Why computers use numerical approximations to solve the equations governing our physical world
In short: Many differential equations are impossible to solve exactly, yet they underpin physics, chemistry, and economics. Numerical methods provide a practical workaround by generating approximate solutions. From the simple Euler method to complex Runge-Kutta algorithms, these computational tools allow us to model systems that would otherwise remain mathematically intractable.
Ordinary differential equations (ODEs) describe how systems change over time, but finding exact analytical solutions is often impossible. Numerical methods, or numerical integration, offer a way to approximate these solutions by breaking down continuous problems into discrete steps. By choosing a step size and applying recursive formulas, computers can trace the path of a solution curve, providing the precision necessary for scientific and economic modeling.
The most fundamental approach is the Euler method, described by Leonhard Euler in 1768. It approximates a curve by moving along a tangent line at each step. While intuitive, it is just one of many techniques. Modern solvers generally fall into two primary categories: linear multistep methods, which leverage data from previous iterations, and Runge-Kutta methods, which utilize intermediate stages to calculate the next point. These methods are further distinguished as explicit, which rely solely on known values, or implicit, which require solving an equation for the next step.
Selecting the right method involves a trade-off between computational cost and stability. Implicit methods, such as Adams-Moulton or backward differentiation formulas, are often necessary for 'stiff' equations—systems where explicit methods might fail or require prohibitively small steps. While implicit schemes demand more processing time due to the need for iterative solvers like Newton-Raphson, their superior stability makes them essential for complex simulations. Newer developments, such as exponential integrators dating back to the 1960s, continue to expand this toolkit by exactly integrating linear terms within a differential equation.
Beyond standard initial value problems, these methods are versatile enough to handle higher-order systems by converting them into larger sets of first-order equations. They even play a role in solving partial differential equations by discretizing them into systems of ODEs. Whether through general linear methods or specialized integrators, these algorithms bridge the gap between abstract mathematical theory and the practical requirements of modern science.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Numerical methods for ordinary differential equations
- Also: ocw.mit.edu — Numerical Methods | Differential Equations | Mathematics | MIT OpenCourseWare
To cite this card, link to https://intellectplus.co.uk/app/science-numerical-methods-for-ordinary-differential-equations.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Calculus and analysis · More in Science · Latest · A–Z · All topics