Mathematics · Notes

The real numbers

12 October 2026

This note collects the definitions and conventions used in Tool 05. It builds on note 03, which defines the maximum and minimum of a set.

Number systems

We write

N={1,2,3,…},Z={…,−2,−1,0,1,2,…},Q={pq:p,q∈Z, q≠0},\mathbb{N} = \{1, 2, 3, \ldots\}, \qquad \mathbb{Z} = \{\ldots, -2, -1, 0, 1, 2, \ldots\}, \qquad \mathbb{Q} = \left\{ \tfrac{p}{q} : p, q \in \mathbb{Z},\ q \ne 0 \right\},

and R\mathbb{R} for the real numbers. In this course, 0∉N0 \notin \mathbb{N}. Some texts include 00; when reading other sources, check their convention.

Order

The real numbers carry the usual order <<, and a≤ba \le b means a<ba < b or a=ba = b. We use the following rules, which are part of the definition of R\mathbb{R} as an ordered field. For all real numbers aa, bb and cc:

  1. exactly one of a<ba < b, a=ba = b and b<ab < a holds;
  2. if a<ba < b and b<cb < c, then a<ca < c;
  3. if a<ba < b, then a+c<b+ca + c < b + c;
  4. if a<ba < b and c>0c > 0, then ac<bcac < bc.

Every other rule for manipulating inequalities follows from these four, so each step in a calculation can be justified by them. Rule 4 requires c>0c > 0. Several exercises in Tool 05 turn on what goes wrong when that condition is forgotten.

Intervals

Let aa and bb be real numbers with a≤ba \le b. The bounded intervals with endpoints aa and bb are

IntervalSetEndpoints included
[a,b][a, b]{x∈R:a≤x≤b}\{x \in \mathbb{R} : a \le x \le b\}both
(a,b)(a, b){x∈R:a<x<b}\{x \in \mathbb{R} : a < x < b\}neither
[a,b)[a, b){x∈R:a≤x<b}\{x \in \mathbb{R} : a \le x < b\}only aa
(a,b](a, b]{x∈R:a<x≤b}\{x \in \mathbb{R} : a < x \le b\}only bb

A square bracket marks an endpoint that belongs to the interval, and a round bracket one that does not. The interval [a,b][a, b] is called closed, and (a,b)(a, b) open. When a=ba = b, the interval [a,a][a, a] is {a}\{a\}, and the other three are empty.

The unbounded intervals are

IntervalSet
(a,∞)(a, \infty){x∈R:x>a}\{x \in \mathbb{R} : x > a\}
[a,∞)[a, \infty){x∈R:x≥a}\{x \in \mathbb{R} : x \ge a\}
(−∞,b)(-\infty, b){x∈R:x<b}\{x \in \mathbb{R} : x < b\}
(−∞,b](-\infty, b]{x∈R:x≤b}\{x \in \mathbb{R} : x \le b\}
(−∞,∞)(-\infty, \infty)R\mathbb{R}

The symbols ∞\infty and −∞-\infty are not real numbers. They record that the interval has no end on that side, so the bracket next to them is always round: [a,∞)[a, \infty) is correct, and [a,∞][a, \infty] is not.

The solution set of an inequality is often a union of intervals, written with ∪\cup, such as (−∞,−1]∪(2,∞)(-\infty, -1] \cup (2, \infty). The empty set ∅\varnothing is the solution set of an inequality with no solutions.

Absolute value

The absolute value of a real number xx is

∣x∣={xif x≥0,−xif x<0.\lvert x \rvert = \begin{cases} x & \text{if } x \ge 0, \\ -x & \text{if } x < 0. \end{cases}

So ∣x∣≥0\lvert x \rvert \ge 0 for every xx, and ∣x∣=0\lvert x \rvert = 0 only when x=0x = 0. For real numbers xx and yy, the number ∣x−y∣\lvert x - y \rvert is the distance between xx and yy on the number line. In particular, ∣x∣\lvert x \rvert is the distance from xx to 00.

To work with an expression containing absolute values, split into cases according to the sign of each quantity inside the bars, and apply the definition in each case.

Bounds

Let AA be a subset of R\mathbb{R}.

An upper bound need not belong to AA, and a set that has one upper bound has many: every number larger than an upper bound is also an upper bound.

Supremum and infimum

A real number ss is the supremum, or least upper bound, of AA if

  1. ss is an upper bound of AA, and
  2. s≤us \le u for every upper bound uu of AA.

We then write s=sup⁡As = \sup A. In the language of note 03, sup⁡A\sup A is the minimum of the set of upper bounds of AA, and so a set has at most one supremum.

Similarly, the infimum, or greatest lower bound, inf⁡A\inf A is the maximum of the set of lower bounds of AA, when that maximum exists.

The supremum and infimum of AA need not belong to AA. Compare the maximum and minimum of note 03, which must belong to the set.

When AA is not bounded above. Then AA has no upper bounds at all, so it has no supremum in R\mathbb{R}, and in Tool 05 the answer is “does not exist”. Some texts write sup⁡A=+∞\sup A = +\infty in this case. That is a convention, not a statement about a real number.

An equivalent test. Condition 2 can be replaced by a condition that is often easier to check. If ss is an upper bound of AA, then

s=sup⁡Aif and only iffor every ε>0 there is a∈A with a>s−ε.s = \sup A \quad\text{if and only if}\quad \text{for every } \varepsilon > 0 \text{ there is } a \in A \text{ with } a > s - \varepsilon .

Here is why. If s=sup⁡As = \sup A and ε>0\varepsilon > 0, then s−ε<ss - \varepsilon < s, so s−εs - \varepsilon is not an upper bound, and some a∈Aa \in A exceeds it. Conversely, suppose that the condition holds, and let u<su < s. Taking ε=s−u\varepsilon = s - u gives some a∈Aa \in A with a>ua > u, so uu is not an upper bound. Hence every upper bound is at least ss.

The condition is a statement about every ε>0\varepsilon > 0. A single ε\varepsilon with no suitable aa shows that it fails, but checking finitely many values of ε\varepsilon can never show that it holds. That needs a rule giving a suitable aa for each ε\varepsilon.

Completeness

The four order rules above hold in Q\mathbb{Q} as well as in R\mathbb{R}. What distinguishes R\mathbb{R} is one further axiom.

The completeness axiom. Every nonempty subset of R\mathbb{R} that is bounded above has a supremum.

The axiom asserts existence only. It does not say how to find the supremum, and it says nothing about sets that are empty or not bounded above.

The same question can be asked of Q\mathbb{Q}: does every nonempty set of rational numbers that is bounded above have a least upper bound that is itself rational? Answering it, and tracing which familiar facts about numbers depend on completeness, is the purpose of the last exercise set in Tool 05.