BUSINESS CALCULUS • DERIVATIVES

Differentiability vs. Continuity

Understanding why a smooth function must be continuous, but a continuous function need not be smooth.

Historical Context & Motivation

The relationship between continuity and differentiability lies at the heart of calculus, and the distinction between these two properties was not always clear—even to the mathematicians who invented the subject. In the seventeenth century, Isaac Newton and Gottfried Wilhelm Leibniz independently developed calculus under the implicit assumption that every "reasonable" function could be drawn without lifting one's pen and that such a curve would always possess a well-defined tangent line at every point. For more than a century, this assumption went largely unchallenged, because the functions encountered in physics and commerce were overwhelmingly polynomials, exponentials, and trigonometric functions—all of which happen to be both continuous and differentiable everywhere on their domains.

It was not until the nineteenth century that mathematicians began to construct pathological counterexamples showing that continuity alone does not guarantee smoothness. These discoveries forced the mathematical community to develop the rigorous epsilon-delta definitions of limits, continuity, and derivatives that we use today. For students of business calculus, this distinction is far from academic: cost functions with kinks, tax schedules with breakpoints, and option payoff diagrams all illustrate scenarios where a function is continuous but fails to be differentiable at economically significant points.

1684
Leibniz Publishes the First Calculus Paper
Gottfried Wilhelm Leibniz publishes Nova Methodus, introducing differential notation. He implicitly assumes that every curve possessing no breaks will also have a tangent—an assumption that would take nearly two centuries to refute rigorously.
1821
Cauchy Formalizes Continuity
Augustin-Louis Cauchy provides the first rigorous definition of continuity in his Cours d'Analyse, separating the concept from differentiability and laying the groundwork for modern analysis.
1861
Weierstrass's Nowhere-Differentiable Function
Karl Weierstrass presents a function that is continuous everywhere but differentiable nowhere, shattering the intuition that continuity implies differentiability. This landmark example reshapes the foundations of analysis.
1872
Formal ε-δ Definitions Widely Adopted
Weierstrass's epsilon-delta formulation of limits and continuity becomes the standard framework, enabling mathematicians to state precisely under what conditions a function is continuous, differentiable, or neither.
1973
Black–Scholes and Non-Smooth Payoffs
The Black–Scholes option pricing model highlights the practical importance of non-differentiable functions in finance: option payoff functions are continuous but have corner points at the strike price, requiring special analytical treatment.

The central question this lesson addresses is deceptively simple: If a function has no gaps or jumps, does it necessarily have a well-defined slope at every point? As we will see, the answer is no—and understanding exactly why and where differentiability can fail is essential for correctly applying derivative-based optimization in business contexts.

Core Principles & Definitions

Before exploring the relationship between continuity and differentiability, we need precise definitions of both concepts. These definitions rely on the notion of a limit, which you should already be familiar with from earlier coursework. The key insight is that continuity is a statement about a function's value agreeing with a limit, while differentiability is a statement about a particular ratio of changes having a limit. Differentiability is therefore a strictly stronger requirement.

1

Continuity at a Point

A function f is continuous at x = a if three conditions hold: f(a) is defined, lim(x→a) f(x) exists, and lim(x→a) f(x) = f(a). Informally, you can draw the graph through x = a without lifting your pen.
2

Differentiability at a Point

A function f is differentiable at x = a if the limit of the difference quotient exists: lim(h→0) [f(a + h) − f(a)] / h is a finite number. Geometrically, this means the graph has a unique, non-vertical tangent line at that point.
3

The Implication Theorem

If f is differentiable at x = a, then f is continuous at x = a. The converse is false: continuity does not guarantee differentiability. This one-way relationship is the central theme of this lesson.
4

Three Ways Differentiability Fails

A continuous function can fail to be differentiable at a point if the graph has a corner (sharp turn), a cusp (pointed tip), or a vertical tangent.
KEY TAKEAWAY
Think of continuity and differentiability like the requirements for a road. A continuous road has no gaps—you can drive along it without your car leaving the pavement. A differentiable road is not only gap-free but also has no sharp corners—your steering wheel turns smoothly at every point. Every smooth road is gap-free (differentiability implies continuity), but a road with a sharp 90-degree turn is gap-free without being smooth (continuity does not imply differentiability).

Visual Explanation

The diagram below illustrates the one-way relationship between differentiability and continuity. On the left, you see a function that is both continuous and differentiable—the graph is an unbroken, smooth curve with a well-defined tangent line at the highlighted point. On the right, three classic failure modes are shown: a corner, a cusp, and a vertical tangent. Each of these functions is continuous at the marked point (no gap in the graph), but none is differentiable there.

Top row: four function behaviors at a single point. The first panel (cyan) shows a smooth, differentiable curve with a unique tangent line. The remaining three panels show continuous functions that fail differentiability due to a corner, cusp, or vertical tangent. Bottom: the implication arrow flows one way—differentiability implies continuity, but not the reverse.

Notice that in every failure case, the graph itself has no break—you can trace along the curve through the marked point without interruption. This confirms that each function is continuous at that point. However, if you try to draw a single tangent line at the marked point, you encounter a problem: at a corner, the left-hand slope and right-hand slope disagree; at a cusp, one or both one-sided slopes blow up to ±∞; and at a vertical tangent, the slope is infinite. In each scenario, the difference quotient limit fails to produce a finite number, so the function is not differentiable.

Mathematical Framework

Let us formalize the definitions and prove the key theorem connecting differentiability and continuity. Throughout this section, we assume that f is a real-valued function defined on an open interval containing the point x = a.

CONTINUITY DEFINITION
f is continuous at x = a ⟺ lim(x→a) f(x) = f(a)
This single equation actually encodes three requirements: (1) f(a) exists, (2) lim(x→a) f(x) exists, and (3) the two are equal. Failure of any one condition means discontinuity at a.
DIFFERENTIABILITY DEFINITION
f'(a) = lim(h→0) [f(a + h) − f(a)] / h
The derivative f'(a) exists if and only if this limit is a finite real number. If the limit is ±∞ or does not exist (e.g., the left-hand and right-hand limits differ), then f is not differentiable at a.

Proof: Differentiability Implies Continuity

Suppose f is differentiable at x = a, so that f'(a) = lim(h→0) [f(a + h) − f(a)] / h exists as a finite number. We want to show that lim(h→0) f(a + h) = f(a), which is equivalent to saying f is continuous at a. The key algebraic trick is to write the numerator of the difference quotient by itself:

ALGEBRAIC MANIPULATION
f(a + h) − f(a) = [f(a + h) − f(a)] / h × h
We have simply multiplied and divided by h (valid for h ≠ 0). Now take the limit as h → 0 on both sides.
TAKING THE LIMIT
lim(h→0) [f(a + h) − f(a)] = lim(h→0) {[f(a + h) − f(a)] / h} × lim(h→0) h = f'(a) × 0 = 0
Since f'(a) is finite and lim(h→0) h = 0, the product is zero. Therefore lim(h→0) f(a + h) = f(a), which is exactly the definition of continuity at a. Q.E.D.
⚠️ Why the Converse Fails
The proof above relies critically on f'(a) being a finite number. If we only know that f is continuous, we have no information about whether the difference quotient limit exists. Consider f(x) = |x| at x = 0: the function is continuous (lim(x→0) |x| = 0 = f(0)), but the left-hand slope is −1 while the right-hand slope is +1, so the two-sided limit of the difference quotient does not exist.

Classifying Non-Differentiable Points

When analyzing a function in a business calculus context—say, a piecewise cost function or a demand curve—it is important to identify exactly where and why differentiability fails. The three classic failure modes each arise from a distinct behavior of the one-sided derivatives. The diagram and table below provide a systematic classification.

Three graphs showing the classic non-differentiable points, each annotated with the one-sided slopes. At a corner, both one-sided derivatives exist but differ. At a cusp, they diverge to infinity with opposite signs. At a vertical tangent, they diverge to infinity with the same sign.
Common non-differentiable point types with business contexts
Failure TypeExample FunctionBusiness Application
Cornerf(x) = |x|, piecewise-linear functionsTax brackets, tiered pricing, shipping cost functions with breakpoints
Cuspf(x) = x2/3Power utility functions, certain production functions near boundary inputs
Vertical Tangentf(x) = x1/3Supply curves with vertical asymptotes, marginal cost near capacity limits

Worked Example: Piecewise Cost Function

A company's total shipping cost (in dollars) for sending q units is modeled by the piecewise function:

PIECEWISE COST FUNCTION
C(q) = { 5q + 20, if 0 ≤ q ≤ 100 | 3q + 220, if q > 100 }
The company offers a reduced per-unit rate for orders exceeding 100 units. We want to determine whether C(q) is (a) continuous and (b) differentiable at q = 100.
Testing Continuity and Differentiability at q = 100
1
Step 1 — Evaluate C(100) from the first pieceSince q = 100 falls in the domain of the first piece (0 ≤ q ≤ 100), we compute C(100) = 5(100) + 20 = 500 + 20.
C(100) = 520
2
Step 2 — Compute the right-hand limitAs q approaches 100 from the right (q > 100), we use the second piece: lim(q→100⁺) C(q) = lim(q→100⁺) (3q + 220) = 3(100) + 220 = 300 + 220.
lim(q→100⁺) C(q) = 520
3
Step 3 — Compute the left-hand limitAs q approaches 100 from the left (q < 100), we use the first piece: lim(q→100⁻) C(q) = lim(q→100⁻) (5q + 20) = 5(100) + 20 = 520. Since both one-sided limits equal C(100) = 520, the function is continuous at q = 100.
Continuous at q = 100? YES ✓
4
Step 4 — Compute the left-hand derivativeThe derivative of the first piece is C'(q) = 5 for all q in (0, 100). Therefore the left-hand derivative at q = 100 is: lim(h→0⁻) [C(100 + h) − C(100)] / h = 5. This represents a marginal shipping cost of $5 per unit.
C'(100⁻) = 5
5
Step 5 — Compute the right-hand derivativeThe derivative of the second piece is C'(q) = 3 for all q > 100. Therefore the right-hand derivative at q = 100 is: lim(h→0⁺) [C(100 + h) − C(100)] / h = 3. The marginal cost drops to $3 per unit once the volume discount kicks in.
C'(100⁺) = 3
6
Step 6 — Conclusion on differentiabilitySince C'(100⁻) = 5 ≠ 3 = C'(100⁺), the two one-sided derivatives do not agree. The difference quotient limit does not exist at q = 100. The function has a corner at this breakpoint—the graph changes slope abruptly from 5 to 3.
Differentiable at q = 100? NO ✗ (corner point)
💼 Business Interpretation
The non-differentiability at q = 100 has a concrete meaning: the marginal cost (rate of change of total cost per additional unit) is not well-defined at the exact breakpoint. Below 100 units the marginal cost is $5; above 100 units it drops to $3. At q = 100 itself, the instantaneous rate of change is ambiguous—a fact that matters when setting up optimization models or using derivative-based decision rules.

Continuity vs. Differentiability: Side-by-Side

Having explored both concepts individually, it is helpful to consolidate the comparison into a single reference table. The table below highlights the key dimensions along which continuity and differentiability differ, from their geometric meaning to their algebraic tests and their implications for business modeling.

Continuity vs. differentiability across six key dimensions
DimensionContinuityDifferentiability
Informal meaningNo gaps, jumps, or holes in the graphThe graph is smooth—has a unique, non-vertical tangent at each point
Limit testlim(x→a) f(x) = f(a)lim(h→0) [f(a+h) − f(a)] / h exists and is finite
StrengthWeaker conditionStronger condition (implies continuity)
Failure examplesJump discontinuities, removable holes, vertical asymptotesCorners, cusps, vertical tangents (plus all discontinuities)
Business contextEnsures small input changes produce small output changes—stability in modelsEnables marginal analysis: computing marginal cost, marginal revenue, elasticity
Optimization roleNeeded for the Extreme Value Theorem (max/min on closed intervals)Needed for the first-derivative test and critical-point analysis
KEY TAKEAWAY
In business modeling, differentiability is like having a speedometer that works at every moment—it lets you read the instantaneous rate of change. Continuity is the weaker guarantee that the car's position doesn't teleport; it's necessary for the speedometer to exist, but it's not sufficient. When you encounter a piecewise-defined cost or revenue function, always check the breakpoints: continuity ensures the function doesn't jump, while differentiability ensures the marginal rate is well-defined. If differentiability fails, you can still optimize—but you must treat those points as separate critical points in your analysis.

Connections to Advanced Theory

The distinction between continuity and differentiability serves as a gateway to several deeper ideas that arise in more advanced business mathematics and economics courses. Understanding this distinction now will prepare you for the richer analytical frameworks you will encounter later, from higher-order smoothness conditions in optimization to the handling of non-smooth functions in operations research.

How this lesson's concepts extend into advanced coursework
This LessonAdvanced ExtensionWhere It Appears
Differentiability implies continuityContinuously differentiable (C¹) functions: f' exists and is itself continuousSecond-order conditions in constrained optimization (Lagrange multipliers)
Corner points where f' does not existSubdifferential calculus: replacing derivatives with sets of 'generalized slopes' at non-smooth pointsLinear programming, piecewise-linear models in supply-chain optimization
Checking one-sided derivativesDirectional derivatives and gradients in multivariable calculusMultivariate profit maximization, sensitivity analysis in econometrics
Piecewise functions with breakpointsSpline interpolation: constructing smooth curves that pass through data pointsFinancial curve fitting, yield-curve construction in fixed income

In particular, many optimization algorithms used in machine learning and data science—such as gradient descent—require the objective function to be differentiable (or at least to possess subgradients at non-smooth points). The theoretical foundations you are building in this lesson—understanding exactly what differentiability demands beyond mere continuity—are the same foundations that support modern quantitative methods in business analytics. As you move into courses on mathematical economics or financial engineering, you will see that the hierarchy of smoothness conditions (continuous → differentiable → twice-differentiable → infinitely differentiable) progressively unlocks more powerful analytical tools.

Practice Problems

PROBLEM 1CONCEPTUAL
A student claims: 'If a function is continuous at x = 3, then it must be differentiable at x = 3.' Is this statement true or false? Explain your reasoning, and provide a specific counterexample if the statement is false.
PROBLEM 2BASIC CALCULATION
Let f(x) = { 2x² + 1, if x ≤ 1 | 4x − 1, if x > 1 }. Determine whether f is (a) continuous and (b) differentiable at x = 1.
PROBLEM 3INTERMEDIATE
A company's revenue function is R(q) = { 10q, if 0 ≤ q ≤ 50 | −0.1(q − 50)² + 500, if q > 50 }. (a) Show that R is continuous at q = 50. (b) Determine whether R is differentiable at q = 50. (c) Interpret the result in terms of marginal revenue.
PROBLEM 4APPLIED
A tax function is defined as T(x) = { 0.10x, if 0 ≤ x ≤ 10,000 | 0.20x − 1,000, if x > 10,000 }, where x is taxable income. (a) Verify that T is continuous at x = 10,000. (b) Is T differentiable at x = 10,000? (c) A financial planner wants to compute the instantaneous marginal tax rate at exactly x = 10,000. What should she report, and why?
PROBLEM 5CRITICAL THINKING
Consider the function f(x) = { ax² + bx, if x ≤ 2 | 3x + 2, if x > 2 }. Find the values of the constants a and b such that f is differentiable at x = 2. Then prove that your solution also guarantees continuity at x = 2 (without checking the continuity conditions separately).

Lesson Summary

This lesson explored the fundamental distinction between continuity and differentiability in the context of business calculus. A function is continuous at a point if its limit equals its value there—geometrically, the graph has no gaps. A function is differentiable at a point if the difference quotient has a finite limit—geometrically, the graph has a unique, non-vertical tangent line. The central theorem is a one-way implication: differentiability implies continuity, but the converse is false. A continuous function can fail to be differentiable at corners (where one-sided derivatives disagree), cusps (where one-sided derivatives diverge with opposite signs), and vertical tangents (where the slope is infinite).

In business applications, piecewise-defined cost, revenue, and tax functions frequently have corner points at breakpoints such as quantity thresholds or tax brackets. At these points, the function is continuous (no jump in total cost or revenue), but the marginal rate changes abruptly, so differentiability fails. When performing optimization, these non-differentiable points must be treated as additional critical points alongside the zeros of the derivative. The proof that differentiability implies continuity—based on factoring the numerator of the difference quotient and applying the product rule for limits—is a foundational result that you should be able to reproduce and apply to verify whether specific piecewise functions admit well-defined marginal analyses at their transition points.

Varsity Tutors • Business Calculus • Differentiability vs. Continuity