← Back to MATH 4630

Continuous Functions

Motivation: Why continuity needs a definition

You have been using the word continuous since precalculus: a function is continuous if you can draw its graph without lifting your pen. This picture is compelling, but it is not a definition — it is an intuition. And intuitions can fail.

Consider the following:

x f c Removable x f c Jump x f c Continuous
Both of the first two functions have a well-defined value at \(c\), and both look "almost" continuous there. What precisely separates them from the third?

We now have a rigorous definition of \(\lim_{x\to c}f(x)\) from the last handout. It is time to use it.

Definition of continuity

Definition

A function f: A → ℝ is continuous at a point c ∈ A if, for all ε > 0, there exists δ > 0 such that whenever |x − c| < δ (and x ∈ A): |f(x) − f(c)| < ε. If f is continuous at every point in A, we say f is continuous on A.

Takeaways: Continuity at \(c\) means three things simultaneously:
  1. \(f(c)\) is defined (\(c\) is in the domain),
  2. \(\lim_{x\to c} f(x)\) exists,
  3. \(\lim_{x\to c} f(x) = f(c)\).

Any one of these failing means \(f\) is discontinuous at \(c\).

x f(x) c f(c)

Reading the picture: Same orange/red band structure as before. The one change: the point \(c\) is now solid on the curve and included in the orange strip. We must have \(f(c)\) land inside the red band — not just nearby points.

Three equivalent characterizations

Just as functional limits had an \(\varepsilon\)–\(\delta\) version and a sequential version, continuity has multiple equivalent formulations. Each is useful in different situations.

Theorem

Let \(f: A \to \mathbb{R}\) and \(c \in A\). The function \(f\) is continuous at \(c\) if and only if any one of the following holds:

  1. (\(\varepsilon\)–\(\delta\) definition): For all \(\varepsilon > 0\), there exists \(\delta > 0\) such that whenever \(|x - c| < \delta\) (and \(x \in A\)): \(|f(x) - f(c)| < \varepsilon\).
  2. (Topological version): For every \(\varepsilon\)-neighborhood \(V_\varepsilon(f(c))\) of \(f(c)\), there exists a \(\delta\)-neighborhood \(V_\delta(c)\) of \(c\) such that for all \(x \in V_\delta(c)\) (with \(x \in A\)): \(f(x) \in V_\varepsilon(f(c))\).
  3. (Sequential): For all sequences \((x_n) \to c\) with \(x_n \in A\): \(f(x_n) \to f(c)\).

If \(c\) is a limit point of \(A\), all three are also equivalent to:

  1. \(\lim_{x\to c} f(x) = f(c)\).
When to use each characterization.
Version Best used for …
\(\varepsilon\)–\(\delta\) Direct proofs that a specific function IS continuous
Sequential (iii) Proving discontinuity: exhibit one bad sequence \((x_n)\to c\) with \(f(x_n)\not\to f(c)\)
(iv) Inheriting results from the last handout (Algebraic Limit Theorem, etc.)
Corollary (Criterion for Discontinuity)

Let f: A → ℝ, c ∈ A a limit point. If there exists a sequence (xₙ) ⊆ A with (xₙ) → c but f(xₙ) ↛ f(c), then f is not continuous at c.

Examples

Example 1. Prove that the function \(f(x) = 3x - 1\) is continuous at every \(c \in \mathbb{R}\).

Scratch work:

Formal proof:

Note: \(\delta = \varepsilon/3\) works for every \(c\in\mathbb{R}\): it does not depend on \(c\) at all. This will become important when we study uniform continuity.

Example 2 (Greatest Integer Function). Consider the function \(h\) defined by \(h(x) = \lfloor x \rfloor.\) Show \(h\) is discontinuous at each \(m\in\mathbb{Z}\), but continuous at each \(c\notin\mathbb{Z}\).
x h(x) 1 2 3 1 2 3

Discontinuity at integers. Let \(m \in \mathbb{Z}\). We want to show \(h = \lfloor x \rfloor\) is not continuous at \(m\) using the sequential criterion.

Continuity away from integers. Let \(c\notin\mathbb{Z}\), so \(n < c < n+1\) for some \(n\in\mathbb{Z}\). Choose \(\delta = \)

Explain why \(|x-c|<\delta\) forces \(h(x) = h(c)\) and hence \(|h(x)-h(c)| < \varepsilon\) for any \(\varepsilon>0\).

Observation: Here \(\delta\) does not depend on \(\varepsilon\) at all — the same \(\delta\) works for every \(\varepsilon>0\).

x h(x) c h(c)
Example 3. Prove that the function \(f(x) = \sqrt{x}\) is continuous on \(A = [0,\infty)\).

Case 1: \(c = 0\).

Case 2: \(c > 0\).

Scratch work:

Using the bound above, what should \(\delta\) be? Then write the formal proof.

Example 4. Consider \(g(x) = \begin{cases} x\sin(1/x) & x \neq 0 \\ 0 & x = 0. \end{cases}\) We saw that \(\lim_{x\to 0}\sin(1/x)\) does not exist. But multiplying by \(x\) kills the oscillation. Is \(g\) continuous at \(0\)?
x g(x) O
Solution:

The Algebraic Continuity Theorem

We do not need to prove continuity from scratch for every function. The Sequential Criterion lets us inherit the Algebraic Limit Theorem:

Theorem

Assume \(f: A\to\mathbb{R}\) and \(g: A\to\mathbb{R}\) are both continuous at \(c\in A\). Then:

  1. \(kf(x)\) is continuous at \(c\) for all \(k\in\mathbb{R}\),
  2. \(f(x)\pm g(x)\) is continuous at \(c\),
  3. \(f(x)\cdot g(x)\) is continuous at \(c\),
  4. \(\frac{f(x)}{g(x)}\) is continuous at \(c\), provided \(g(c)\neq 0\).
Consequence: All polynomials are continuous on \(\mathbb{R}\).

Why? Start from two facts:

  1. The constant function \(f(x)=k\) is continuous everywhere (\(\delta=1\) works for any \(\varepsilon>0\)).
  2. The identity function \(g(x)=x\) is continuous everywhere (\(\delta=\varepsilon\) works).

By the Algebraic Continuity Theorem (repeatedly applying parts (i)–(iii)):

\(p(x) = a_0 + a_1 x + a_2 x^2 + \cdots + a_n x^n\)

is a sum and product of continuous functions, hence continuous on \(\mathbb{R}\).

By part (iv): rational functions are continuous wherever the denominator is nonzero.

Continuity of Compositions

The Algebraic Continuity Theorem handles sums and products. But what about \(\sqrt{3x^2+5}\)? We need to compose two continuous functions. The Algebraic Continuity Theorem does not cover this.

Theorem

Let \(f: A \to \mathbb{R}\) and \(g: B \to \mathbb{R}\) with \(f(A)\subseteq B\). If \(f\) is continuous at \(c\in A\) and \(g\) is continuous at \(f(c)\in B\), then \(g\circ f\) is continuous at \(c\).

Domain A c Domain B f(c) Range of g g(f(c)) f g
Proof sketch: Let \((x_n)\to c\) with \(x_n\in A\).

Since \(f\) is continuous at \(c\): \(f(x_n) \to\)

Since \(g\) is continuous at \(f(c)\) and \(f(x_n)\to f(c)\): \(g(f(x_n)) \to\)

That is: \((g\circ f)(x_n) \to g(f(c)) = (g\circ f)(c)\). By the sequential criterion, \(g\circ f\) is continuous at \(c\). \(\square\)

Example 5. \(h(x) = \sqrt{3x^2+5}\) is continuous on all of \(\mathbb{R}\).

Dirichlet's Function

Define a function \(g: \mathbb{R} \to \mathbb{R}\) as follows:

\(g(x) = \begin{cases}1 & x\in\mathbb{Q}\\ 0 & x\notin\mathbb{Q}.\end{cases}\)

x g(x) 1 rationals irrationals
Fact. Dirichlet's function \(g\) is nowhere continuous on \(\mathbb{R}\): it is discontinuous at every single point of the real line.
Fact. If \(c\) is an isolated point of \(A\), then \(f\) is automatically continuous at \(c\), for any \(f\).

Why?

Activity

Problem 1. Use Definition 1 (the \(\varepsilon\)–\(\delta\) definition of continuity) to prove that \(f(x) = x^2 + 1\) is continuous at \(c = 2\).
Problem 2. True/False: If \(f\) and \(g\) are both discontinuous at \(c\), then \(f+g\) is discontinuous at \(c\).