Calculus & Change

Measuring change - the maths of motion!

Junior level — plain language, no maths

Throw a ball straight up. It rises fast, then slower, hangs for a heartbeat at the very top, then drops faster and faster. Here's a sneaky question that stumped people for centuries: how fast is it moving at one exact instant - not over a second, not even a millisecond, but at a single frozen moment? Freeze time completely and the ball hasn't moved at all, so the usual "distance divided by time" just hands you a baffling zero over zero.

Newton and Leibniz cracked it in the 1600s, working separately, and their idea is beautifully simple: zoom in on any smooth curve closely enough and it starts to look like a straight line. The slope of that line is the speed at that instant - and, more generally, the derivative, the rate at which something is changing right now. Steepness becomes speed.

Run the same machine in reverse - add up endlessly many endlessly thin slices - and you get the integral, which hands you areas, volumes, distances and grand totals. Derivatives and integrals are two sides of one coin, and together they may be the most powerful tools mathematics has ever produced, quietly running physics, engineering, medicine, economics and very nearly every science there is.

Derivatives, integrals, and the Fundamental Theorem of Calculus

Student level — the core equations

The derivative is built from a limit that stares the "zero over zero" problem down and beats it: \(f'(x) = \lim_{h\to 0} \dfrac{f(x+h) - f(x)}{h}\). Geometrically it's the slope of the tangent line; physically, if \(f(t)\) is position then \(f'(t)\) is velocity and \(f''(t)\) acceleration. Two rules carry most of the load: the power rule \(\dfrac{d}{dx}x^n = n x^{n-1}\), and the chain rule \(\dfrac{d}{dx}f(g(x)) = f'(g(x))\,g'(x)\) for peeling apart nested functions.

The definite integral \(\int_a^b f(x)\,dx\) is the signed area under a curve, defined as the limit of ever-finer rectangles - a Riemann sum \(\lim_{n\to\infty}\sum_i f(x_i)\,\Delta x\). Then comes the punchline that binds the whole subject together, the Fundamental Theorem of Calculus: if \(F' = f\), then \(\int_a^b f(x)\,dx = F(b) - F(a)\). Differentiation and integration look like utterly different jobs - finding slopes versus finding areas - and yet they are exact inverses. That single fact is the hinge the entire subject swings on.

Aim calculus at time and you get differential equations, the language nature uses to state its laws. The simplest, \(\dfrac{dy}{dt} = ky\), captures anything whose growth keeps pace with its own size - populations, compound interest, radioactive decay - and solves neatly to \(y(t) = y_0 e^{kt}\). Add a little complexity and the same machinery delivers Newton's second law, the wave equation and the heat equation, the differential equations that run essentially all of physics and engineering.

Real analysis, measure theory, and the Lebesgue integral

Scholar level — full mathematical depth

01Where Riemann runs out

The integral you meet first - Riemann's - chops the horizontal axis into thin strips and totals their areas. It works beautifully for the smooth functions of physics, but it chokes on anything too jagged. The standard horror is the Dirichlet function, equal to 1 on the rationals and 0 on the irrationals: between any two points it jumps infinitely often, no strip ever settles, and the Riemann integral flatly refuses to exist. To do analysis honestly - to trade limits and integrals without crossing your fingers - you need a sturdier definition.

02Lebesgue's idea: slice the range, not the domain

Lebesgue's move (1902) was to turn the picture on its side. Instead of partitioning the input, partition the output: ask how much of the domain lands in each thin horizontal band of values, and weight by that amount - \(\int f\,d\mu = \sup \sum_i y_i\,\mu\!\left(f^{-1}[y_i, y_{i+1}]\right)\). The Dirichlet function instantly becomes trivial: the rationals, where it equals 1, form a set of measure zero, so they contribute nothing and the integral is simply 0. The space \(L^1(\mu)\) of integrable functions turns into a complete Banach space - a proper home for analysis.

03Measure, and the things you cannot measure

Beneath it all sits measure theory, the craft of consistently assigning a "size" to sets - length on the line, area in the plane, probability in a sample space - that survives countable unions. It is general enough to size oddities like the Cantor set, yet it has a hard edge: granting the Axiom of Choice, you can build sets so pathological that no consistent measure exists for them at all. The Banach–Tarski paradox lurks here, slicing a sphere into a few unmeasurable scraps and reassembling them into two spheres identical to the first. Measure theory is precisely the language that explains why that ought to feel impossible.

04The convergence theorems - the real prize

Analysts didn't switch for elegance; they switched for power. Lebesgue's integral arrives with theorems for swapping a limit and an integral - the step physicists take constantly, usually without a second thought. The Dominated Convergence Theorem is the workhorse: if \(f_n \to f\) pointwise and all are capped by a single integrable \(g\), with \(|f_n| \le g\), then \(\lim \int f_n\,d\mu = \int f\,d\mu\). Riemann offers no such clean guarantee, and nearly every interchange of sum, limit and integral in modern physics is quietly leaning on this result.

05L² and Fourier: functions as geometry

The square-integrable functions \(L^2\) form a Hilbert space - an infinite-dimensional geometry equipped with lengths and angles, in which the complex exponentials \(\{e^{inx}/\sqrt{2\pi}\}\) serve as perpendicular axes. Expanding a function in them is just dropping perpendiculars, and Parseval's identity \(\|f\|^2 = \sum_n |\hat{c}_n|^2\) is Pythagoras in disguise: a signal's energy equals the energy of its spectrum. The continuous version, the Fourier transform \(\hat{f}(\xi) = \int f(x)\,e^{-2\pi i \xi x}\,dx\), carries the same geometry onto the whole real line.

06Where it all lands

None of this is ornamental - it is the floor modern science stands on. Quantum mechanics is the study of state vectors in a Hilbert space, with \(|\hat{\psi}(p)|^2\) the momentum probability density and operators standing in for observables. Probability theory is measure theory under another name, and the Central Limit Theorem falls out cleanly through characteristic functions. Every MP3, JPEG and MRI scan is Fourier analysis cashed out in silicon. The shaky "infinitesimals" of Newton and Leibniz took two centuries to make rigorous - and that rigour turned out to be the bedrock of twentieth-century physics.