Skip to content

The Real Number System

This chapter assumes the reader is comfortable with:

  • Proof techniques: direct proof, proof by contradiction, mathematical induction.
  • Basic set theory: sets, subsets, unions, intersections, the power set.
  • Functions: domain, range, injectivity, surjectivity, composition.
  • Informal calculus: limits, continuity, differentiation, and integration as studied at A-Level or IB Mathematics.

The axioms stated below are taken as given; the purpose of this chapter is to derive their consequences, not to justify the axioms themselves. Readers without a proof-writing background should consult an introductory discrete mathematics text first.

The real numbers R\mathbb{R} form a complete ordered field. The field axioms guarantee closure Under addition, subtraction, multiplication, and division (by non-zero elements), together with the Usual commutative, associative, and distributive laws.

The order relation ≤\leq on R\mathbb{R} satisfies:

  1. Reflexivity: a≤aa \leq a
  2. Antisymmetry: a≤ba \leq b and b≤ab \leq a implies a=ba = b
  3. Transitivity: a≤ba \leq b and b≤cb \leq c implies a≤ca \leq c
  4. Totality: for all a,ba, bEither a≤ba \leq b or b≤ab \leq a
  5. Compatibility with addition: a≤ba \leq b implies a+c≤b+ca + c \leq b + c
  6. Compatibility with multiplication: a≤ba \leq b and 0≤c0 \leq c implies ac≤bcac \leq bc

The completeness axiom (also called the least upper bound property) is what distinguishes R\mathbb{R} from Q\mathbb{Q}:

Axiom (Completeness). Every non-empty subset of R\mathbb{R} that is bounded above has a least Upper bound (supremum) in R\mathbb{R}.

Let S⊆RS \subseteq \mathbb{R} be a non-empty set that is bounded above.

Definition. The supremum (or least upper bound) of SSDenoted sup⁡(S)\sup(S)Is the real number uu satisfying:

  1. uu is an upper bound: s≤us \leq u for all s∈Ss \in S.
  2. uu is the least upper bound: if vv is any upper bound of SSThen u≤vu \leq v.

Similarly, the infimum (or greatest lower bound), inf⁡(S)\inf(S)Is the greatest number ll such that l≤sl \leq s for all s∈Ss \in S.

Proposition 1.1. sup⁡(S)\sup(S) exists if and only if SS is non-empty and bounded above.

Proposition 1.2 (Approximation Property). If u=sup⁡(S)u = \sup(S)Then for every ε>0\varepsilon > 0There Exists s∈Ss \in S such that u−ε<s≤uu - \varepsilon \lt s \leq u.

Proof. If no such ss existed, then u−εu - \varepsilon would be an upper bound of SS strictly less Than uuContradicting the definition of sup⁡(S)\sup(S). ■\blacksquare

Example. Let S={x∈R:x2<2}S = \{x \in \mathbb{R} : x^2 \lt 2\}. Then sup⁡(S)=2\sup(S) = \sqrt{2}. Note that 2∉Q\sqrt{2} \notin \mathbb{Q}, so Q\mathbb{Q} does not satisfy the completeness axiom.

Theorem 1.1 (Archimedean Property). For every x∈Rx \in \mathbb{R}There exists n∈Nn \in \mathbb{N} Such that n>xn \gt x.

Proof. Suppose, for contradiction, that N\mathbb{N} is bounded above. By the completeness axiom, s=sup⁡(N)s = \sup(\mathbb{N}) exists in R\mathbb{R}. Then s−1s - 1 is not an upper bound for N\mathbb{N} So there exists n∈Nn \in \mathbb{N} with n>s−1n \gt s - 1I.e., n+1>sn + 1 \gt s. But n+1∈Nn + 1 \in \mathbb{N} Contradicting that ss is an upper bound. ■\blacksquare

Corollary 1.2. For every ε>0\varepsilon > 0There exists n∈Nn \in \mathbb{N} such that 1/n<ε1/n \lt \varepsilon.

Proof. By the Archimedean property, choose n∈Nn \in \mathbb{N} with n>1/εn \gt 1/\varepsilon. Then 1/n<ε1/n \lt \varepsilon. ■\blacksquare

Corollary 1.3 (Density of Q\mathbb{Q}). Between any two distinct real numbers a<ba \lt bThere Exists a rational number q∈Qq \in \mathbb{Q} with a<q<ba \lt q \lt b.

Proof. Since b−a>0b - a > 0By Corollary 1.2 there exists n∈Nn \in \mathbb{N} with 1/n<b−a1/n \lt b - a So 1<n(b−a)=nb−na1 \lt n(b - a) = nb - na. Let m=⌊na⌋+1∈Zm = \lfloor na \rfloor + 1 \in \mathbb{Z}. Then m−1≤na<mm - 1 \leq na \lt m Giving m≤na+1<na+n(b−a)=nbm \leq na + 1 \lt na + n(b - a) = nb. Hence a<m/n<ba \lt m/n \lt bAnd m/n∈Qm/n \in \mathbb{Q}. ■\blacksquare

Proposition 1.4. If AA and BB are non-empty bounded subsets of R\mathbb{R}Then sup⁡(A+B)=sup⁡(A)+sup⁡(B)\sup(A + B) = \sup(A) + \sup(B)Where A+B={a+b:a∈A,b∈B}A + B = \{a + b : a \in A, b \in B\}.

Proof. For all a∈Aa \in A and b∈Bb \in B: a≤sup⁡(A)a \leq \sup(A) and b≤sup⁡(B)b \leq \sup(B)So a+b≤sup⁡(A)+sup⁡(B)a + b \leq \sup(A) + \sup(B). Thus sup⁡(A)+sup⁡(B)\sup(A) + \sup(B) is an upper bound for A+BA + BSo sup⁡(A+B)≤sup⁡(A)+sup⁡(B)\sup(A + B) \leq \sup(A) + \sup(B).

For the reverse inequality, let ε>0\varepsilon > 0. By the approximation property, there exist a∈Aa \in A And b∈Bb \in B with a>sup⁡(A)−ε/2a > \sup(A) - \varepsilon/2 and b>sup⁡(B)−ε/2b > \sup(B) - \varepsilon/2. Then a+b>sup⁡(A)+sup⁡(B)−εa + b > \sup(A) + \sup(B) - \varepsilonSo sup⁡(A+B)≥sup⁡(A)+sup⁡(B)−ε\sup(A + B) \geq \sup(A) + \sup(B) - \varepsilon. Since ε>0\varepsilon > 0 is arbitrary, sup⁡(A+B)≥sup⁡(A)+sup⁡(B)\sup(A + B) \geq \sup(A) + \sup(B). ■\blacksquare

Proposition 1.5. For any non-empty bounded set S⊆RS \subseteq \mathbb{R}, inf⁡(S)=−sup⁡(−S)\inf(S) = -\sup(-S) Where −S={−s:s∈S}-S = \{-s : s \in S\}.

Proof. Let u=sup⁡(−S)u = \sup(-S). Then −s≤u-s \leq u for all s∈Ss \in SSo s≥−us \geq -u for all s∈Ss \in S Meaning −u-u is a lower bound for SS. If vv is any lower bound for SSThen −v-v is an upper bound For −S-SSo u≤−vu \leq -vI.e., −u≥v-u \geq v. Hence −u=inf⁡(S)-u = \inf(S). ■\blacksquare

Worked Example: Find $\sup$ and $\inf$ of $S = \{(-1)^n + 1/n : n \in \mathbb{N}\}$

Solution. The first few terms are 0,3/2,−2/3,5/4,−4/5,7/6,…0, 3/2, -2/3, 5/4, -4/5, 7/6, \ldots.

For even n=2kn = 2k: (−1)2k+1/(2k)=1+1/(2k)(-1)^{2k} + 1/(2k) = 1 + 1/(2k)Which decreases toward 11 from above. For odd n=2k−1n = 2k-1: (−1)2k−1+1/(2k−1)=−1+1/(2k−1)(-1)^{2k-1} + 1/(2k-1) = -1 + 1/(2k-1)Which increases toward −1-1 from below.

The even terms form the sequence 3/2,5/4,7/6,…3/2, 5/4, 7/6, \ldots with limit 11So sup⁡(S)=3/2\sup(S) = 3/2 (the first even term). The odd terms form 0,−2/3,−4/5,…0, -2/3, -4/5, \ldots with limit −1-1And since 00 Is an odd-indexed term, inf⁡(S)=−1\inf(S) = -1 (approached but not attained). ■\blacksquare

1.6 Construction of R\mathbb{R} via Dedekind Cuts

Section titled “1.6 Construction of R\mathbb{R}R via Dedekind Cuts”

Remark. The following outline shows how R\mathbb{R} can be constructed from Q\mathbb{Q}Making The completeness axiom a theorem rather than an axiom.

Definition (Dedekind Cut). A Dedekind cut is a subset α⊆Q\alpha \subseteq \mathbb{Q} satisfying:

  1. α≠∅\alpha \neq \emptyset and α≠Q\alpha \neq \mathbb{Q}
  2. If p∈αp \in \alpha and q<pq \lt p (with q∈Qq \in \mathbb{Q}), then q∈αq \in \alpha (downward closure)
  3. α\alpha has no greatest element: for every p∈αp \in \alphaThere exists q∈αq \in \alpha with p<qp \lt q

Definition. The set of real numbers R\mathbb{R} is defined as the set of all Dedekind cuts.

The order, addition, and multiplication are defined as follows:

  • Order: α<β\alpha \lt \beta if and only if α⊊β\alpha \subsetneq \beta
  • Addition: α+β={p+q:p∈α,q∈β}\alpha + \beta = \{p + q : p \in \alpha, q \in \beta\}
  • Multiplication: For α,β≥0∗\alpha, \beta \geq 0^*: α⋅β={p⋅q:p∈α,q∈β,p≥0,q≥0}∪{r∈Q:r<0}\alpha \cdot \beta = \{p \cdot q : p \in \alpha, q \in \beta, p \geq 0, q \geq 0\} \cup \{r \in \mathbb{Q} : r \lt 0\}

Here 0∗={q∈Q:q<0}0^* = \{q \in \mathbb{Q} : q \lt 0\} represents the real number 00.

Theorem. With these definitions, R\mathbb{R} is a complete ordered field, and Q\mathbb{Q} embeds Into R\mathbb{R} via q↦{r∈Q:r<q}q \mapsto \{r \in \mathbb{Q} : r \lt q\}.

Proof (sketch). Verifying the field axioms and order axioms is lengthy but straightforward. The key Step is the completeness axiom: if A\mathcal{A} is a non-empty set of Dedekind cuts bounded above, Then α=⋃β∈Aβ\alpha = \bigcup_{\beta \in \mathcal{A}} \beta is itself a Dedekind cut and α=sup⁡(A)\alpha = \sup(\mathcal{A}). ■\blacksquare

The completeness axiom can be formulated in several equivalent ways. Each implies the others:

  1. Least Upper Bound Property: Every non-empty set bounded above has a supremum.
  2. Monotone Convergence Theorem: Every bounded monotone sequence converges.
  3. Nested Interval Property: Every nested sequence of closed intervals I1⊇I2⊇⋯I_1 \supseteq I_2 \supseteq \cdots with length(In)→0\mathrm{length}(I_n) \to 0 has exactly one point in ⋂In\bigcap I_n.
  4. Bolzano-Weierstrass Property: Every bounded sequence has a convergent subsequence.
  5. Cauchy Completeness: Every Cauchy sequence converges.

Proposition 1.6. In any ordered field, (1)   ⟺  \iff (2)   ⟺  \iff (3)   ⟺  \iff (4)   ⟺  \iff (5).

Proof (outline). We have shown (1)⇒(2)(1) \Rightarrow (2) (MCT in Section 2.2), (2)⇒(4)(2) \Rightarrow (4) (via the bisection argument in Bolzano-Weierstrass), (4)⇒(5)(4) \Rightarrow (5) (Cauchy completeness …/1-number-and-algebra/3_proof-and-logic In Section 2.3), and (5)⇒(1)(5) \Rightarrow (1) can be shown by constructing a Cauchy sequence converging To sup⁡S\sup S from the approximation property. The equivalence (1)⇒(3)(1) \Rightarrow (3) follows from the Nested interval argument, and (3)⇒(1)(3) \Rightarrow (1) follows by constructing nested intervals that Shrink to sup⁡S\sup S. ■\blacksquare

Remark. The field Q\mathbb{Q} satisfies none of these properties, which is why it must be Extended to R\mathbb{R} for analysis.

:::caution Common Pitfall The completeness axiom is often misstated as “every bounded set has a supremum.” The set must be Non-empty. Also, completeness does not say every set has a maximum; sup⁡(S)\sup(S) need not belong to SS. For example, sup⁡{1/n:n∈N}=1\sup\{1/n : n \in \mathbb{N}\} = 1Which belongs to the set, but sup⁡(0,1)=1\sup(0, 1) = 1Which does not belong to (0,1)(0, 1).

:::