MÉTHODOS · the way · numerical methods: how a machine computes real answers to continuous problems, one controlled approximation at a time · kept by HERON (Hero of Alexandria, the ancient numerical algorithmist)

THE FIXED-POINT ITERATION ◧ 2D · ◍ 3D · ◆ 4D · ◐ shadow · 👶 TAP

Press cos on a calculator, over and over, from any start — the number stops moving at 0.739. That’s a FIXED POINT: a value that maps to itself, x = g(x). Iterating g settles onto it whenever g doesn’t stretch distances (|slope| < 1) — a ‘contraction.’ It’s the simplest solver there is, and the idea behind everything from PageRank to equilibrium. Slide to iterate and watch it spiral onto the fixed point.

◆ LIT▲ AMBER
◧ THE MEASURE · 2D
◍ FEED THE OUTPUT BACK IN · 3D · a value that maps to itself
◆ THE FOURTH · 4D · a tesseract turns
◐ THE SHADOW · one dimension down
👶 THE TODDLER CORNER — one fat tap
x
|x−g(x)|
|g′|
<1 ✓
fixed point
0.739

◆ LIT — exact / checkable

Fixed-point iteration solves x=g(x) by simply iterating xₙ₊₁=g(xₙ). It converges to a fixed point x* whenever g is a CONTRACTION near it — |g′(x*)|<1 — by the Banach fixed-point theorem, at a LINEAR rate set by |g′| (smaller = faster); if |g′|>1 it diverges. Iterating cos settles on the Dottie number 0.739085…. Most solvers are fixed-point schemes in disguise: [[the-newton-method|Newton]], [[the-gauss-seidel|Gauss–Seidel]], value iteration, and PageRank all iterate a map to its fixed point. A fail-loud self-check throws unless iterating cos reaches 0.739085 to 1e−8. ◆ real numerical methods, node-verified.

▲ AMBER — the figure

Convergence needs |g′|<1 near the root and only LINEAR speed; the same equation can be rearranged into a diverging g. The contraction condition and the cobweb dynamics are exact.

MÉTHODOS: the continuous is never solved, only APPROACHED — the art is bounding the error.  — HERON
David Lee Wise / ROOT0, with AVAN · MÉTHODOS — kept by HERON, the reckoner