Banach fixed-point theorem

In mathematics, the Banach fixed-point theorem (also known as the contraction mapping theorem or contractive mapping theorem or Banach–Caccioppoli theorem) is an important tool in the theory of metric spaces; it guarantees the existence and uniqueness of fixed points of certain self-maps of metric spaces, and provides a constructive method to find those fixed points. It can be understood as an abstract formulation of Picard's method of successive approximations. The theorem is named after Stefan Banach (1892–1945) who first stated it in 1922.

Statement
Definition. Let $$(X, d)$$ be a metric space. Then a map $$T : X \to X$$ is called a contraction mapping on X if there exists $$q \in [0, 1)$$ such that
 * $$d(T(x),T(y)) \le q d(x,y)$$

for all $$x, y \in X.$$

Banach fixed-point theorem. Let $$(X, d)$$ be a non-empty complete metric space with a contraction mapping $$T : X \to X.$$ Then T admits a unique fixed-point $$x^*$$ in X (i.e. $$T(x^*) = x^*$$). Furthermore, $$x^*$$ can be found as follows: start with an arbitrary element $$x_0 \in X$$ and define a sequence $$(x_n)_{n\in\mathbb N}$$ by $$x_n = T(x_{n-1})$$ for $$n \geq 1.$$ Then $$\lim_{n \to \infty} x_n = x^*$$.

Remark 1. The following inequalities are equivalent and describe the speed of convergence:



\begin{align} d(x^*, x_n) & \leq \frac{q^n}{1-q} d(x_1,x_0), \\[5pt] d(x^*, x_{n+1}) & \leq \frac{q}{1-q} d(x_{n+1},x_n), \\[5pt] d(x^*, x_{n+1}) & \leq q d(x^*,x_n). \end{align} $$

Any such value of q is called a Lipschitz constant for $$T$$, and the smallest one is sometimes called "the best Lipschitz constant" of $$T$$.

Remark 2. $$d(T(x),T(y))<d(x,y)$$ for all $$x \neq y$$ is in general not enough to ensure the existence of a fixed point, as is shown by the map
 * $$T : [1,\infty) \to [1,\infty), \,\, T(x)=x+\tfrac{1}{x}\,,$$

which lacks a fixed point. However, if $$X$$ is compact, then this weaker assumption does imply the existence and uniqueness of a fixed point, that can be easily found as a minimizer of $$d(x,T(x))$$, indeed, a minimizer exists by compactness, and has to be a fixed point of $$T.$$ It then easily follows that the fixed point is the limit of any sequence of iterations of $$T.$$

Remark 3. When using the theorem in practice, the most difficult part is typically to define $$X$$ properly so that $$T(X) \subseteq X.$$

Proof
Let $$x_0 \in X$$ be arbitrary and define a sequence $$(x_n)_{n\in\mathbb N}$$ by setting $$x_n = T(x_{n-1})$$. We first note that for all $$n \in \N,$$ we have the inequality


 * $$d(x_{n+1}, x_n) \le q^n d(x_1, x_0).$$

This follows by induction on n, using the fact that T is a contraction mapping. Then we can show that $$(x_n)_{n\in\mathbb N}$$ is a Cauchy sequence. In particular, let $$m, n \in \N$$ such that $$m > n $$:


 * $$\begin{align}

d(x_m, x_n) & \leq d(x_m, x_{m-1}) + d(x_{m-1}, x_{m-2}) + \cdots + d(x_{n+1}, x_n) \\[5pt] & \leq q^{m-1}d(x_1, x_0) + q^{m-2}d(x_1, x_0) + \cdots + q^nd(x_1, x_0) \\[5pt] & = q^n d(x_1, x_0) \sum_{k=0}^{m-n-1} q^k \\[5pt] & \leq q^n d(x_1, x_0) \sum_{k=0}^\infty q^k \\[5pt] & = q^n d(x_1, x_0) \left ( \frac{1}{1-q} \right ). \end{align}$$

Let ε > 0 be arbitrary. Since $$q \in [0,1)$$, we can find a large $$N \in \N$$ so that


 * $$q^N < \frac{\varepsilon(1-q)}{d(x_1, x_0)}.$$

Therefore, by choosing $$m$$ and $$n$$ greater than $$N$$ we may write:


 * $$d(x_m, x_n) \leq q^n d(x_1, x_0) \left ( \frac{1}{1-q} \right ) < \left (\frac{\varepsilon(1-q)}{d(x_1, x_0)} \right ) d(x_1, x_0) \left ( \frac{1}{1-q} \right ) = \varepsilon.$$

This proves that the sequence $$(x_n)_{n\in\mathbb N}$$ is Cauchy. By completeness of (X,d), the sequence has a limit $$x^* \in X.$$ Furthermore, $$x^*$$ must be a fixed point of T:


 * $$x^*=\lim_{n\to\infty} x_n = \lim_{n\to\infty} T(x_{n-1}) = T\left(\lim_{n\to\infty} x_{n-1} \right) = T(x^*). $$

As a contraction mapping, T is continuous, so bringing the limit inside T was justified. Lastly, T cannot have more than one fixed point in (X,d), since any pair of distinct fixed points p1 and p2 would contradict the contraction of T:


 * $$ d(T(p_1),T(p_2)) = d(p_1,p_2) > q d(p_1, p_2).$$

Applications

 * A standard application is the proof of the Picard–Lindelöf theorem about the existence and uniqueness of solutions to certain ordinary differential equations. The sought solution of the differential equation is expressed as a fixed point of a suitable integral operator on the space of continuous functions under the uniform norm. The Banach fixed-point theorem is then used to show that this integral operator has a unique fixed point.
 * One consequence of the Banach fixed-point theorem is that small Lipschitz perturbations of the identity are bi-lipschitz homeomorphisms. Let Ω be an open set of a Banach space E; let I : Ω → E denote the identity (inclusion) map and let g : Ω → E be a Lipschitz map of constant k < 1. Then
 * 1) Ω′ := (I + g)(Ω) is an open subset of E: precisely, for any x in Ω such that B(x, r) ⊂ Ω one has B((I + g)(x), r(1 − k)) ⊂ Ω′;
 * 2) I + g : Ω → Ω′ is a bi-Lipschitz homeomorphism;
 * precisely, (I + g)−1 is still of the form I + h : Ω → Ω′ with h a Lipschitz map of constant k/(1 − k). A direct consequence of this result yields the proof of the inverse function theorem.


 * It can be used to give sufficient conditions under which Newton's method of successive approximations is guaranteed to work, and similarly for Chebyshev's third-order method.
 * It can be used to prove existence and uniqueness of solutions to integral equations.
 * It can be used to give a proof to the Nash embedding theorem.
 * It can be used to prove existence and uniqueness of solutions to value iteration, policy iteration, and policy evaluation of reinforcement learning.
 * It can be used to prove existence and uniqueness of an equilibrium in Cournot competition, and other dynamic economic models.

Converses
Several converses of the Banach contraction principle exist. The following is due to Czesław Bessaga, from 1959:

Let f : X → X be a map of an abstract set such that each iterate fn has a unique fixed point. Let $$q \in (0, 1),$$ then there exists a complete metric on X such that f is contractive, and q is the contraction constant.

Indeed, very weak assumptions suffice to obtain such a kind of converse. For example if $$f : X \to X$$ is a map on a T1 topological space with a unique fixed point a, such that for each $$x \in X$$ we have fn(x) → a, then there already exists a metric on X with respect to which f satisfies the conditions of the Banach contraction principle with contraction constant 1/2. In this case the metric is in fact an ultrametric.

Generalizations
There are a number of generalizations (some of which are immediate corollaries).

Let T : X → X be a map on a complete non-empty metric space. Then, for example, some generalizations of the Banach fixed-point theorem are:
 * Assume that some iterate Tn of T is a contraction. Then T has a unique fixed point.
 * Assume that for each n, there exist cn such that d(Tn(x), Tn(y)) ≤ cnd(x, y) for all x and y, and that
 * $$\sum\nolimits_n c_n <\infty.$$
 * Then T has a unique fixed point.

In applications, the existence and uniqueness of a fixed point often can be shown directly with the standard Banach fixed point theorem, by a suitable choice of the metric that makes the map T a contraction. Indeed, the above result by Bessaga strongly suggests to look for such a metric. See also the article on fixed point theorems in infinite-dimensional spaces for generalizations.

A different class of generalizations arise from suitable generalizations of the notion of metric space, e.g. by weakening the defining axioms for the notion of metric. Some of these have applications, e.g., in the theory of programming semantics in theoretical computer science.

Example
Banach theorem allows for example fast and accurate calculation of the $\pi$ number using the trigonometric functions which numerically are the power Taylor series.

Because $$\sin(\pi)=0$$ and the π is the fixed point of for example the function $$f(x)=\sin(x)+x$$

i.e.


 * $$f(\pi)=\pi$$

and also the function $$f$$ is around π the contraction mapping from the obvious reasons because its derivative in π vanishes therefore π can be obtained from the infinite superposition for example for the argument value 3:


 * $$\pi=f(f(f(\cdots f(3)\cdots))))$$

Already the triple superposition of this function at $$3$$ gives π with accuracy to 33 digits:


 * $$f(f(f(3)))=3.141592653589793238462643383279502\ldots\,.$$