Brouwer Fixed Point Theorem

The theorem in one sentence

Take any continuous function that maps a closed disc — or more generally any compact, convex set — back into itself. No matter how it stretches, rotates, or squashes the points around, at least one point is guaranteed to land exactly on top of where it started. That is Brouwer's fixed point theorem, proved in full generality by L.E.J. Brouwer in the early 1910s, building on earlier results in two dimensions. The statement sounds almost trivial, but it is one of the most quietly consequential results in all of mathematics, with applications reaching from pure topology into economics, game theory and numerical computing.

live demo · stirring a disc while one point stays fixed● LIVE

Why continuity and convexity both matter

Drop either hypothesis and the theorem breaks immediately. Take an annulus — a disc with a hole punched out of its middle — and rotate it by some fixed nonzero angle. Every point slides along its own circle and none of them ever return to their starting position: there is no fixed point at all, because the hole means the shape is not simply connected the way a solid disc is. Continuity matters just as much: a function that abruptly swaps the left half of the disc with the right half, jumping discontinuously across the middle, can also leave no point fixed. Both requirements — a continuous map, and a domain shaped like a solid ball rather than something with a hole in it — are load-bearing, not decoration.

The coffee-cup intuition, and the easy 1D case

The one-dimensional version is nothing more than the intermediate value theorem: any continuous function f: [0,1] → [0,1] must cross the diagonal line y = x somewhere, because g(x) = f(x) − x is non-negative at x = 0 and non-positive at x = 1, so it must equal zero somewhere in between. The popular two-dimensional illustration is stirring a cup of coffee: however you continuously swirl the liquid's surface and let it settle back within the cup's original circular outline, some point of coffee ends up exactly where it started — the same idea applied to a disc instead of an interval.

A constructive proof sketch: Sperner's lemma

Brouwer's original argument only proves that a fixed point exists — it hands you no way to locate one. Sperner's lemma, proved by Emanuel Sperner in 1928, gives a concrete combinatorial route instead: triangulate the shape finely, label each vertex according to a rule tied to the function f, and Sperner's lemma guarantees that at least one small, fully-labelled triangle must exist no matter how the triangulation is refined, as long as the labelling is consistent along the boundary. As the triangulation gets finer and finer, these fully-labelled triangles shrink down toward an actual fixed point of f — turning a pure existence proof into a genuinely constructive method, the basis of practical algorithms (Scarf's algorithm and its successors) for approximating fixed points to arbitrary precision.

Applications beyond pure math

John Nash's 1950 proof that every finite game has at least one equilibrium in mixed strategies invokes Kakutani's fixed point theorem, a direct generalisation of Brouwer's result to set-valued maps — the existence of a Nash equilibrium is, at bottom, a topological fact, requiring no game-specific reasoning whatsoever. The same family of theorems underlies the Arrow-Debreu proof that a general economic equilibrium of market prices exists, and the infinite-dimensional Schauder fixed point theorem is used routinely to establish that certain differential equations have solutions before anyone tries to compute them numerically. More recently, some implicit-layer and equilibrium-propagation neural network architectures lean on the same kind of fixed-point existence argument to guarantee that a solution exists at all before a numerical solver goes looking for it.

Frequently asked questions

What exactly does Brouwer's fixed point theorem guarantee?

That any continuous function mapping a closed disc, or more generally any compact, convex set, to itself must leave at least one point unmoved. It guarantees the existence of such a point, but says nothing about how many there are or exactly where to find it.

Why doesn't a rotation of a ring, or annulus, have a fixed point even though it seems similar to a disc?

Because an annulus has a hole in the middle, it is not simply connected the way a solid disc is, and a nonzero rotation slides every point along its own circle with no point ever mapping back to itself — showing that the shape of the domain, not just continuity of the function, is essential to the theorem.

Does the theorem tell you how to find the fixed point?

Not directly — Brouwer's original proof is a pure existence argument. Sperner's lemma gives a combinatorial, constructive route instead, approximating the fixed point to any desired precision by refining a triangulation, and this idea underlies practical fixed-point-finding algorithms used in economics and game theory.

Try it live

Everything above runs in your browser — open Brouwer Fixed Point Theorem and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Brouwer Fixed Point Theorem simulation

Brouwer's Fixed Point Theorem: A Point That Cannot Move