Skip to content
Get started

Numerical simulation

A calculated temperature has more digits than a thermometer. Which digits should you believe? This book follows that question from one decaying state to a spatial field, then compares two ways of approximating the same equation.

You can arrive directly from your favorite subject. A cooling body, an RC circuit, and a relaxing mechanical system all introduce time scales; heat conduction and elasticity both lead to spatial systems. The examples here are small enough that we can predict their behavior before computing it.

Question Lesson Mathematics used
My solver stopped. Is the answer accurate? Errors and residuals Vectors, matrix multiplication, norms
Why can a decaying system produce oscillations? Time integration Derivatives, exponentials, geometric sequences
How does a differential equation become a matrix? Weak forms and finite elements Integration by parts, piecewise linear functions
How does a mesh conserve a quantity? Finite-volume balance Flux, integration, outward normals
What does a refinement study actually show? Refinement and reproducibility Integrals, logarithms, error ratios

Read in this order for a connected introduction. For physical context, start with heat transfer and return here when you want to understand the numerical choices. The short modeling foundations explain quantities, relations, and boundary conditions when you need them.

The time experiment uses the same decay.eqi as Get started. The spatial experiment uses a sine-shaped Poisson solution, first on an interval and then conceptually on a square. Each lesson follows the same habit: state a prediction, express the mathematics, compute, and try to disprove the interpretation with a controlled change.

Follow Get started for the Python installation. Keep its eqiora-source checkout: the Poisson comparison runs its maintained Cargo example from that directory. The source revision attached to these pages supplies the models and programs together.

The goal is a useful explanation of an answer: which equation it approximates, where it is sampled, what changes under refinement, and which physical assumptions still need examination.

Start: Errors and residuals