Lesson 01 · Unit 12 · Differentiation

What a derivative is

A straight line has one gradient. A curve has a different gradient at every point. This lesson builds the tool that measures it, then gives you the six derivatives you must know by heart.

You already know how to find the gradient of a straight line. Two points, rise over run, and the answer is the same wherever you measure it. That is what makes a line straight.

A curve is not like that. On \(y=x^{2}\) the curve is almost flat near the origin and climbing steeply by \(x=3\). So asking “what is the gradient of this curve” is not a fair question. You have to say where.

Measuring something we cannot measure yet

Fix a point \(P(1,1)\) on \(y=x^{2}\). We want the gradient of the curve exactly at P. We cannot do that directly — gradient needs two points, and we only have one.

So take a second point Q further along the curve and join them. The straight line PQ is called a chord, and its gradient we can work out, because now we have two points. Then slide Q toward P and watch what happens.

Interactive The chord closing onto the tangent

Here are the numbers, written out. Each row moves Q closer to P.

Q at \(x=\) \(\delta x\) \(\delta y\) gradient of chord \(=\dfrac{\delta y}{\delta x}\)
2133
1.50.51.252.5
1.10.10.212.1
1.010.010.02012.01
1.0010.0010.0020012.001

The gradients are not settling near 2 by luck. Here is why they must.

Worked example The chord gradient, in algebra

Let Q be the point on \(y=x^{2}\) with \(x = 1+h\), where \(h\) is any small number.

  • \(P = (1,\ 1) \qquad Q = \big(1+h,\ (1+h)^{2}\big)\)both points are on the curve
  • \(\delta x = (1+h) - 1 = h\)the run
  • \(\delta y = (1+h)^{2} - 1 = 1 + 2h + h^{2} - 1 = 2h + h^{2}\)the rise, expanded
  • \(\dfrac{\delta y}{\delta x} = \dfrac{2h + h^{2}}{h} = \dfrac{h(2+h)}{h} = 2 + h\)\(h \neq 0\), so we may cancel

So the gradient of the chord is \(2 + h\) — exactly, for every \(h\). Put \(h = 1\) and you get 3. Put \(h=0.01\) and you get 2.01. The table was never going to say anything else.

Now make \(h\) smaller and smaller. The chord gradient \(2+h\) gets as close to 2 as you like. We write this

\(\dfrac{\delta y}{\delta x} \to 2 \quad \text{as} \quad \delta x \to 0\)

and we say the gradient of the curve at P is 2.

Marks note

You will never be asked to do this in the exam. Differentiation from first principles is not on the 0606 syllabus. It is here so that you know what the number means — because in two lessons’ time you will be asked what a gradient of \(-3\) tells you about a curve, and a student who only learned the recipe cannot answer that.

Marks note

“The gradient of the curve at the point” and “the gradient of the tangent at the point” are the same number. Questions use both wordings and expect you to know they are the same.

Guided The same argument at a different point

Take the same curve \(y=x^{2}\), but this time fix P at \(x=3\), so \(P(3,9)\).

1

Let Q be at \(x=3.5\), so \(\delta x = 0.5\). Find \(\delta y\).

2

Now find the gradient of that chord.

3

Repeat with Q at \(x=3.1\). What is the chord gradient now?

4

In general the chord gradient here is \(6+h\). So what is the gradient of the curve at \(x=3\)?

At \(x=1\) the gradient was 2. At \(x=3\) it is 6. Both are double the \(x\)-value. Hold on to that — it is the whole of the next tab but one.

Fluency Chord gradients on \(y=x^{2}\) 12 questions
  1. On \(y=x^{2}\), the chord from \(x=2\) to \(x=3\). What is its gradient?
  2. On \(y=x^{2}\), the chord from \(x=2\) to \(x=2.5\).
  3. On \(y=x^{2}\), the chord from \(x=2\) to \(x=2.1\).
  4. On \(y=x^{2}\), the chord from \(x=2\) to \(x=2.01\).
  5. Those four are closing in on something. What is the gradient of the tangent at \(x=2\)?
  6. On \(y=x^{2}\), the chord from \(x=5\) to \(x=7\).
  7. On \(y=x^{2}\), the chord from \(x=-1\) to \(x=3\).
  8. On \(y=x^{2}\), the chord from \(x=-2\) to \(x=-1\).
  9. On \(y=x^{2}\), \(x\) goes from \(3\) to \(3.2\). What is \(\delta y\)?
  10. And what is the gradient of that chord?
  11. On \(y=x^{2}\), the chord from \(x=4\) to \(x=4+h\) has gradient \(8+h\). What is it when \(h=0.5\)?
  12. And as \(h\) shrinks towards zero?

Every chord on this curve has gradient equal to the sum of its two \(x\)-values — which is why the last one in each run is so easy to see coming.

Give the gradient of the chord as a decimal. Each one is the difference in \(y\) divided by the difference in \(x\), and nothing more than that.