Calculus Didn’t Have to Be This Hard

Calculus asks two questions about anything that changes: how fast is it changing right now (a rate, pictured as the slope of a graph), and how much has built up altogether (a total, pictured as the area under a rate graph); steady rates need only rate × time, and calculus is what handles rates that won't hold still.

A rate is one change divided by another, its units are a fraction ("per" means divide) that tell you whether to multiply or divide, a flipped rate (pace, gallons per 100 miles) often tells the truth more plainly, and most real rates change from moment to moment.

On a distance–time graph the steepness is the speed (steep is fast, flat is stopped, straight is steady, bending up is speeding up, bending over is slowing down), on a speed–time graph the height is the speed and the area under it is the distance, and a curve means the speed itself is changing.

Slope is rise ÷ run (written Δy ÷ Δx), it is the same everywhere on a straight line, it can be zero or negative, it is always a rate with units, and grade, pitch and ramp ratios are the same idea in trade language; a curve has a different slope at every point.

An average rate is total change ÷ total change over a stretch, the slope of the secant line joining its ends; it hides what happened in between, it must be worked from totals (never by averaging rates held for different times), a smooth trip hits its average at some instant (the mean value theorem), and shrinking the stretch toward one point makes the averages settle on a single number.

The speed (or any rate) at a single instant is the number that average rates settle on as the stretch of time shrinks toward zero; in a picture it's the slope of the tangent line, the straight line a smooth curve looks like when you zoom in far enough.

A limit is the value a quantity heads for as its input closes in on something, whether or not the input ever arrives; it lets calculus talk about zero time without dividing by zero, turns 0 ÷ 0 into a signal to tidy the algebra first, and also describes long-run values and the places (jumps) where no single value is approached.

The derivative of a function is a second function that gives its slope (rate of change) at every input, written f′(x) or dy/dx, in output units per input unit; it is positive where the function rises, negative where it falls and zero at its flat tops and bottoms, and you can sketch it straight from a graph.

To differentiate a whole-number power of x (the rule holds for other powers too), bring the power down in front and lower it by one; constants have slope zero, and multiples and sums go through term by term, so any polynomial can be differentiated in one line, and the pictures (a square's two strips, a cube's three slabs, a circle's rim) show why.

The second derivative is the rate of change of the rate of change: acceleration for motion, the bend of a graph (positive is a cup, negative a frown), and the difference between "rising" and "rising more slowly"; where the bend flips is the inflection point, where the rate itself is biggest.

When one quantity drives another, rates multiply down the chain (dy/dx = dy/du × du/dx, "the outside's rate times the inside's"), and units cancel link by link; when two changing quantities are multiplied, the product's rate is each one's change times the other, added (the two strips of a growing rectangle), because the little corner vanishes for small changes.

At a smooth peak or valley the slope is zero, so the candidates are where f′(x) = 0; the sign of the slope before and after (or the sign of f″) tells a peak from a valley from a mere pause, and when the question has limits, the ends are candidates too.

Optimization is five steps (draw it, name the letters, write the thing to make best, use the rule to leave one letter, set the slope to zero and check the ends), plus a sixth that the math can't do: ask what the formula left out, as the soda can shows.

Profit is revenue minus cost, revenue is price × quantity sold with quantity falling as price rises (a demand line you must assume and then test), and the most profit comes where marginal revenue equals marginal cost; with a straight demand line, the best price is the revenue-best price plus half the unit cost, and many cost trade-offs of the form a/x + bx are cheapest where the two parts are equal.

When two quantities are tied by a rule that holds at every moment, write the rule, take the rate of both sides with respect to time (the chain rule gives each changing quantity its own rate), and only then put in the moment's numbers; the unknown rate is the known rate divided by the link between them, which can change even when the known rate is steady.

Near a point you know, a smooth curve is almost its tangent line, so new value ≈ known value + slope × step; that gives mental square roots, the rules that a small percent change in a length becomes about twice that in an area and three times in a volume (and that percent changes of a product add), how measuring errors grow, and Newton's method; the guess drifts as the step grows, by about the square of the step.

Draw a rate against time and the area under the graph is the total: miles from speed, gallons from flow, kilowatt-hours from power; straight-edged graphs split into rectangles and triangles, and area below the axis counts as negative, so you can find both the net change and the total movement.

When a rate curves, cut the area under it into thin slices, add each slice's height × width (left edge, right edge, midpoint or trapezoid), and make the slices thinner until the total stops changing; with readings instead of a formula, the trapezoid rule turns any list into a total.

The definite integral is the total that thin-slice sums settle on; read ∫ as "add up", the small numbers as start and stop, and dt as a thin slice's width; it gives the net change (negative pieces subtract), and divided by the length of the stretch it gives the true average.

The area-so-far under a curve grows at the curve's height, so the derivative of a total is its rate (the Fundamental Theorem of Calculus); to find the area from a to b, find any antiderivative F (run the power rule backward and check by differentiating) and compute F(b) − F(a), with + C standing for the starting value when you build a formula.

Any total that builds up from a changing rate is an integral: name the rate, choose what to slice along, make the units match, then slice, multiply and add; the same move gives a day's water or electricity, a growing income, a cone's volume by stacked disks, and braking distance, which grows with the square of speed.

When something grows at a rate proportional to its size, it follows the number e (about 2.71828), the limit of compounding more and more often; e to the x is the curve (up to a constant multiple) whose slope equals its own height; its doubling time is ln 2 divided by the rate, about 70 divided by the percentage.

Anything that loses a fixed fraction of itself in each equal stretch of time decays exponentially: half is gone in each half-life, the rate constant is ln 2 divided by the half-life, and Newton's law of cooling is the same rule applied to the gap between a hot object and its surroundings.

When the rate of change depends on the amount itself, the rule is a differential equation, and you can always step it: Euler's method moves forward with the rate you have now, smaller steps give smaller errors, and setting the rate to zero finds the steady state where flow in equals flow out.

Calculus is two questions about change, how fast right now (the derivative, a slope) and how much altogether (the integral, an area), and the link that each one undoes the other; an ordinary week is full of both, and a calculus course adds notation, techniques, series and differential equations to the same ideas.