STUDY GUIDE · 2 INTERACTIVE LABS · ABOUT 12 MIN

Nature doesn't tell you where things are. It tells you how they change.

Falling objects, cooling coffee, swinging pendulums, spreading viruses and orbiting planets all obey rules about rates of change. A differential equation is such a rule; solving it predicts the future. Step through one yourself.

FAMOUS DIFFERENTIAL EQUATIONS
Newton's second lawF = m·d²x/dt²Force sets acceleration: the root of all mechanics.
Exponential growthdP/dt = kPBacteria, compound interest, early pandemics.
Newton's coolingdT/dt = −k(T − Tₐ)Why coffee cools fast at first, then slowly.
Simple harmonic motionx″ = −ω²xSprings, pendulums, sound waves.
Heat equation∂u/∂t = α∇²uHow heat spreads through a solid (a PDE).

What is a differential equation?

A differential equation relates a function to its own rates of change (its derivatives). Instead of giving y directly, it gives a rule like "y grows at a rate proportional to itself". Solving it means finding every function that obeys the rule; a starting value then picks out one.

Ordinary differential equations (ODEs) involve one variable, usually time. Partial differential equations (PDEs), like the heat and wave equations, involve several. Most real ones can't be solved exactly, so computers approximate them step by step.

dy/dt = f(t, y)First orderInvolves only the first derivative.
y″ + y = 0Second orderInvolves acceleration: vibrations and orbits.
ay″ + by′ + cy = 0LinearSolutions can be added together (superposition).
dx/dt = x(1 − x)Non-linearCan produce chaos; usually solved numerically.
LAB 1 · SLOPE FIELDS AND EULER'S METHOD

Follow the arrows, one step at a time

Each little dash shows the slope the equation demands at that point. Euler's method starts at y₀, walks a short way in that direction, re-reads the slope, and repeats. Smaller steps track the true solution (dashed) more closely.

12.814Euler y(4)
27.299exact y(4)
53.1%8 steps · error
y = y₀eˣ · Exponential growth: the bigger y gets, the faster it grows. Euler always undershoots here because the slope keeps increasing within each step. Halving h roughly halves the error: Euler is a 'first-order' method.
Euler's approximationexact solutionslope field
LAB 2 · THE SIR EPIDEMIC MODEL

Three coupled equations predict an outbreak

Split a population into Susceptible, Infected and Recovered. Infections happen when S meets I, at rate β; people recover at rate γ. The ratio R₀ = β/γ, the number each case infects in a fully susceptible population, decides everything.

SHARE OF POPULATIONSusceptibleInfectedRecovered
day 0day 50day 100day 150day 200
dS/dt = −β·S·I
dI/dt = β·S·I − γ·I
dR/dt = γ·I
3.00R₀ = β ÷ γ
30.4%peak share infected
day 39peak timing
94%ever infected by day 200
R₀ is above 1, so cases grow until enough people are immune that each case causes fewer than one more. Reducing β (distancing, masks) or adding vaccination flattens and delays the peak. Herd immunity threshold here: 67%.
A teaching model: it assumes everyone mixes equally, immunity is permanent and nothing else changes. Real epidemiology adds age groups, behaviour, vaccines and much more. Starts with 0.1% infected.

Core ideas at a glance

OrderThe highest derivative in the equation.
Initial value problemAn equation plus a starting value; usually has exactly one solution.
Separation of variablesRearrange to ∫dy/g(y) = ∫f(x)dx and integrate both sides.
Integrating factorSolves linear first-order equations y′ + p(x)y = q(x).
Characteristic equationFor ay″ + by′ + cy = 0, try y = eʳˣ and solve ar² + br + c = 0.
EquilibriumA constant solution where the derivative is zero; stable if nearby solutions approach it.
Euler's methodyₙ₊₁ = yₙ + h·f(xₙ, yₙ): simple but needs small steps.
Runge-Kutta (RK4)The workhorse numerical method: much more accurate per step than Euler.

From falling apples to chaos

1687Newton's Principia.Laws of motion written, in effect, as differential equations.
1768Euler's method.Leonhard Euler described stepping along a solution numerically.
1822Fourier's heat equation.Solving it led him to represent functions as sums of sine waves.
1927The SIR model.Kermack and McKendrick modelled epidemics with three coupled equations.
1963Chaos.Edward Lorenz found that a simple weather model was wildly sensitive to its starting values: the butterfly effect.
1969Apollo.Numerical ODE solvers on the Apollo Guidance Computer helped navigate to the Moon.

Further reading

Elementary Differential EquationsBoyce & DiPrima · the standard first course
Nonlinear Dynamics and ChaosSteven Strogatz · where equations get interesting
Differential Equations, Dynamical Systems, and an Introduction to ChaosHirsch, Smale & Devaney · a modern bridge to dynamics
Test yourself: 20 differential equations questionsCalculus guide

Predicting change is reasoning at its best. Measure yours.

Our full IQ test covers numerical, logical, spatial and verbal reasoning: 60 questions, about 30 minutes.

Take the full IQ test