Derivatives and Slope
A derivative answers one question: how fast is this changing right here?
A road climbs out of a valley along a curve: nearly flat at the bottom, steeper with every step. Ask for its grade and there's no one number to give. Yet a car parked anywhere on it sits on one definite grade. That grade at a single spot is a derivative, and a line or two of algebra will get it from the curve's equation.
Straight lines first
A straight road is the easy case, because its grade is the same everywhere. Pick any two points, and the slope is rise over run:
A road that climbs 3 ft per 100 ft across has a slope of 0.03, a 3% grade. The Handbook's Straight Line section prints three ways to write a line:
| Form | Equation | Its slope |
|---|---|---|
| Slope-intercept | m, in plain sight | |
| Point-slope | m, with a known point | |
| General |
Adapted from the FE Reference Handbook, Straight Line
Point-slope is the definition of slope with the fraction cleared, so use it whenever you know a slope and a point. The general form hides its slope, and the Handbook doesn't print it, so solve for to dig it out:
So the slope is : for , it's .
Two lines are parallel when they tilt the same way, so they share a slope. Perpendicular takes a moment more. Draw a slope triangle with a run of 4 and a rise of 3, then give it a quarter turn counterclockwise. The run of 4 now points up, so it's a rise of 4. The rise of 3 now points left, so it's a run of −3. The new slope is : the old fraction, flipped, with its sign changed.
The Handbook states this as , and slopes related that way are called negative reciprocals.
The slope of a curve
Now back to the curved road. The ground under your feet looks flat even though the Earth is a ball, because you only see a tiny piece of it. Any smooth curve behaves the same way: zoom in far enough around a point, and it looks like a straight line.
That line is the tangent line, and its slope is the derivative at that point. The Handbook's Differential Calculus section says it flatly: is "the slope of the curve." Slope is rise per unit of run, so the derivative is also the rate of change: how fast is changing with , right at that spot. If were a tank's water level and were time, it would be how fast the level is climbing at this instant.
Rise over run takes two points, though, and a single point has no run. So take a second point a distance away, compute the slope the ordinary way, then slide the second point in. For starting from :
| Run, | Second point | Rise | Rise ÷ run |
|---|---|---|---|
| 1.5 | (2.5, 6.25) | 5.25 | 3.5 |
| 1 | (2, 4) | 3 | 3 |
| 0.5 | (1.5, 2.25) | 1.25 | 2.5 |
| 0.1 | (1.1, 1.21) | 0.21 | 2.1 |
| 0.01 | (1.01, 1.0201) | 0.0201 | 2.01 |
The slopes settle toward exactly 2. You never set to zero, since that would give 0/0. You only ask what value the slopes are closing in on. That's a limit, and it's how the Handbook defines the derivative:
You'll see the derivative written , , or , all meaning the same thing. Read as a tiny rise over a tiny run.
The power rule
Nobody wants to grind through that limit for every curve, and there's a pattern that saves you the trouble. Think of as the area of a square with side . Stretch the side by a sliver , and the area grows by two strips, each long and thick. The tiny corner, , is a sliver of a sliver, and it vanishes as shrinks. So the area grows at a rate of , and at that's the 2 the table was closing in on.
Make it a cube, . Stretching the side grows slabs on three faces, each with area , so the rate is . The pattern keeps going whether or not you can draw it:
Bring the exponent down, then drop it by one. The Handbook's Derivatives list prints this as . That extra matters when is a whole expression, which is the business of Differentiation Rules. When is just , it equals 1. The rule also holds for negative and fractional powers, though this Lesson sticks to polynomials.
The same list gives the rest of the toolkit, and each entry matches a picture:
- A constant has zero slope, . Sliding a curve up or down doesn't change its tilt.
- A constant multiple carries through, . Stretch a curve vertically by , and every rise gets times bigger.
- Sums and differences go term by term. The rises simply add.
Together, those handle any polynomial:
The derivative is itself a function, so you can differentiate again. For , you get and . That , the derivative of the derivative, says how fast the slope is changing, and it comes into its own in Maxima, Minima, and Inflection Points.
For the slope at a point, put its into the derivative.
Working backward from a slope
So far you've been handed an and asked for the slope. The exam also runs it backward: here's a slope, so where on the curve does it occur? Set the derivative equal to that slope and solve for . To find where has a slope of 5, write , which gives .
The slope may be hiding behind the word "parallel." A tangent parallel to a given line has that line's slope, so get the slope from the line first.
Tangent and normal lines
You now have everything needed to write the line that just touches a curve. A line takes a point and a slope, and on a curve those two come from different places, which is where most mistakes live. The point's height comes from the function, . The slope comes from the derivative, . Then point-slope form does the rest:
The normal line goes through the same point but stands square to the tangent, the way a wheel's spoke stands square to the road at the spot where the tire touches it. So its slope is the negative reciprocal, . The Handbook prints no formula for either line, and it doesn't need to: its slope statement and its point-slope form are enough.
Reading the derivative from a graph
You can do a lot with no equation at all. Walk along a graph from left to right and, at each spot, ask one question: am I going uphill or downhill? Uphill means is positive. Downhill means is negative. At the top of a peak or the bottom of a valley you're level for an instant, and is zero.
Notice what's missing from that list: your height. You can walk uphill below sea level. Where crosses the x-axis it's just passing through a height of zero, and its slope there can be anything. In the figure below, crosses the axis at in the middle of its steepest descent, with .
Steepness shows up too: the steeper is, the farther sits from zero. To match a function to the graph of its derivative, find the flat spots first, because they fix where is zero. Then check the signs between them.