Graphs, coupling, and spectraHearing the shape of a network — a full workshop on the most wall-like word in mathematics

"Eigenvalue" guards the door to half of physics and most of this website, and it is almost always taught backwards — symbols first, meaning never. This module goes the other way, with your hands on everything: you will pluck a chain of springs and read off the recipe of natural wobbles your pluck was made of, build your own network and play its chord, operate the one simple machine that generates every mode (a single question: how far am I from my neighbors' average?), watch diffusion kill the wiggly patterns first, and finally meet the Fiedler mode — the gentlest way a network can disagree with itself, which finds the weakest seam every time and whose eigenvalue is a live fragility meter. By the end, the sentence in the learning-rule chapter that sent you here reads like plain language.

Three sections, each interactive end to end. Assumes only primer 00.

Modes: a network's natural wobblesPluck a spring chain and hear which overtones you excited — including the shape that rings one pure tone. Step through a real network's modes. Then build your own graph, edge by edge, and play its chord.
The Laplacian: the disagreement machineDrag values onto a network and watch every node measure its own disagreement. Let go, and diffusion drains it — wiggliest modes first, the cluster-vs-cluster pattern always last. The same machine runs the site's coupling dynamics.
The Fiedler mode: the network's early warningThe gentlest disagreement paints the bottleneck on a barbell, a ring, and a grid. Weaken a bridge and watch λ₂ slide continuously toward zero — then cut it, and hear the chord lose its lowest voice.

The curriculum Start: Modes