From ℚ to ℝ — why a richer set?
N ⊂ Z ⊂ Q is enough for most calculations, but some very simple geometric lengths don't fit into it.
The limits of ℚ
We have $\mathbb{N} \subset \mathbb{Z} \subset \mathbb{Q}$. These sets meet most computational needs, but some geometric entities fall outside them: the length $x$ of the diagonal of a square with side $1$ satisfies $x^2 = 2$, yet it can be shown that $x \notin \mathbb{Q}$.
$\sqrt{2}$ is irrational
If $x$ is a solution of $x^2 = 2$, then $x \notin \mathbb{Q}$.
Suppose $x = \frac{p}{q}$ with $p\in\mathbb{Z}$, $q\in\mathbb{Z}^*$ coprime. Then $p^2 = 2q^2$, so $p^2$ is even, so $p$ is even: $p = 2k$. Hence $q^2 = 2k^2$, so $q$ is also even — contradicting the fact that $p$ and $q$ are coprime.
Characterization of ℚ
$$x \in \mathbb{Q} \iff x \text{ has a finite or periodic decimal expansion.}$$
Decimal expansions and rational numbers
$\frac{3}{5}=0{,}6$ (finite), $\frac{1}{3}=0{,}333\ldots$ (periodic): these are rational numbers.
For $x = 12{,}34\overline{2021}$ (periodic), we multiply by powers of $10$ to eliminate the periodic part:
Commutative field structure
ℝ is built to satisfy a list of axioms that naturally extend those of ℚ.
Axioms of $(\mathbb{R}, +, \times)$
- Closure: $x+y \in \mathbb{R}$ and $x \times y \in \mathbb{R}$.
- Commutativity, associativity, distributivity of $+$ and $\times$.
- Neutral elements: $\exists\, 0, 1 \in \mathbb{R}$, $x+0=x$ and $x \times 1 = x$.
- Opposites / inverses: $\exists\,(-x)$, $(-x)+x=0$; and for $x\neq 0$, $\exists\, x^{-1}$, $x \times x^{-1}=1$.
We say that $(\mathbb{R},+,\times)$ is a commutative field.
Immediate consequences
- $x=y \iff x-y=0$
- $0 \times x = 0$
- $(-x)y = -(xy) = x(-y)$
- $xy=0 \iff x=0$ or $y=0$
- $(xy)^{-1} = x^{-1}y^{-1}$
Order relation on ℝ
The relation ≤ equips ℝ with a total order compatible with the field operations.
Order relation ≤
- Reflexivity: $x \le y$ or $y \le x$ (in particular $x \le x$).
- Antisymmetry: $x\le y$ and $y \le x \Rightarrow x=y$.
- Transitivity: $x\le y$ and $y\le z \Rightarrow x \le z$.
Compatible with addition: $x \le y \Rightarrow x+z \le y+z$. Partial compatibility with multiplication: $0\le z$ and $x\le y \Rightarrow zx \le zy$.
Useful notation
$x<y$ means $x\le y$ and $x\ne y$. $x \ge y$ means $y \le x$. These can be combined: $x \le y \le z$.
Inequality rules (I)
- $x\ge 0 \iff -x \le 0$
- $x\le y \iff -y\le -x$
- $x\le y \iff x-y\le 0$
- $y\ge 0 \Rightarrow x-y \le x \le x+y$
- $x^2 = xx \ge 0$ for all $x$
- $x<0$ and $y>0 \Rightarrow xy<0$
Inequality rules (II)
- $x\le y$ and $z>0 \Rightarrow xz \le yz$
- $x>0 \Rightarrow x^{-1}=\frac1x>0$
- $x,y\ge0$: $x\le y \iff x^2 \le y^2$
- $0<x<y \Rightarrow 0<\frac1y<\frac1x$
- $(x\le y$ and $s\le t) \Rightarrow x+s\le y+t$
- $(x\ge y>s$ or $x>y\ge s) \Rightarrow x>s$
Intervals of ℝ
- $[a,b]=\{x: a\le x \le b\}$, $\;]a,b[=\{x: a<x<b\}$
- $[a,b[=\{x: a\le x <b\}$, $\;]a,b]=\{x: a<x\le b\}$
- $[a,\infty[$, $\;]a,\infty[$, $\;]-\infty,b]$, $\;]-\infty,b[$
- $\mathbb{R} = \;]-\infty,\infty[\,$, $\;\overline{\mathbb{R}}=[-\infty,\infty]$
- $\mathbb{R}^+=[0,\infty[\,$, $\;\mathbb{R}^{*+}=]0,\infty[\,$, $\;\mathbb{R}^-=]-\infty,0]\,$, $\;\mathbb{R}^{*-}=]-\infty,0[$
Absolute value
The distance to 0 on the real line — a key tool for handling inequalities and approximations.
Absolute value $|x|$
$$|x| = \begin{cases} x & \text{if } x\ge 0 \\ -x & \text{otherwise} \end{cases}$$
We always have $|x| = |-x| = \max(x,-x)$.
Main properties
- $|x|=0 \iff x=0$
- $|xy|=|x||y|$, and $|x|^2=|x^2|=x^2$
- For $y>0$: $|x|\le y \iff -y\le x\le y$
- Triangle inequality: $|x+y|\le|x|+|y|$
- $\big||x|-|y|\big| \le |x-y|$
Max and min via absolute value
Show that for all $x,y \in \mathbb{R}$:
Upper and lower bounds
The property that truly distinguishes ℝ from ℚ: every bounded set has a bound in ℝ.
Upper bound, lower bound, bounded sets
$E$ is bounded below by $m$ if $m\le x$ for all $x\in E$. The greatest lower bound is $\inf E$ (infimum).
$E$ is bounded above by $M$ if $x\le M$ for all $x\in E$. The least upper bound is $\sup E$ (supremum).
$E$ is bounded if it is both bounded below and bounded above.
Bounds of three sets
- $E=\{-3,0,4,10\}$: bounded below by $-3$, bounded above by $\sqrt{101}$.
- $E=\,]{-1},\infty[$: bounded below, $\inf E = -1$; not bounded above.
- $E=\{x\in\mathbb{Q}: x^2<2\}$: bounded above by $\sqrt2$, with $\sqrt2 \notin E$ and $\sqrt2 = \sup E$.
ε-characterization of $\inf E$
ε-characterization of $\sup E$
$$b=\sup E \iff \big(\forall x\in E,\, x\le b\big) \text{ and } \big(\forall \varepsilon>0,\, \exists x_\varepsilon\in E: b-\varepsilon<x_\varepsilon\le b\big)$$
Uniqueness of the bound
When the lower (or upper) bound of a set exists, it is unique.
Passing to the opposite
If $E\subset\mathbb{R}$ is nonempty and bounded below, show that $F=\{-x : x\in E\}$ is nonempty and bounded above, with $\inf F = -\sup E$.
Convention at infinity
If $E$ is nonempty and not bounded above (resp. not bounded below), we set $\sup E = \infty$ (resp. $\inf E = -\infty$).
The founding property of ℝ
Every nonempty, bounded-above set $E$ has a least upper bound in $\mathbb{R}$; every nonempty, bounded-below set has a greatest lower bound in $\mathbb{R}$.
It is this property — not the Archimedean property — that truly characterizes $\mathbb{R}$ relative to $\mathbb{Q}$: $\mathbb{Q}$ is Archimedean but does not satisfy it (e.g. $\{x\in\mathbb{Q}:x^2<2\}$ has no rational least upper bound).
The Archimedean property
A traveler setting out on foot from point A always eventually reaches a point C, in a finite number of steps.
The Archimedean property
$$\text{If } x,y\in\mathbb{R},\ 0<y<x,\ \text{then} \ \exists\, n\in\mathbb{N}: x<ny$$
We say that $\mathbb{R}$ is Archimedean. This property follows from the least upper bound property.
Archimedes alone doesn't characterize ℝ
$\mathbb{Q}$ is Archimedean but does not satisfy the least upper bound property. So it is indeed the least upper bound property that distinguishes $\mathbb{R}$ from $\mathbb{Q}$, not the Archimedean property.
Bounding by integers
For every $x \in \mathbb{R}\setminus\mathbb{Z}$, there exists a unique $n_x \in \mathbb{Z}$ such that:
A direct consequence of the Archimedean property.
Integer part
The integer part places any real number between two consecutive integers.
Integer part $[x]$
For every $x \in \mathbb{R}$, the integer part of $x$, denoted $E(x)$ or $[x]$, is the unique integer satisfying:
Case $x \in \mathbb{Z}$ and $x \notin \mathbb{Z}$
If $x\in\mathbb{Z}$ then $[x]=x$. If $x\notin\mathbb{Z}$, setting $n_x=[x]$, the inequality becomes strict on both sides :
Computing integer parts
$[2]=2$, $\;[2{,}3]=2$, $\;[-2{,}3]=-3$, $\;[\sqrt2]=1$.
Properties of the integer part
- Show that $n=[x] \iff x-1<n\le x$.
- Show that $[x+n]=[x]+n$ for all $n\in\mathbb{Z}$.
- Show that $[x+y]-[x]-[y] \in \{0,1\}$.
Density and approximations
Every real number can be approximated as closely as desired by decimals — and ℚ is dense in ℝ.
Decimal number
$d\in\mathbb{R}$ is decimal if it has a finite decimal expansion: $d\in\mathbb{D} \iff \exists n\in\mathbb{N}: 10^n d \in \mathbb{Z}$. We have $\mathbb{D}\subset\mathbb{Q}$.
E.g.: $11{,}319$ is decimal ($10^3x\in\mathbb{Z}$); $\frac13=0{,}333\ldots$ is not (infinite period).
Decimal approximation to within $10^{-n}$
- Set $x_n = [x\times10^n]\times10^{-n}$.
- This gives the bound $x_n \le x < x_n+10^{-n}$.
- $x_n$: approximation from below ; $x_n+10^{-n}$: approximation from above.
ℚ is dense in ℝ
$$\forall\, x,y\in\mathbb{R},\ x<y,\ \exists\, r\in\mathbb{Q}: x<r<y$$
Take $t=\frac{x+y}{2}$ (rational or not). If it is not rational, approximate it by a decimal $t_n$ close enough (within $10^{-n}$, with $n$ chosen via the Archimedean property) to stay within $]x,y[$.
Corollary of density
Every interval of $\mathbb{R}$ containing at least two reals contains infinitely many rationals and infinitely many irrationals.
Approximating $\sqrt2$ by continued fractions
Method: $[\sqrt2]=1$, then $x_1=\dfrac{1}{\sqrt2-1}=\sqrt2+1$ (by rationalizing), so $[x_1]=2$ — a pattern that repeats indefinitely.
The sequence $(u_n)$ converges rapidly to $\sqrt2 \approx 1{,}41421$.
Approximating $\sqrt7$ by continued fractions
$[\sqrt7]=2$ (since $4<7<9$). The same process gives a periodic pattern of quotients $[1,1,1,4]$, with $u_n^2$ converging rapidly to $7$:
Each step tightens the bound on $\sqrt7$ between two rationals ever closer together.
Approximating $\sqrt{17}$
Using the same method ($[\sqrt{17}]=4$, then iterating $x_{k+1}=\dfrac{1}{x_k-[x_k]}$), find the approximation $u_5$ of $\sqrt{17}$.