imagine. The idea that there is should seem absurd. $\endgroup$
    – fleablood
    Commented Oct 7, 2017 at 4:32
1
$\begingroup$

A Dedekind cut on $\Bbb Q$ can also be defined as a pair $(A,B)$ of subsets of $\Bbb Q$ such that

(i). $A\ne \phi \ne B$ and $A\cap B=\phi$ and $A\cup B=\Bbb Q $.

(ii). Every $a\in A$ is less than every $b\in B.$

(iii). $A$ has no largest member.

For $q\in \Bbb Q$ let $A_q=\{r\in \Bbb Q: r<q\}$ and $B_q=\{r\in \Bbb Q:r\geq q\}.$ Then $(A_q,B_q)$ is a Dedekind cut on $\Bbb Q.$ In this case the second member $B_q$ of the Dedekind cut has a least member ($q$).

There are Dedekind cuts $(A,B)$ on $\Bbb Q$ for which $B$ has no least member. For example let $A=\{q\in \Bbb Q:q\leq 0\lor (q>0\land q^2<2)\}$ and $B=\Bbb Q \setminus A.$ Such Dedekind cuts correspond to members of $\Bbb R \setminus \Bbb Q.$

In fact we $define$ $\Bbb R$ in such a way that

(i'). If $r\in \Bbb R\setminus \Bbb Q$ then there is a unique Dedekind cut $(A_r,B_r)$ on $\Bbb Q$ such that $a<r<b$ for all $a\in A_r$ and $b\in B_r$, and such that $B_r$ has no least member .

(ii'). If $(A,B)$ is a Dedekind cut on $\Bbb Q$ such that $B$ has no least member then there is a unique $r\in \Bbb R\setminus \Bbb Q$ such that $a<r<b$ for every $a\in A$ and every $b\in B$.

$\endgroup$
1
$\begingroup$

For any rational number $p$, there is a corresponding subset of rational numbers defined by

$\quad I_p = \{q \in \mathbb Q \; | \; q \lt p \}$

So we have a function

$\tag 1 \mathcal R: \mathbb{Q} \to \mathcal{P}(\mathbb{Q})$

with domain $\mathbb Q$ and target the power set of $\mathbb Q$; it is defined by

$\tag 2 p \mapsto I_p = \mathcal R(p)$

Proposition 1: The function $\mathcal R$ is an injection.
Proof: Exercise.

Proposition 2: Each subset $I_p$ of $\mathbb Q$ satisfies the following conditions:

$\tag 3 I_p \ne \emptyset$
$\tag 4 I_p \ne \mathbb Q$
$\tag 5 \text{IF } q \in I_p \text{ And } r \lt q \text{ THEN } r \in I_p$
$\tag 6 \text{IF } s \in I_p \text{ THEN There Exists } t \in I_p \text{ such that } t \gt s$
Proof: Exercise.

Define $D_{x^2 = 2\text{?}} = \{q \in \mathbb Q \; | \; q^2 \lt 2 \} \; \bigcup \; I_0$.

Proposition 3: The set $D_{x^2 = 2\text{?}} \subset \mathbb Q$ in not in the range of $\mathcal R$. Moreover, this set satisfies the conditions (3) thru (6).
Proof: Exercise.

So the set of all Dedekind Cuts $x\subset\mathbb{Q}$ contains all the $I_p$ but also 'other stuff'. In fact, it contains a 'whole bunch of other stuff'.

$\endgroup$

You must log in to answer this question.

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