Post

Contraction Mapping Theorem

Metric Space

Definition of metric space

Definition. A metric space is an ordered pair $(M, d)$ where $M$ is a set and $d$ is a metric on $M$, i.e., a function $d: M\times M \to \mathbb{R}$ satisfying the following axioms for all points $x, y, z \in M:$

  1. The distance from a point to itself is zero: $d(x,x) = 0.$
  2. (Positivity) The distance between two distinct points is always positive: If $x\ne y,$ then $d(x,y)>0.$
  3. (Symmetry) The distance from x to y is always the same as the distance from y to x: $d(x,y) = d(y,x)$
  4. The triangle inequality holds: $d(x,z)\le d(x,y)+d(y,z).$

Note that $0 = d(x,x)\le d(x,y)+d(y,x),$ so the second axiom can be weakened to “If $x\ne y,$ then $d(x,y)\ne 0$” and combined with the first axiom to “$d(x,y) = 0 \Leftrightarrow x = y.$”

Cauchy sequence

Definition. Let $\set{x_t}_{t}^\infty$ be a sequence in a metric space $(M,d),$ then it is a Cauchy sequence, if for every positive real number $\epsilon>0,$ there is a positive integer $N$ such that for all positive integers $m,n > N,$ the distance $d(x_m, x_n) < \epsilon.$

Symbolically, this is:

\[\forall \epsilon > 0 (\exists N\in\mathbb{N} (\forall m,n\in \mathbb{N}(m,n\ge N \Rightarrow d(x_m,x_n)< \epsilon))).\]

Convergent sequence

Definition. Let $\set{x_t}_{t}^\infty$ be a sequence in a metric space $(M,d),$ then it is a convergent sequence, if there is a $x\in M$ such that for every positive real number $\epsilon>0,$ there is a positive integer $N$ such that for all positive integer $n\ge N,$ the distance $d(x_n, x) < \epsilon.$

Definition. A point $x$ of the metric space $(M, d)$ is the limit of the sequence $(x_n)$ if: For each $0<\epsilon\in\mathbb{R},$ there is $N\in\mathbb{N}$ such that, for every $N\le n\in \mathbb{N},$ we have $d(x_n, x)<\epsilon.$

Symbolically, this is:

\[\forall \epsilon > 0 (\exists N\in\mathbb{N} (\forall n\in \mathbb{N}(n\ge N \Rightarrow d(x_n,x)< \epsilon))).\]

Completeness

Definition. A metric space $(M,d)$ is called complete, if every Cauchy sequence in it converges to an element of $M.$

Examples:

  1. $\mathbb{R}$ with usual distance is complete;
  2. $\mathbb{Q}$ (the set of rational numbers) with usual distance is not complete. Because a sequence of rational numbers can converge to an irrational number, e.g. $\pi.$

Contraction Mapping

Definition. A contraction mapping on a metric space $(M,d)$ is a function $f:M\to M$ with the property that there is some real number $0\le k< 1$ such that for all $x$ and $y$ in $M,$

\[d(f(x),f(y))\le k\cdot d(x,y).\]

Fixed Point

Definition. Let $(M,d)$ be a metric space and $f:M\to M$ be a function. If $x\in M$ and $x=f(x),$ then $x$ is called a fixed point of $f.$

Contraction Mapping → Cauchy Sequence

Lemma

Lemma. Let $(M,d)$ be a metric space, $f:M\to M$ be a function, $x_0\in M$, and $x_{t+1} = f(x_t), t=0,1,\ldots,$ then $\set{x_t}_{t}^\infty$ is a Cauchy sequence.

Proof

Since $f$ is a contraction mapping, there must exist a $\beta \in (0,1)$ such that

\[d(x_2, x_1) = d(f(x_1), f(x_0)) \leq \beta d(x_1, x_0)\]

and

\[d(x_3, x_2) \leq \beta d(x_2, x_1) \leq \beta^2 d(x_1, x_0), \cdots.\]

Generally, we have

\[d(x_{t+1}, x_t) \leq \beta d(x_t, x_{t-1}) \leq \cdots \leq \beta^t d(x_1, x_0).\]

Let $N$ be a positive integer. Let positive integers $m, n$ satisfy $m > n \geq N$. Then by the triangle inequality, we have

\[\begin{aligned} d(x_m, x_n) & \leq d(x_m, x_{m-1}) + d(x_{m-1}, x_{m-2}) + \cdots + d(x_{n+2}, x_{n+1}) + d(x_{n+1}, x_n) \\ & \leq \left[\beta^{m-1} + \beta^{m-2} + \cdots + \beta^{n+1} + \beta^n\right] d(x_1, x_0) \\ & = \beta^n \left[\beta^{m-1-n} + \beta^{m-2-n} + \cdots + \beta^1 + 1\right] d(x_1, x_0) \\ & = \beta^n \frac{1-\beta^{m-n}}{1-\beta} d(x_1, x_0) \\ & = \frac{\beta^n - \beta^m}{1-\beta} d(x_1, x_0) \\ & \leq \frac{\beta^N}{1-\beta} d(x_1, x_0) . \end{aligned}\]

Let $\epsilon > 0$. If $d(x_1, x_0) > 0$ (the case of $d(x_1, x_0) = 0$ will be discussed later), then for a positive integer $N$ satisfying

\[N > \frac{\ln \frac{(1-\beta) \epsilon}{d(x_1, x_0)}}{\ln \beta},\]

we have for any $m, n \geq N$, that $d(x_m, x_n) \leq \frac{\beta^N}{1-\beta} d(x_1, x_0) < \epsilon$, hence $\set{x_t}_{t=0}^{\infty}$ is a Cauchy sequence.

If $d(x_1, x_0) = 0$, then $\set{x_t}_{t=0}^{\infty}$ is a constant sequence, and therefore also a Cauchy sequence. $\blacksquare$

Contraction Mapping Theorem

Theorem

Theorem. Let $(M, d)$ be a complete metric space, and $f: M \to M$ be a contraction mapping. Then $f$ has a unique fixed point $x^* \in M$. Furthermore, starting from any point $x_0 \in M$, the sequence $\left(x_0, f(x_0), f(f(x_0)), \cdots\right)$ converges to $x^*$.

Proof

Uniqueness: First, let’s prove that if a point is a fixed point of $f$, then it is the only fixed point of $f$. Suppose $x_1, x_2 \in M$ and $x_1=f(x_1)$, $x_2=f(x_2)$. Since $f$ is a contraction mapping, there exists a $\beta \in (0, 1)$ such that

\[d(x_1, x_2) = d(f(x_1), f(x_2)) \leq \beta d(x_1, x_2).\]

The above inequality holds only when $d(x_1, x_2) = 0$, i.e., $x_1 = x_2$.

Convergence: Since $M$ is complete, there exists a point $x^* \in M$ such that the sequence $\set{x_t}$ converges to $x^*$.

Fixed Point: Let $\epsilon > 0$. Since $\set{x_t}$ is a Cauchy sequence and converges to $x^*$, there exists a positive integer $T$ such that for any $t \geq T$, we have $d(x^*, x_t) < \frac{\epsilon}{3}$ and $d(x_{t+1}, x_t) < \frac{\epsilon}{3}$. For $t \geq T$, by the triangle inequality, we have

\[d(x^*, f(x^*)) \leq d(x^*, x_t) + d(x_t, f(x_t)) + d(f(x_t), f(x^*)).\]

Note that the middle term on the right-hand side is equal to $d(x_t, x_{t+1})$. Hence, the first two terms on the right-hand side are both less than $\frac{\epsilon}{3}$ by our earlier discussion. Now, consider the third term. We have

\[d(f(x_t), f(x^*)) \leq \beta d(x_t, x^*) < \frac{\epsilon}{3}.\]

Thus, $d(x^*, f(x^*)) < \epsilon$. Since $\epsilon$ was arbitrary, we have $d(x^*, f(x^*)) = 0$, i.e., $x^* = f(x^*)$. $\blacksquare$

This post is licensed under CC BY 4.0 by the author.