An Intuitive Introduction to the Derivative (Difference Quotient)

Every idea this block has built so far — limits, asymptotes, continuity — was preparation for a single question: how fast is a function changing at one exact instant? Average rate of change over an interval is easy: divide the change in output by the change in input. But the rate of change at a point, where the interval has shrunk to nothing, seems at first to require dividing by zero. This chapter shows how the limit concept resolves that apparent paradox and defines the derivative.

You will need the limit techniques of Chapter 1 (especially factoring and rationalising to resolve $\frac00$ forms) and the continuity criterion of Chapter 3, since continuity turns out to be a necessary condition for everything built here.

1. Average Rate of Change

For a function $f$, the average rate of change between $x=a$ and $x=a+h$ (for some nonzero $h$) is the familiar "rise over run":

Geometrically, this is the gradient of the secant line — the straight line joining the two points $(a, f(a))$ and $(a+h, f(a+h))$ on the graph of $f$.

Worked example. For $f(x) = x^2$, find the average rate of change between $x=1$ and $x=3$.

This tells us that, on average, $f$ increased by $4$ units of output per unit of input across that interval — but it says nothing about whether $f$ was changing faster at the start or the end of the interval.

2. From Secant to Tangent: Shrinking the Interval

To find the rate of change at a single point $x=a$, rather than across an interval, imagine the second point $x=a+h$ sliding towards $a$, so $h\to0$. The secant line pivots as it does so, and in the limit it settles into the tangent line — the line that just grazes the curve at $(a, f(a))$, matching its direction exactly at that instant.

Definition. The difference quotient of $f$ at $a$ is $\dfrac{f(a+h)-f(a)}{h}$. The derivative of $f$ at $a$, written $f'(a)$, is the limit of the difference quotient as $h\to0$:

$$f'(a) = \lim_{h\to0} \frac{f(a+h)-f(a)}{h}$$

when this limit exists. Geometrically, $f'(a)$ is the gradient of the tangent line to $y=f(x)$ at $x=a$.

The derivative is not a new kind of object invented from nothing — it is exactly the average-rate-of-change calculation from Section 1, with a limit applied to shrink the interval to a single instant. Every technique from Chapter 1 for resolving indeterminate forms is about to earn its keep, because direct substitution of $h=0$ into the difference quotient always gives $\frac00$.

3. Computing a Derivative from the Definition

Worked example. Find $f'(2)$ for $f(x) = x^2$ directly from the definition.

Expand the numerator:

so

The factor of $h$ that caused the $\frac00$ form cancels exactly, which is no coincidence: $h=0$ is always a root of the numerator of a difference quotient, because $f(a+0)-f(a) = 0$. This guarantees a common factor of $h$ exists to cancel, provided $f$ is built from polynomials.

4. An Alternative Form of the Definition

Setting $x = a+h$ (so $h = x-a$, and $h\to0$ corresponds to $x\to a$), the derivative can equivalently be written as

This form makes the connection to Chapter 1 explicit: $f'(a)$ is precisely $\displaystyle\lim_{x\to a}$ of the slope between a general point $x$ and the fixed point $a$ — another $\frac00$ limit, resolved by exactly the factor-and-cancel method practised there.

Worked example. Use this form to find $f'(3)$ for $f(x) = x^2 - 2x$.

Factor the numerator: $x^2-2x-3=(x-3)(x+1)$, so

5. Why Continuity Is Necessary (But Not Sufficient) for Differentiability

Key distinction. If $f'(a)$ exists, $f$ is called differentiable at $a$. Every differentiable function is continuous at that point, but a continuous function need not be differentiable.

The first half follows from the algebra of limits: if $f'(a) = \displaystyle\lim_{h\to0}\frac{f(a+h)-f(a)}{h}$ exists, then

so $\displaystyle\lim_{h\to0}f(a+h) = f(a)$, which is exactly the continuity condition. The converse fails: $f(x)=|x|$ is continuous at $x=0$ (Chapter 3), yet its difference quotient there is $\dfrac{|h|-0}{h} = \dfrac{|h|}{h}$, which equals $1$ for $h>0$ and $-1$ for $h<0$ — a jump, so the limit (and hence $f'(0)$) does not exist. A corner in a graph is continuous but not differentiable: the left-hand and right-hand tangent "directions" disagree.

6. The Derivative as an Instantaneous Rate of Change

Beyond its geometric meaning as a tangent gradient, $f'(a)$ has a physical reading: it is the instantaneous rate of change of $f$ with respect to $x$, at $x=a$ — the value the average rate of change approaches as the interval over which it is measured shrinks to nothing. This is why the derivative of a position function with respect to time gives instantaneous velocity (a theme developed fully in a later chapter of this block): velocity is nothing but the rate of change of position, evaluated at an instant rather than averaged over a stretch of time.

7. Estimating a Derivative Numerically

When an algebraic formula for $f$ is unavailable or unwieldy, $f'(a)$ can be estimated by evaluating the difference quotient for a small, fixed value of $h$ (not a limit, since $h$ is not actually sent to $0$, but a close approximation):

Worked example. Estimate $f'(1)$ for $f(x)=\sqrt{x}$ using $h=0.01$.

This is already close to the exact value $f'(1)=\tfrac12$, found algebraically once the general power rule is established in the next chapter. Smaller $h$ gives a closer estimate, at the cost of needing more decimal precision to avoid rounding error swamping the genuinely small numerator.

Common Exam Pitfalls

• Substituting $h=0$ directly into the difference quotient and stopping at $\frac00$. This is always the starting indeterminate form, never the final answer — algebraic simplification (expansion and cancellation) must happen first.
• Forgetting to expand $(a+h)^2$ or similar fully. A dropped cross term is the most common algebra slip in this entire calculation.
• Treating the two forms of the definition as different results. $\lim_{h\to0}\frac{f(a+h)-f(a)}{h}$ and $\lim_{x\to a}\frac{f(x)-f(a)}{x-a}$ are the same derivative, just with a different substitution for the shrinking interval.
• Believing continuity guarantees differentiability. $f(x)=|x|$ at $x=0$ is the standard counterexample: continuous, but with disagreeing one-sided difference-quotient limits.
• Confusing the average rate of change over an interval with the instantaneous rate at a point. They coincide only if the function's rate of change happens to be constant over that interval (e.g. a straight line).
• Treating a numerical derivative estimate as exact. A finite $h$ gives an approximation whose error shrinks as $h\to0$, not the true limiting value itself.

Summary

• The average rate of change of $f$ from $a$ to $a+h$ is $\dfrac{f(a+h)-f(a)}{h}$, the gradient of the secant line.
• The derivative $f'(a) = \displaystyle\lim_{h\to0}\dfrac{f(a+h)-f(a)}{h} = \displaystyle\lim_{x\to a}\dfrac{f(x)-f(a)}{x-a}$ is the gradient of the tangent line, found by resolving an inevitable $\frac00$ form.
• Computing a derivative from the definition always involves expanding the numerator, factoring out $h$ (or $x-a$), cancelling, then substituting.
• Differentiability implies continuity, but not conversely: $|x|$ at $x=0$ is continuous with a corner, where the one-sided difference quotients disagree.
• The derivative is the instantaneous rate of change, the limiting value the average rate of change approaches as the interval shrinks to a point.
• A derivative can be estimated numerically using a small fixed $h$ when an exact algebraic computation is unavailable, with accuracy improving as $h\to0$.