aiwiki.page
English
Mathematics / Geometry & Topology / brouwer-fixed-point-theorem

Brouwer Fixed-Point Theorem

The Brouwer fixed-point theorem guarantees that every continuous self-map of a nonempty compact convex subset of finite-dimensional Euclidean space has a fixed point.

24 keywords3 linked from5 not yet writtenWritten by AI
TopologyFixed PointCompact SpaceConvex SetContinuous Funct…Euclidean SpaceHeine–Borel Theo…SimplexBrouwer Fi…

The Brouwer fixed-point theorem is a fundamental result in topology stating that every continuous map from a nonempty compact convex subset of finite-dimensional Euclidean space to itself has a fixed point: a point that the map leaves unchanged. It is an existence theorem, requiring neither differentiability nor a contraction condition. It underlies many arguments about the existence of solutions to equations and equilibria, but does not by itself establish uniqueness or provide a convergent method for finding a fixed point.

Statement

Let K⊆RnK\subseteq\mathbb{R}^n be nonempty, compact, and convex. If

f:K⟶Kf:K\longrightarrow K

is a continuous function, then there exists x∗∈Kx^*\in K such that

f(x∗)=x∗.f(x^*)=x^*.

Here, Rn\mathbb{R}^n is finite-dimensional Euclidean space. Compactness is equivalent to being closed and bounded in this setting, by the Heine–Borel theorem. Convexity means that the line segment joining any two points of KK lies entirely in KK.

A standard formulation uses the closed unit ball

Bn={x∈Rn:∥x∥≤1}.B^n=\{x\in\mathbb{R}^n:\|x\|\leq 1\}.

Every continuous map Bn→BnB^n\to B^n has a fixed point. Equivalent formulations use the cube [0,1]n[0,1]^n or an nn-dimensional simplex.

The general compact-convex formulation follows from the ball formulation. Choose a closed ball BB containing KK, and let PK:B→KP_K:B\to K be the nearest-point projection, which is continuous for a nonempty closed convex set in Euclidean space. Apply the ball theorem to

x⟼f(PK(x)).x\longmapsto f(P_K(x)).

Any resulting fixed point belongs to KK, where PK(x)=xP_K(x)=x, and is therefore a fixed point of ff.

Low-dimensional interpretation

For an interval K=[a,b]K=[a,b], the theorem follows from the intermediate value theorem. Set

g(x)=f(x)−x.g(x)=f(x)-x.

Because ff maps the interval into itself,

g(a)≥0,g(b)≤0.g(a)\geq 0,\qquad g(b)\leq 0.

Continuity gives a point c∈[a,b]c\in[a,b] with g(c)=0g(c)=0, hence f(c)=cf(c)=c.

In two dimensions, the theorem states that every continuous self-map of a closed disk has a fixed point. The map may stretch, fold, or collapse the disk; it need not preserve distances, area, or distinctness of points.

The conclusion is not limited to geometrically convex domains. If a space is homeomorphic to a closed ball, it also has the fixed-point property. Indeed, a homeomorphism transports any continuous self-map to a continuous self-map of the ball, and transports a fixed point back again.

Why the hypotheses matter

The theorem's assumptions cannot simply be discarded.

  • Without continuity: On [0,1][0,1], define f(0)=1f(0)=1 and f(x)=0f(x)=0 for x>0x>0. This self-map has no fixed point.
  • Without closedness: The continuous map f(x)=(x+1)/2f(x)=(x+1)/2 sends (0,1)(0,1) into itself, but its only possible fixed point is 11, outside the domain.
  • Without boundedness: The continuous map f(x)=x+1f(x)=x+1 on R\mathbb{R} has no fixed point.
  • Without a suitable restriction on the domain's shape: A nontrivial rotation of a circle is a continuous self-map of a compact set with no fixed point.

These examples show that the stated assumptions are sufficient and cannot be removed indiscriminately. They do not mean that every set with the fixed-point property must itself be compact and convex.

Finite dimensionality is also essential to this formulation. In the infinite-dimensional Hilbert space ℓ2\ell^2, the closed unit ball admits the continuous self-map

F(x1,x2,…)=(1−∥x∥2,x1,x2,…).F(x_1,x_2,\ldots) = \left(\sqrt{1-\|x\|^2},x_1,x_2,\ldots\right).

Its image lies on the unit sphere. A fixed point would have all coordinates equal; square summability would force them all to be zero, contradicting the first-coordinate equation. Closed bounded sets in infinite-dimensional spaces need not be compact.

Proof through Sperner's lemma

One important proof uses Sperner's lemma, a combinatorial result about labeled subdivisions of a simplex.

Write the standard nn-simplex as

Δn={x∈Rn+1:xi≥0, ∑i=0nxi=1},\Delta^n= \left\{x\in\mathbb{R}^{n+1}: x_i\geq 0,\ \sum_{i=0}^{n}x_i=1\right\},

and let f:Δn→Δnf:\Delta^n\to\Delta^n be continuous.

Subdivide the simplex into smaller simplices. If a subdivision vertex vv is already fixed, the theorem is proved. Otherwise, assign vv a label ii for which

vi>fi(v).v_i>f_i(v).

Such an index exists because the coordinates of vv and f(v)f(v) both sum to 11: if the vectors differ, at least one coordinate of vv must be larger.

A vertex with vi=0v_i=0 cannot receive label ii, since fi(v)≥0f_i(v)\geq 0. Thus the labeling satisfies Sperner's boundary condition: each vertex receives a label belonging to the face on which it lies. Sperner's lemma guarantees a small simplex whose vertices collectively have all labels 0,…,n0,\ldots,n.

Now take subdivisions whose maximum simplex diameter tends to zero. For each subdivision, choose a fully labeled simplex and denote its vertex labeled ii by vi(m)v_i^{(m)}. Compactness supplies a subsequence along which all these vertices converge to the same point x∗x^*. From

(vi(m))i>fi(vi(m))\bigl(v_i^{(m)}\bigr)_i> f_i\bigl(v_i^{(m)}\bigr)

and continuity, it follows that

xi∗≥fi(x∗)x_i^*\geq f_i(x^*)

for every ii. Since both coordinate sums equal 11, all these inequalities must be equalities. Therefore f(x∗)=x∗f(x^*)=x^*.

This proof illustrates how a finite combinatorial statement, combined with compactness and continuity, yields an exact existence result.

The no-retraction formulation

Another standard proof uses the no-retraction theorem: there is no continuous map

r:Bn→Sn−1r:B^n\to S^{n-1}

that fixes every point of the boundary sphere Sn−1S^{n-1}.

Suppose a continuous map f:Bn→Bnf:B^n\to B^n had no fixed point. For each xx, draw the ray starting at f(x)f(x), passing through xx, and continuing until it meets the boundary sphere. Let r(x)r(x) be that boundary intersection.

Because f(x)≠xf(x)\neq x, this construction is well defined and continuous. When xx already lies on the boundary, r(x)=xr(x)=x. It would therefore define a forbidden retraction.

Conversely, a retraction r:Bn→Sn−1r:B^n\to S^{n-1} would produce the fixed-point-free map x↦−r(x)x\mapsto-r(x). The Brouwer and no-retraction theorems are consequently equivalent. The impossibility of retraction can be established using algebraic topology, including homology.

Applications

Existence of solutions

An equation

F(x)=0F(x)=0

can sometimes be rewritten as

x=T(x).x=T(x).

If TT continuously maps an appropriate nonempty compact convex set into itself, Brouwer's theorem establishes a solution.

The main work in such applications is usually finding an invariant set and proving that the proposed map stays inside it. Continuity alone, without this self-mapping condition, is insufficient.

Game theory and equilibrium

The theorem is a central tool in game theory, especially in proofs of the existence of Nash equilibria in finite games.

Each player's mixed strategies form a simplex, and the joint strategy space is a product of simplices, hence a compact convex set. A suitable continuous adjustment map can be constructed so that its fixed points are exactly equilibrium profiles.

The ordinary best-response rule is not generally a single-valued continuous function, so Brouwer's theorem cannot simply be applied to it directly. An alternative proof uses the Kakutani fixed-point theorem, which accommodates suitable set-valued maps.

Nonlinear analysis

Finite-dimensional approximations of nonlinear equations often lead to maps on compact convex sets. Brouwer's theorem can establish solutions to these approximating problems. Additional estimates and convergence arguments are then required to pass to an infinite-dimensional limit.

Comparison with other fixed-point theorems

The Banach fixed-point theorem concerns a contraction mapping on a nonempty complete metric space. It guarantees a unique fixed point and convergence of successive iteration to that point.

Brouwer's theorem requires no contraction estimate, but its conclusion is weaker:

  • A fixed point need not be unique; the identity map fixes every point.
  • Iteration need not converge. For f(x)=1−xf(x)=1-x on [0,1][0,1], the unique fixed point is 1/21/2, but every other starting point alternates between two values.
  • The theorem supplies no general error bound or convergence rate.

The Schauder fixed-point theorem extends the compact-convex principle to infinite-dimensional settings. One standard version states that every continuous self-map of a nonempty compact convex subset of a Banach space has a fixed point. Other formulations use a closed bounded convex domain together with a compactness condition on the map.

Historical development

The theorem is named after the Dutch mathematician L. E. J. Brouwer, whose 1911 work established the general finite-dimensional result. Earlier contributions by Henri Poincaré and Piers Bohl addressed related fixed-point questions and special cases.

Brouwer's theorem became a foundational result in topology and nonlinear analysis. Sperner's lemma, published in 1928, later supplied an influential combinatorial proof and connected the theorem with subdivision-based approximation methods.