Absolute Value and the Real Line
Absolute Value and the Real Line
The absolute value $\lvert x \rvert$ is one of the most used tools in all of analysis — yet its power rests on just four simple properties. In this post we define it carefully from the ordered field structure of $\mathbb{R}$, prove all four properties by case analysis, derive the celebrated Triangle Inequality and its reverse form, and interpret $\lvert x - y \rvert$ as the distance between two points on the real line.
Definition of Absolute Value
For $x \in \mathbb{R}$, the absolute value of $x$ is:
\(\lvert x \rvert = \begin{cases} x & \text{if } x \geq 0, \\ -x & \text{if } x < 0. \end{cases}\)
Equivalently, $\lvert x \rvert = \max{x, -x}$.
Geometric meaning: $\lvert x \rvert$ is the distance of $x$ from $0$ on the real line. More generally, $\lvert x - y \rvert$ is the distance between $x$ and $y$: we write $d(x,y) = \lvert x - y \rvert$.
$\lvert x \rvert$ = distance from $x$ to $0$; $\lvert x - y \rvert$ = distance between $x$ and $y$
📖 Reference: Rudin, W., Principles of Mathematical Analysis, 3rd ed., Ch. 1, Definition 1.32 and Theorem 1.33. Also: Apostol, T.M., Mathematical Analysis, 2nd ed., Ch. 1, §1.11–1.12.
← Order Relations and Ordered Sets — $\lvert x \rvert$ is defined using the order on $\mathbb{R}$; sign of $x$ determines which case applies.
← Consequences of Field Axioms — sign rules $(-a)(-b)=ab$ and $-(-a)=a$ are used in the proofs.
← Logic and Proof Methods — case-split proofs and mathematical induction are the main techniques here.
Properties, Intuition, and History
The four properties of $\lvert \cdot \rvert$ can be remembered through the geometric picture: $\lvert x \rvert$ is a length (always non-negative), zero only when the point is at the origin, multiplication scales lengths, and the triangle inequality says the “straight-line” path is never longer than the two-step path through an intermediate point.
Triangle Inequality (P4): $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$. Proof: Since $-\lvert x \rvert \leq x \leq \lvert x \rvert$ and $-\lvert y \rvert \leq y \leq \lvert y \rvert$, adding gives $-(\lvert x \rvert + \lvert y \rvert) \leq x+y \leq \lvert x \rvert + \lvert y \rvert$, which is exactly $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$.
Reverse Triangle Inequality: $\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x - y \rvert$. Proof: Apply (P4) to $x = (x-y)+y$: $\lvert x \rvert \leq \lvert x-y \rvert + \lvert y \rvert$, so $\lvert x \rvert - \lvert y \rvert \leq \lvert x-y \rvert$. Swapping $x,y$: $\lvert y \rvert - \lvert x \rvert \leq \lvert x-y \rvert$. Together these give the result.
Key usage: (P4) gives an upper bound for $\lvert x+y \rvert$; the reverse form gives a lower bound for $\lvert x-y \rvert$. Both are essential in analysis proofs.
Origin Story. The vertical-bar notation $\lvert x \rvert$ was introduced by Karl Weierstrass in the 1840s as part of his systematic programme to arithmetise analysis — replacing geometric intuition with precise algebraic inequalities. Augustin-Louis Cauchy had used absolute values informally in his 1821 Cours d’Analyse to state convergence conditions, but without a clean notation or an axiomatic treatment.
The Problem It Solved. Before Weierstrass, mathematicians relied on vague phrases like “a quantity becomes small” to describe convergence. The absolute value gave a precise language: “$f(x)$ is close to $L$” became $\lvert f(x) - L \rvert < \varepsilon$ — making the $\varepsilon$-$\delta$ definition of a limit possible.
Interesting Fact. The Triangle Inequality became the defining axiom of a norm when Stefan Banach developed his theory of complete normed spaces (now called Banach spaces) in 1922. Every result proved here for $\lvert \cdot \rvert$ on $\mathbb{R}$ generalises directly to norms on infinite-dimensional function spaces — the backbone of modern functional analysis.
Why It Matters Today. Every $\varepsilon$-$\delta$ proof in real analysis, every convergence argument in sequences and series, and every continuity proof ultimately reduces to the Triangle Inequality — making it the single most applied inequality in mathematics.
परम मान (Absolute Value) $\lvert x \rvert$: $x \geq 0$ हो तो $\lvert x \rvert = x$; $x < 0$ हो तो $\lvert x \rvert = -x$। यह $x$ की $0$ से दूरी है। चार गुण: (P1) $\lvert x \rvert \geq 0$; (P2) $\lvert x \rvert = 0 \Leftrightarrow x=0$; (P3) $\lvert xy \rvert = \lvert x \rvert \lvert y \rvert$; (P4) त्रिभुज असमिका $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$। उत्क्रम त्रिभुज असमिका: $\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x-y \rvert$। दूरी: $d(x,y) = \lvert x-y \rvert$।
Solved Examples
$\lvert -7 \rvert = -(-7) = 7$ (since $-7 < 0$).
$\lvert 3-8 \rvert = \lvert -5 \rvert = 5$.
$\lvert (-3)(4) \rvert = \lvert -12 \rvert = 12$.
Verify (P3): $\lvert -3 \rvert \cdot \lvert 4 \rvert = 3 \cdot 4 = 12 = \lvert (-3)(4) \rvert$. $\checkmark$ $\blacksquare$
Rewrite: $(x+y)-2 = (x-3)+(y+1)$.
Apply (P4): $\lvert (x+y)-2 \rvert = \lvert (x-3)+(y+1) \rvert \leq \lvert x-3 \rvert + \lvert y+1 \rvert \leq 1 + 2 = 3$.
Therefore $\lvert (x+y)-2 \rvert \leq 3$. $\blacksquare$
Key technique: always decompose the expression inside $\lvert \cdot \rvert$ into pieces whose absolute values you know.
Step 1. Factor: $\lvert x^2-y^2 \rvert = \lvert (x-y)(x+y) \rvert = \lvert x-y \rvert \cdot \lvert x+y \rvert$ by (P3).
Step 2. By the Reverse Triangle Inequality: $\lvert x-y \rvert \geq \bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert$.
Step 3. By the Triangle Inequality: $\lvert x+y \rvert \geq \bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert$ (apply reverse TI to $\lvert x+y \rvert$ and $\lvert -y \rvert$: $\lvert x+y \rvert \geq \lvert \lvert x \rvert - \lvert y \rvert \rvert$).
Step 4. Multiply: $\lvert x^2-y^2 \rvert = \lvert x-y \rvert \cdot \lvert x+y \rvert \geq \bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert^2 = \bigl(\lvert x \rvert - \lvert y \rvert \bigr)^2$. $\blacksquare$
Claim: $\displaystyle\Bigl\lvert \sum_{k=1}^{n} a_k \Bigr\rvert \leq \sum_{k=1}^{n} \lvert a_k \rvert$ for all $a_1,\ldots,a_n \in \mathbb{R}$, $n \geq 1$.
Base case $n=1$: $\lvert a_1 \rvert \leq \lvert a_1 \rvert$. $\checkmark$
Inductive step: Assume the inequality holds for $n=m$. Then:
$$\Bigl\lvert \sum_{k=1}^{m+1} a_k \Bigr\rvert = \Bigl\lvert \sum_{k=1}^{m} a_k + a_{m+1} \Bigr\rvert \leq \Bigl\lvert \sum_{k=1}^{m} a_k \Bigr\rvert + \lvert a_{m+1} \rvert \leq \sum_{k=1}^{m} \lvert a_k \rvert + \lvert a_{m+1} \rvert = \sum_{k=1}^{m+1} \lvert a_k \rvert.$$
By induction, the inequality holds for all $n \geq 1$. $\square$
Bound: If $\lvert a_k \rvert \leq M$ for all $k$, then $\displaystyle\Bigl\lvert \sum_{k=1}^{n} a_k \Bigr\rvert \leq \sum_{k=1}^{n} \lvert a_k \rvert \leq \sum_{k=1}^{n} M = nM$. $\blacksquare$
Quick Revision Cards
📊 A — Key Properties & Formulae
- (P1) $\lvert x \rvert \geq 0$
- (P2) $\lvert x \rvert = 0 \Leftrightarrow x=0$
- (P3) $\lvert xy \rvert = \lvert x \rvert \lvert y \rvert$
- (P4) $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$
- Rev. TI: $\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x-y \rvert$
- $d(x,y) = \lvert x-y \rvert \geq 0$
- Gen. TI: $\bigl\lvert \sum a_k \bigr\rvert \leq \sum \lvert a_k \rvert$
⚙️ B — Edge Cases & Conditions
- $\sqrt{x^2} = \lvert x \rvert$ (NOT $x$)
- $\lvert x+y \rvert = \lvert x \rvert + \lvert y \rvert$ iff $xy \geq 0$
- $\lvert x \rvert \leq M \Leftrightarrow -M \leq x \leq M$ ($M>0$)
- $\lvert x/y \rvert = \lvert x \rvert / \lvert y \rvert$ for $y \neq 0$
- TI gives upper bound; Rev. TI gives lower bound
- $\lvert x-y \rvert = \lvert y-x \rvert$ (distance symmetric)
🎯 C — Exam Tips
- 🔵 CSIR NET: Bound $\lvert f(x) \rvert$: split into pieces, apply TI to each
- 🟢 GATE: Know both forms of Rev. TI separately
- 🟠 IIT JAM: Gen. TI with bound $nM$ — standard estimation technique
- 🔴 B.Sc. Raj.: Case-split (P3) proof: cover all 4 sign combinations
- All: $\sqrt{x^2}=\lvert x \rvert$ is a classic MCQ trap
Common Mistakes
❌ Errors to Avoid
Wrong: "$\sqrt{x^2} = x$ for all $x \in \mathbb{R}$." | Correct: $\sqrt{x^2} = \lvert x \rvert$. This equals $x$ only when $x \geq 0$. For $x = -3$: $\sqrt{(-3)^2} = \sqrt{9} = 3 = \lvert -3 \rvert \neq -3$.
Wrong: "$\lvert x+y \rvert \geq \lvert x \rvert + \lvert y \rvert$." | Correct: The Triangle Inequality gives $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$ (upper bound). The reverse inequality $\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x-y \rvert$ is a lower bound on $\lvert x-y \rvert$, not on $\lvert x+y \rvert$.
Wrong: Using equality in the Triangle Inequality without checking. | Correct: Equality holds if and only if $xy \geq 0$ (same sign or at least one is zero). Counterexample: $\lvert 1+(-1) \rvert = 0 \neq 2 = \lvert 1 \rvert + \lvert -1 \rvert$.
Wrong: Checking only $x \geq 0, y \geq 0$ and concluding $\lvert xy \rvert = \lvert x \rvert \lvert y \rvert$ always. | Correct: All four cases must be handled: $(+,+)$, $(+,-)$, $(-,+)$, $(-,-)$. Each yields a different expression for $\lvert xy \rvert$ and $\lvert x \rvert \lvert y \rvert$ that must be verified separately.
Real-World Applications
Signal Processing
Total distortion after $n$ processing stages is bounded by $\sum \lvert e_k \rvert$ — a direct application of the generalised Triangle Inequality for bounding cumulative error.
Numerical Analysis
Each floating-point operation introduces rounding error $\lvert \delta_k \rvert \leq \varepsilon$. The Triangle Inequality gives the absolute bound $\leq n\varepsilon$ after $n$ operations.
Metric Spaces & Topology
$d(x,y) = \lvert x-y \rvert$ makes $\mathbb{R}$ a metric space. The Triangle Inequality for $d$ is exactly (P4). All convergence and continuity definitions on $\mathbb{R}$ use this distance.
LASSO Regression
The $L^1$ norm $\sum \lvert \beta_k \rvert$ is used in LASSO to penalise model complexity. Its non-differentiability at $0$ (unlike $\lvert \cdot \rvert^2$) is what promotes sparse solutions.
Summary Table & Key Result
| Property | Statement | Equality condition | Reference |
|---|---|---|---|
| (P1) Non-neg. | $\lvert x \rvert \geq 0$ | $\lvert x \rvert = 0 \Leftrightarrow x=0$ | Rudin 1.33 |
| (P3) Product | $\lvert xy \rvert = \lvert x \rvert \lvert y \rvert$ | Always | Rudin 1.33 |
| (P4) Triangle | $\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$ | iff $xy \geq 0$ | Rudin 1.33 |
| Reverse TI | $\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x-y \rvert$ | iff $xy \geq 0$ | Derived |
| Gen. TI | $\bigl\lvert \sum_{k=1}^n a_k \bigr\rvert \leq \sum_{k=1}^n \lvert a_k \rvert$ | By induction | Example 4 |
| Distance | $d(x,y) = \lvert x-y \rvert$ | $d(x,y)=0 \Leftrightarrow x=y$ | Remark |
Triangle Inequality: $\;\lvert x+y \rvert \leq \lvert x \rvert + \lvert y \rvert$
Reverse Triangle Inequality: $\;\bigl\lvert \lvert x \rvert - \lvert y \rvert \bigr\rvert \leq \lvert x - y \rvert$
Both follow from the order structure of $\mathbb{R}$ and are the foundation of every convergence and continuity argument in analysis.
Cross-References & Related Posts
← Prerequisites: Order Relations and Ordered Sets — $\lvert x \rvert$ is defined using the order on $\mathbb{R}$ | Consequences of Field Axioms — sign rules used in proofs | Logic and Proof Methods — case-split and induction proofs.
→ Next Topic: Bounded Sets, Supremum and Infimum — absolute values appear throughout LUB arguments and the Archimedean property.
📖 Further Reading: Rudin, Ch. 1, §§1.32–1.38; Apostol, Ch. 1, §1.11–1.12; Bartle & Sherbert, Ch. 2, §2.2.
How did you find this post?
Tap a reaction — counts are saved permanently on our server.
Have a question, doubt, or thought about this post? Choose how you want to join the discussion below.
💬 Comment on Telegram — No account needed