Continuity

⚡ Key Takeaways

A function is continuous at a point if its graph has no break, jump, or hole there — formally, the limit of the function as you approach the point equals the function's actual value. Key result: $\lim_{x\to a} f(x) = f(a)$ . Applies across calculus, analysis, and physics wherever smooth, unbroken behaviour matters. Rigorously formulated by Cauchy, Bolzano, and Weierstrass in the 19th century.

💡 The Core Insight

Continuity is a local promise about predictability: if you know where a point is heading (its limit), continuity guarantees the function actually arrives there. The whole concept reduces to forcing the predicted value (the limit) and the actual value (the function value) to coincide. Everything else — sums of continuous functions, the IVT, integrability — flows from this single equality.

DefinitionContinuity at a Point

Let $f$ be a function defined on an open interval containing $a$ . We say $f$ is continuous at $a$ if:

1

\quad \lim_{x\to a} f(x) = f(a)

Derivation hint: This compact statement secretly bundles three requirements — that $f(a)$ exists, that the limit exists, and that they match. For this single equation to be valid, all three of the following must hold:

2

\quad f(a)\text{ is defined}, \qquad \lim_{x\to a} f(x)\text{ exists}, \qquad \lim_{x\to a} f(x) = f(a)

Derivation hint: Each clause rules out one failure mode — a hole, a non-existent (or infinite/jumping) limit, and a misplaced value, respectively. Here $x$ is the input variable, $a$ is the point of interest, $f(a)$ is the function's value at $a$ , and $\lim_{x\to a} f(x)$ is the value $f(x)$ approaches as $x$ gets arbitrarily close to $a$ (from both sides).

DefinitionThe ε–δ Definition

The rigorous (Weierstrass) form: $f$ is continuous at $a$ if for every $\varepsilon > 0$ there exists a $\delta > 0$ such that:

3

\quad |x - a| < \delta \;\implies\; |f(x) - f(a)| < \varepsilon

Derivation hint: Read it as a challenge–response game: name any tolerance $\varepsilon$ around the output $f(a)$ ; continuity guarantees you can find an input window $\delta$ that keeps you inside it. Here $\varepsilon$ (epsilon) is the allowed output error and $\delta$ (delta) is the required input closeness. Note that no " $0 < |x-a|$ " is needed, unlike the limit definition, because $x = a$ trivially satisfies the conclusion.

DefinitionContinuity on an Interval

$f$ is continuous on an open interval $(c,d)$ if it is continuous at every point of $(c,d)$ . On a closed interval $[c,d]$ , continuity at the endpoints uses one-sided limits:

4

\quad \lim_{x\to c^+} f(x) = f(c) \qquad\text{and}\qquad \lim_{x\to d^-} f(x) = f(d)

Derivation hint: At an endpoint you can only approach from inside the interval, so the two-sided limit is replaced by the appropriate one-sided limit.

TheoremIntermediate Value Theorem

If $f$ is continuous on the closed interval $[a,b]$ and $N$ is any value strictly between $f(a)$ and $f(b)$ , then there exists at least one $c \in (a,b)$ such that:

5

\quad f(c) = N

Derivation hint: A continuous curve joining a point below $N$ to a point above $N$ cannot teleport across $N$ ; it must cross it.

TheoremExtreme Value Theorem

If $f$ is continuous on a closed, bounded interval $[a,b]$ , then $f$ attains both an absolute maximum value $M$ and an absolute minimum value $m$ somewhere on $[a,b]$ :

6

\quad \exists\, x_{\max}, x_{\min} \in [a,b] \text{ such that } f(x_{\min}) = m \le f(x) \le M = f(x_{\max}) \;\;\forall x \in [a,b]

Derivation hint: Continuity prevents the function from escaping to infinity or sneaking up on a supremum it never reaches, on a closed bounded domain.

ProofSum of Continuous Functions is Continuous

Claim: If $f$ and $g$ are both continuous at $a$ , then $h = f + g$ is continuous at $a$ .

Step 1. Since $f$ is continuous at $a$ , $\lim_{x\to a} f(x) = f(a)$ . Since $g$ is continuous at $a$ , $\lim_{x\to a} g(x) = g(a)$ .

Step 2. Apply the Sum Law for limits, which states the limit of a sum is the sum of the limits (provided both exist):

7

\quad \lim_{x\to a}\big(f(x)+g(x)\big) = \lim_{x\to a} f(x) + \lim_{x\to a} g(x)

Derivation hint: This is itself proved from the ε–δ definition by splitting the total error budget ε into ε/2 for each function.

Step 3. Substitute the known limits from Step 1:

8

\quad \lim_{x\to a}\big(f(x)+g(x)\big) = f(a) + g(a) = h(a)

Derivation hint: Replace each limit with the function's value, then recognise the result as $h$ evaluated at $a$ .

Step 4. Equation [8] is exactly the definition of continuity for $h$ at $a$ . Therefore $h = f+g$ is continuous at $a$ . $\blacksquare$ The same argument structure (using the corresponding limit laws) proves that differences, products, and quotients (where the denominator is nonzero) of continuous functions are continuous.

∎

mathematician · Augustin-Louis Cauchy | 1789–1857

Cauchy was the French mathematician who, in his 1821 Cours d'analyse, gave the first widely-adopted limit-based definition of continuity, writing that an infinitely small increment in $x$ produces an infinitely small increment in $f(x)$ . Though he still leaned on the language of "infinitesimals," his work decisively shifted analysis away from vague geometric intuition toward arithmetical rigor. His foundational program was later sharpened by Bolzano (who had similar ideas slightly earlier but in obscurity) and completed by Weierstrass's ε–δ formalism, giving us the definitions in this article.

🧠 Intuition

Think of continuity through three mental pictures: 1. The pen test. A function is continuous on an interval if you can draw its entire graph there without lifting your pen off the paper. Jumps, holes, and vertical-asymptote blowups all force you to lift the pen. 2. The zoom-in test. Pick the point $a$ . As you demand the output stay within an ever-tighter band of width $\varepsilon$ around $f(a)$ , you can always satisfy this by restricting the input to a small-enough window of width $\delta$ around $a$ . The function "settles down" to $f(a)$ as you approach. 3. The prediction test. Stand at $a$ and ask: "Based only on where the function is heading from nearby points, where do I predict it to be?" That prediction is the limit. Continuity means the function keeps its promise — it actually is where you predicted. The deep point: continuity is about agreement between trend and reality. A discontinuity is a broken promise.

Figure

A continuous function (gold) versus a jump discontinuity at x = a

ExampleChecking Continuity of a Piecewise Function

Given:

f(x) = \begin{cases} 2x + 1 & x \le 3 \\ x^2 - 2 & x > 3 \end{cases}

Is $f$ continuous at $x = 3$ ?

Step 1 — Function value. Use the $x \le 3$ branch: $f(3) = 2(3) + 1 = 7$ . ✓ (defined)

Step 2 — Left-hand limit. Approaching from below uses the same branch:

\lim_{x\to 3^-} f(x) = 2(3) + 1 = 7

Step 3 — Right-hand limit. Approaching from above uses the $x>3$ branch:

\lim_{x\to 3^+} f(x) = (3)^2 - 2 = 9 - 2 = 7

Step 4 — Compare. Left limit $=7$ , right limit $=7$ , so the two-sided limit exists and equals $7$ . Function value is also $7$ . Since $\lim_{x\to 3} f(x) = 7 = f(3)$ :

\boxed{f \text{ is continuous at } x = 3}

ExampleFinding a Root via the Intermediate Value Theorem

Given: Show that $f(x) = x^3 + x - 1$ has a root in the interval $[0, 1]$ , and estimate it.

Step 1 — Confirm continuity. $f$ is a polynomial, hence continuous everywhere, in particular on $[0,1]$ . ✓

Step 2 — Evaluate endpoints.

f(0) = 0^3 + 0 - 1 = -1 \quad(\text{negative})

f(1) = 1^3 + 1 - 1 = 1 \quad(\text{positive})

Step 3 — Apply the IVT. Since $f(0) = -1 < 0 < 1 = f(1)$ , the value $N = 0$ lies between $f(0)$ and $f(1)$ . By the Intermediate Value Theorem there exists $c \in (0,1)$ with $f(c) = 0$ .

Step 4 — Refine by bisection. Test the midpoint:

f(0.5) = 0.125 + 0.5 - 1 = -0.375 \quad(\text{negative})

Root lies in $(0.5, 1)$ . Test $f(0.7) = 0.343 + 0.7 - 1 = 0.043$ (positive). Root lies in $(0.5, 0.7)$ .

\boxed{\text{A root exists, approximately } c \approx 0.68}

⚠️ Common mistake

"Continuous means differentiable." False. Differentiability is strictly stronger than continuity. Every differentiable function is continuous, but not vice versa. The function $f(x) = |x|$ is continuous everywhere yet has no derivative at $x=0$ because of the sharp corner — the slope jumps from $-1$ to $+1$ .

"A function with a hole 'almost' equals its value, so it's basically continuous." No. If $f(a)$ is undefined or set to the wrong value, $f$ is discontinuous at $a$ , full stop. Continuity is binary at each point.

"If a function has a discontinuity, it must be a vertical jump." Not necessarily — discontinuities also include removable holes, infinite (asymptotic) discontinuities, and oscillatory discontinuities like $\sin(1/x)$ near $0$ .

CounterexampleContinuous Everywhere, Differentiable Nowhere

The Weierstrass function, introduced in 1872, is defined by:

9

\quad W(x) = \sum_{n=0}^{\infty} a^n \cos(b^n \pi x)

Derivation hint: With $0<a<1$ the series converges uniformly (geometric bound), giving continuity; with $b$ a large enough odd integer and $ab > 1 + \tfrac{3}{2}\pi$ , the wiggles never smooth out, killing differentiability everywhere. This function is continuous at every real number but has a corner at every single point — you can draw it without lifting your pen, yet it has no well-defined slope anywhere. It shattered the 19th-century intuition that "continuous curves are smooth except at isolated points."

CounterexampleThe Removable Discontinuity

Consider:

g(x) = \frac{x^2 - 4}{x - 2}

For $x \ne 2$ this simplifies to $g(x) = x + 2$ , so $\lim_{x\to 2} g(x) = 4$ . But $g(2)$ is undefined (division by zero). The limit exists; the function value does not. Condition [2] fails, so $g$ is discontinuous at $x = 2$ — a removable discontinuity (a hole), since redefining $g(2) = 4$ would patch it.

connection

Continuity sits at a crossroads of introductory analysis:

It is built directly on top of Limit of a Function — equation [1] is a statement about a limit equalling a value.
It is a necessary precondition for Differentiability: a function must be continuous at a point to be differentiable there (the converse fails).
It guarantees Riemann Integrability: every function continuous on $[a,b]$ is integrable there.
It powers two cornerstone existence theorems, the Intermediate Value Theorem and the Extreme Value Theorem, which underlie root-finding and optimization.
A stronger cousin, Uniform Continuity, demands a single $\delta$ that works across the whole domain, not just locally.

🔬 Where this appears

⏳ History

- c. 300 BCE — Euclid & the Greeks. Implicit geometric notions of "unbroken" magnitudes, but no formal definition; continuity was visual.

- 1665–1700 — Newton & Leibniz. Calculus invented using "flowing" quantities and infinitesimals; continuity assumed intuitively, never rigorously defined.

- 1817 — Bernard Bolzano. Published Rein analytischer Beweis, giving an essentially modern definition of continuity and proving an early form of the IVT — but his work went largely unnoticed.

- 1821 — Augustin-Louis Cauchy. Cours d'analyse defined continuity via limits/infinitesimals, the first influential rigorous treatment, anchoring analysis in arithmetic.

- 1854 — Bernhard Riemann. Connected continuity to integrability, defining the Riemann integral and clarifying which functions can be integrated.

- 1872 — Karl Weierstrass. Published the ε–δ definition and the everywhere-continuous, nowhere-differentiable function, demolishing naive intuition.

- 1872 — Georg Cantor & Richard Dedekind. Rigorous constructions of the real numbers (Cauchy sequences, Dedekind cuts) finally gave continuity a solid foundation — the completeness of

\mathbb{R}

is what makes the IVT true.

- 1906 — Maurice Fréchet. Generalized continuity to abstract metric spaces, paving the way for topology.

- 1914 — Felix Hausdorff. Formalized topological spaces, recasting continuity purely in terms of open sets ("preimages of open sets are open"), the most general modern viewpoint.

open-problem

While continuity for functions $\mathbb{R}\to\mathbb{R}$ is fully settled, deep subtleties persist at higher levels:

Classifying "how" a function fails continuity led to the rich theory of Baire classes and Borel hierarchies; characterizing the set of discontinuities of arbitrary functions (which must be an $F_\sigma$ set) is subtle.
Generalizing continuity to abstract spaces raises questions about which topologies make familiar operations continuous; some natural function spaces have no metric capturing the "right" notion.
The interplay of continuity and measurability (e.g. Lebesgue's theorem that a bounded function is Riemann-integrable iff its discontinuity set has measure zero) sits at the boundary of analysis and is fully understood only with measure theory.
In computability, whether a "continuous" object can be computed to arbitrary precision is an active concern in computable analysis, where some classically continuous functions resist algorithmic evaluation.

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also