Calculus

Calculus is the maths of things that change. It has two halves, and they fit together so neatly that one undoes the other.

The first half is about speed, or rate. Imagine a car. Its distance keeps climbing, but you want to know how fast it is going at one exact instant, not its average over the whole trip. Calculus gives you a way to zoom in on a single moment and read off the rate right there. On a graph, that rate is the steepness of the line at that point. The tool that finds it is called the derivative.

The second half is about totals. Suppose instead you know the speed at every moment and want to work out how far the car travelled. You chop the journey into tiny slivers of time, work out the small distance covered in each, and add them all up. Slice it finely enough and the answer is exact. On a graph, that total is the area sitting under the curve. The tool that finds it is called the integral.

Here is the surprise that makes calculus one idea and not two. Speed and distance are reverses of each other. If you find the rate of a total, you get back the original speed. So the derivative and the integral are opposite moves, like multiplying and dividing. That link is the heart of the whole subject, and once you have it, you can describe almost anything that moves, grows, cools or flows.

Two thousand years of geometry could handle straight lines and static shapes. What it could not do was pin down a rate that changes from instant to instant, or the area of a region with a curved edge. Calculus, built by Newton and Leibniz in the 1600s, cracked both at once using a single tool: the limit, the value something heads toward as you take a step smaller and smaller without ever quite reaching zero.

The derivative. To get the slope of a curve at a point, start with two points and draw the straight line through them, a secant. Its slope is the rise over the run:

\[ \frac{f(x+h) - f(x)}{h}. \]

Now slide the second point in toward the first by shrinking \(h\). The secant pivots, and as \(h\) goes to zero it settles onto the tangent, the line that just grazes the curve. The slope it lands on is the derivative, written \(f'(x)\) or \(\tfrac{dy}{dx}\):

\[ f'(x) = \lim_{h \to 0} \frac{f(x+h) - f(x)}{h}. \]

That single number is the instantaneous rate of change. For position it gives velocity. For velocity it gives acceleration. There are tidy rules for working it out, so \(x^2\) has derivative \(2x\), and \(\sin x\) has derivative \(\cos x\), with no limits to grind through by hand once you know them.

The integral. To get the area under a curve between \(a\) and \(b\), slice the region into thin rectangles, add up their areas, then let the rectangles get thinner and more numerous:

\[ \int_a^b f(x)\,dx = \lim_{n \to \infty} \sum_{i=1}^{n} f(x_i)\,\Delta x. \]

The widget above shows this directly. A handful of fat rectangles is a rough guess; a hundred thin ones trace the curve almost exactly. The limit of that sum is the integral.

The bridge. The fundamental theorem of calculus says the two halves are inverses. If you integrate a function and then differentiate the result, you are back where you started. In practice it means you can compute an area without adding up infinitely many slivers: find a function \(F\) whose derivative is \(f\), and the area is just

\[ \int_a^b f(x)\,dx = F(b) - F(a). \]

That shortcut is why calculus is a calculating tool and not only a beautiful idea. It runs through physics, engineering, economics and statistics, anywhere a quantity changes and you need to track the rate or the accumulation.

What a limit actually means. Newton and Leibniz reasoned with "infinitesimals", quantities smaller than any positive number yet not zero, and it worked spectacularly while resting on shaky ground. Bishop Berkeley mocked them as the ghosts of departed quantities. The repair came in the 1800s from Cauchy and Weierstrass, who replaced the vague infinitesimal with a precise statement. To say \(\lim_{x\to c} f(x) = L\) is to say that for every tolerance \(\varepsilon > 0\) there is a margin \(\delta > 0\) such that \(0 < |x - c| < \delta\) forces \(|f(x) - L| < \varepsilon\). Everything in calculus rests on that one definition.

The derivative, precisely. A function is differentiable at \(x\) if the limit \(f'(x) = \lim_{h\to 0}\frac{f(x+h)-f(x)}{h}\) exists. Differentiability is strictly stronger than continuity: every differentiable function is continuous, but not the reverse. The absolute value \(|x|\) is continuous everywhere yet has no derivative at \(0\), where its graph has a corner and the slope from the left disagrees with the slope from the right. Weierstrass went further and built a function continuous at every point and differentiable at none, a curve that is all corners.

The integral, precisely. The Riemann integral is the common limit of upper and lower sums as the partition is refined, when those two bounds squeeze together to the same value. It handles continuous functions and many others, but it has limits of its own. Lebesgue later rebuilt integration by partitioning the range rather than the domain, which integrates a far wider class of functions and behaves better when you swap a limit and an integral, the foundation modern analysis and probability are built on.

The fundamental theorem, stated. It comes in two parts. First, if \(f\) is continuous and \(F(x) = \int_a^x f(t)\,dt\), then \(F'(x) = f(x)\): differentiating an accumulated area gives back the integrand. Second, if \(F\) is any antiderivative of \(f\), then \(\int_a^b f = F(b) - F(a)\). Together they convert the hard problem of summing infinitely many slivers into the easy one of evaluating an antiderivative at two endpoints.

Where it goes from here. Hold one variable still and differentiate in another and you have partial derivatives and multivariable calculus, the language of fields and surfaces. Let an equation relate a function to its own derivatives and you have a differential equation, the form almost every law of physics takes, from a cooling cup of coffee to the gravitational dance of the three-body problem. One of the great unsolved questions, the Navier–Stokes problem among the Millennium Prize Problems, asks whether the calculus describing fluid flow always has smooth, sensible solutions. The basic machinery is centuries old and still opening new ground.