0
$\begingroup$

From what I understand, most mathematicians don't actually think of the rational numbers as equivalence classes of ordered pairs of integers—rather, that is how they are modelled in set theory. The "everyday" definition of a rational number is that is a number of the form $a/b$, where $a$ and $b$ are integers and $b\neq0$. This description is closer to how we think of rational numbers when we are not doing set theory. My question is: is there a similar "everyday" definition of a real number?

I have heard some descriptions, but none of them feel particularly satisfying:

  • "A real number is any number that can be represented as an infinite decimal." This description is slanted towards a completely arbitrary way of representing real numbers, and doesn't seem to get any closer to what a real number actually is.
  • "A real number is a rational or irrational number". This definition is circular, as an irrational number is defined as a real number which is not rational. Of course, many "everyday" definitions are ultimately circular, but this description just doesn't seem very enlightening to me.

To be clear, I'm not asking about a specific construction of the reals like Dedekind cuts or cauchy sequences of rationals—rather, I am asking about what "everyday" notion these constructions are trying to capture. I'm also interested in everyday notions of real numbers that perhaps can't be turned into formal constructions, but are enlightening nonetheless.

$\endgroup$
8
  • 4
    $\begingroup$ The "everyday" definition of a rational number is not just the form $a/b$. People also see $1/2$, $2/4$, etc. as the same number, so it is equivalence classes of ordered pairs of integers after all, the description simply states what is there in a fanciful abstract way. The "everyday" notion of real number is harder to pinpoint because the thing is a late artifact of long historical developments. At the root they were points on the geometric line, and "filling the holes" between rationals was a major theme in arithmetical approaches, with infinite decimals historically first (Stevin, c. 1600). $\endgroup$
    – Conifold
    Commented Aug 14, 2021 at 11:16
  • 9
    $\begingroup$ Are infinite decimals really that arbitrary? In order to specify an irrational number in general (aside from special ones such as $\sqrt{2}$ that have simple geometric interpretations), how would you do it if not by specifying where it is relative to the rationals? One way is by Dedekind cuts, which specifies which rationals are larger and which are smaller than the target number. The other is by a (Cauchy) sequence of rationals that converges to the target number. Infinite decimals are simply one specific and methodical way of constructing such a sequence. $\endgroup$
    – user169852
    Commented Aug 14, 2021 at 11:17
  • 3
    $\begingroup$ I'm not sure what's arbitrary about decimals, unless it the privileging of base ten that bothers you. If that's your objection, you might prefer to think of reals as continued fractions, which give unique representations of reals (if you're a little careful with the definitions) and don't involve any arbitrary choice of base. But I wouldn't claim that's the way mathematicians think of real numbers. $\endgroup$ Commented Aug 14, 2021 at 11:21
  • 6
    $\begingroup$ A single point on the real line $\endgroup$ Commented Aug 14, 2021 at 11:22
  • 1
    $\begingroup$ @GerryMyerson: Yes, it was the base ten that was bothering me. But now I see from your and Bungo's comment that the idea of real numbers being infinite decimals has a nice generalisation, which seems less arbitrary. Thanks for the help. $\endgroup$
    – Joe
    Commented Aug 14, 2021 at 11:29

4 Answers 4

3
$\begingroup$

The everyday definition I always give my students of a real number is something that can express a directed length from a starting point (0) on the number line, with positive being length to the right and negative being length to the left. I do this after defining smaller sets, and showing how they aren't enough to do all lengths via the old 1/1/root 2 right triangle.

$\endgroup$
1
$\begingroup$

The important sets of numbers are characterized by the fundamental operations we expect to be able to do with them. This is also called "closure" of an operation; for example way say that $\mathbb{Z}$ is closed under subtraction. In most cases, constructing these sets from set theory axioms is just aimed at making those operations possible.

The natural numbers $\mathbb{N}$ is a set where we can:

  • Have a unique smallest number (either $0$ or $1$ depending on context and discipline).
  • Take the unique next number after another number (successor operation).
  • Take the smallest number in any non-empty set of numbers.
  • As a consequence of the above properties, we can also add and multiply numbers in $\mathbb{N}$.

($\mathbb{N}$ is actually a fairly nuanced and interesting case! But it's not the focus of this question or answer.)

The integers $\mathbb{Z}$ is a set including $\mathbb{N}$ where we can:

  • Add.
  • Subtract.
  • Multiply.

Here "including" means for every day use we can think of the relation as $\subset$, but in the set theory definitions this is not true, and we instead have an injective function which preserves the operations like addition which make sense in both sets.

The rational numbers $\mathbb{Q}$ is a set including $\mathbb{Z}$ where we can:

  • Add.
  • Subtract.
  • Multiply.
  • Divide, when the divisor is not zero.

The algebraic numbers $\overline{\mathbb{Q}}$ is a set including $\mathbb{Q}$ where we can:

  • Add.
  • Subtract.
  • Multiply.
  • Divide, when the divisor is not zero.
  • Take the roots of any polynomial with coefficients in $\overline{\mathbb{Q}}$.

Although it's quite common to discuss polynomial roots in $\mathbb{C}$ or $\mathbb{R}$, those sets actually have properties not required for that purpose and "extra" numbers which are not roots of polynomials over $\mathbb{Q}$ (i.e., transcendental numbers).

The real numbers $\mathbb{R}$ is a set including $\mathbb{Q}$ where we can:

  • Add.
  • Subtract.
  • Multiply.
  • Divide, when the divisor is not zero.
  • Take the limit of any converging sequence.

The complex numbers $\mathbb{C}$ is a set including $\mathbb{R}$ (and, it turns out, $\overline{\mathbb{Q}}$), where we can:

  • Add.
  • Subtract.
  • Multiply.
  • Divide, when the divisor is not zero.
  • Take the limit of any converging sequence.
  • Have a number $i$ where $i^2 = -1$.
  • Using the above properties, we can prove the Fundamental Theorem of Algebra, meaning we can take the roots of any polynomial with coefficients in $\mathbb{C}$.

Note each time we add a new operation type, the set theory development will first define the new number set in terms of the operation's action just on an existing number set. Then we can also prove a similar operation makes sense on the entire new sense in a way consistent with the first limited operation, and without needing any additional numbers in the set (this is the "closure" property again). For example, $\mathbb{Z}$ is usually defined based on pairs in $\mathbb{N} \times \mathbb{N}$ with equivalence classes inspired by the idea of subtracting any two natural numbers; then we can prove we can also subtract two of these new integers. Likewise, the set theory definitions of $\mathbb{R}$ use rational numbers as the elements of Cauchy-convergent sequences or Dedekind cuts, and then we can prove limit properties in $\mathbb{R}$ more generally.

So I'd say the "everyday" definitions of these sets are just the fundamental properties, and a few of the most important theorems, listed above. Related to that are the important rules about the operations themselves, like $x+y=y+x$ and $x \cdot 1 = x$ and $(x+y)-y=x$, but I think of that as more "what addition, subtraction, and multiplication are", beyond the fact that "integers are numbers we can subtract". For everyday use we're not really concerned with how set theory can define the sets of numbers and their operations and show that the operations satisfy our expected rules, just that the numbers and operations are there.

From this point of view, the essential thing the real numbers have when compared to the rational numbers is closure of converging limits. (We do need to specify "converging", just to rule out unbounded sequences which don't converge at all.) A decimal representation (or other base $b$ representation), Cauchy-convergent sequences of rationals, and Dedekind cuts are all ways of adding this concept and operation to $\mathbb{Q}$. When study of geometry suggests we need numbers with properties like $x^2 = 2$ (which we can give names like $\sqrt{2}$), the intuition such numbers "ought to exist" probably reasons from the fact that the square of a rational number can get arbitrarily close, but never exactly equal, to $2$, and again the limit concept comes into play. So introducing the set $\mathbb{R}$ makes sense, although the other set $\overline{\mathbb{Q}}$ would also suffice for finding square roots. Again, every day use doesn't need to care about how real numbers are defined in set theory, just that the numbers and their properties including sequence limits are there.

$\endgroup$
0
$\begingroup$

The real numbers are exactly like the rational numbers, except that ever sequence that should have a limit (because the sequence elements get closer and closer together) actually has a limit.

For example, if you use Newton iteration to find better and better approximations to the square root of 2, starting with 1, that sequence should have a limit, the square root of 2. But in the rational numbers, that limit doesn’t exist because it is not rational. In the real numbers, that kind of limit is a real number and exists.

(I think this is easier to understand than the “least upper bound” axiom. And the fact that certain sequences don’t have limits in the rational numbers is naively unexpected, so the real numbers are “rational numbers without the defects”).

$\endgroup$
-2
$\begingroup$

An element of a dedekind-complete ordered field? Usually in math we don't think about what things "are" but how they behave. And the axioms of dedekind-complete ordered fields describe completely how $\mathbb R$ behaves. We only need to construct the real numbers (as dedekind-cuts, or classes of equivalence of cauchy sequences or whatever else) to prove they exist. But in almost all practical situations we only need these axioms.

$\endgroup$
4
  • 1
    $\begingroup$ This is a 2 year old question asking for an "everyday" definition. It seems clear that your answer doesn't fit what was asked. $\endgroup$
    – Brick
    Commented Jul 2 at 19:24
  • $\begingroup$ @Brick It doesn't seem clear to me that my answer doesn't fit what was asked. Could you elaborate? $\endgroup$
    – Carla_
    Commented Jul 2 at 20:25
  • $\begingroup$ Someone asking for an “everyday” definition is almost surely asking for something way less technical than “dedekind-complete”. $\endgroup$
    – Brick
    Commented Jul 2 at 20:58
  • $\begingroup$ @Brick How is "every non-empty bounded set has a least upper bound" technical? $\endgroup$
    – Carla_
    Commented Jul 2 at 23:25

You must log in to answer this question.

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