LeoMath

Foundations of calculus · Section 2 of 5

The derivative

The limit of a rate of change, and why eˣ is its own derivative.

about 9 min

Start from a problem

Problem

An object's position is x(t). Its average velocity from t to t+Δt is

x(t+Δt)−x(t)Δt.

But what we really want to ask is: at the instant t, how fast is it going?

An instant has no length; at Δt=0 the quotient is 00. Yet "instantaneous speed" clearly means something: the speedometer shows it. So we are forced to consider the limit of the average velocity as Δt→0. If that limit exists, it deserves the name instantaneous velocity:

x′(t)=limΔt→0⁡x(t+Δt)−x(t)Δt.

The derivative is not a notation handed down from above; it is the only reasonable answer to the question "what is the rate of change right now?"

Observe

Play first, read later. Below are f(x)=ax and its difference quotient

f(x+h)−f(x)h.

Drag the base a and make the step h small. The difference-quotient curve always looks like ax itself, just taller or shorter. Is there a base a for which the two curves coincide exactly?

Interactive experimentWhy eˣ is its own derivative
ax+h−axh=ax⋅ah−1h
difference quotient ÷ function 0.8284

What did you observe?Play first, then answer. Read the explanation only after you have thought it through.

  1. 1.Fix h and drag a from 1.5 to 4. Does the difference-quotient curve rise or fall relative to the function?
  2. 2.At a = 2 make h small. What number does the ratio in the corner approach? Is it related to ln 2 ≈ 0.693?
  3. 3.Why does the difference-quotient curve always have the shape of aˣ, only taller or shorter, whatever a is?
Conjecture

The difference-quotient curve is a constant multiple of ax; the constant ah−1h depends only on a and h. As h→0 it tends to some number c(a) depending only on a. c(a) increases with a: about 0.69 at a=2, about 1.10 at a=3. So between 2 and 3 there should be a unique a with c(a)=1. We call it e.

First a geometric look: shrink the distance h between two points and the secant turns into the tangent.

Interactive experimentSecant becomes tangent
Function
f(x0+h)−f(x0)hsecant slope -0.0275tangent slope f′(x₀) 0.5403gap 0.5678

What did you observe?Play first, then answer. Read the explanation only after you have thought it through.

  1. 1.Shrink h from 2 to 0.01. How does the gap between secant and tangent slopes change? Does it shrink uniformly?
  2. 2.Choose x² and put x₀ at 0. How does the secant slope depend on h? Can you guess f′(0)?
  3. 3.Choose ln x and drag x₀ very close to 0. What happens to the tangent slope?

Definition

Definition 2.1Derivative

Let f be defined on a neighbourhood of x0. If the limit

f′(x0)=limh→0⁡f(x0+h)−f(x0)h

exists, f is differentiable at x0 and f′(x0) is its derivative there.

Geometrically the difference quotient is the slope of a secant and the derivative the slope of the tangent. Physically, average velocity and instantaneous velocity. The definition contains one limit and nothing else.

Definition 2.2The number e

e is the unique positive real number with limh→0⁡eh−1h=1.

This looks engineered to make (ex)′=ex true, and that is exactly what it is. We now show such a number exists, is unique, and identify it.

Derivation

Proposition 2.1Derivative of a^x

For a>0,

ddxax=ax⋅limh→0⁡ah−1h=axln⁡a.
Proof

The first equality is the factorisation seen in the experiment: ax+h−ax=ax(ah−1).

For the second we need limh→0⁡ah−1h=ln⁡a. Put u=ah−1, so h=loga⁡(1+u)=ln⁡(1+u)ln⁡a and u→0 as h→0. Then

ah−1h=uln⁡aln⁡(1+u)=ln⁡a⋅1ln⁡(1+u)u→ln⁡a,

using limu→0⁡ln⁡(1+u)u=1, which is equivalent to limu→0⁡(1+u)1/u=e, the other common definition e=limn→∞⁡(1+1n)n.

So c(a)=ln⁡a, and c(a)=1⟺a=e. The "coincidence" you saw in the experiment is ln⁡e=1.

Basic rules and their proofs

Theorem 2.2Rules of differentiation

Let f,g be differentiable at x. Then

  1. (f+g)′=f′+g′;
  2. (fg)′=f′g+fg′;
  3. if g(x)≠0, (fg)′=f′g−fg′g2;
  4. (chain rule) if g is differentiable at x and f at g(x), then (f∘g)′(x)=f′(g(x))g′(x).
Proof

We prove the product rule; the method is the same add-and-subtract trick as for products of limits.

f(x+h)g(x+h)−f(x)g(x)h=f(x+h)−f(x)hg(x+h)+f(x)g(x+h)−g(x)h.

As h→0 the first term tends to f′(x)g(x) (differentiability of g implies continuity, so g(x+h)→g(x)) and the second to f(x)g′(x).

Theorem 2.3Differentiable implies continuous

If f is differentiable at x0, then f is continuous at x0.

Proof

f(x0+h)−f(x0)=f(x0+h)−f(x0)h⋅h→f′(x0)⋅0=0.

The converse fails: |x| is continuous at 0 but not differentiable; the one-sided difference quotients tend to −1 and 1.

Common mistake

f′(x0) is a number; f′(x) is a function. In f′(x0)=limh→0⁡f(x0+h)−f(x0)h the point x0 is fixed and h is what moves. Writing "f′(x)=2x" means: for every x, this limit was computed, and it happened to equal 2x.

Applications

ApplicationWhy e appears everywhere

(ex)′=ex says: the rate of change is proportional to the current value, with constant 1. Any process whose growth rate is proportional to its size (compound interest, populations, radioactive decay, a discharging capacitor) is solved by Cekt. This is the first equation, y′=ky, in What is an ODE.

ApplicationTrigonometric functions

From limx→0⁡sin⁡xx=1 and limx→0⁡cos⁡x−1x=0,

sin⁡(x+h)−sin⁡xh=sin⁡xcos⁡h−1h+cos⁡xsin⁡hh→cos⁡x.

Try it in the plotter →

Remark

Observe → conjecture → define → derive → prove → apply. We did not start from "e≈2.718". We first saw that a special base exists, then defined it, then proved it is the base of ln⁡. A definition is the tidying-up of a conclusion, not something handed down from above.

Exercises

0 of 2 done

1
Let f(x)=x3−2x. Find f′(2).

Enter the exact value; a fraction such as 1/3 is fine.

2
Compute limh→0⁡2h−1h to three decimals. This is the derivative of 2x at x=0.

Decimals or fractions; anything within 0.0006 of the answer counts as correct.

Discussion

Found a mistake, have a question or a suggestion? You can also write to 620026600@qq.com

No comments yet. Share your understanding, ask a question, or point out an error on this page.

Log in to join the discussion