PlanetMath (more info)
 Math for the people, by the people. Sponsor PlanetMath
Encyclopedia | Requests | Forums | Docs | Wiki | Random | RSS  
Login
create new user
name:
pass:
forget your password?
Main Menu
Revision difference : proof of Banach fixed point theorem
Version current Version 1
Let $(X,d)$ be a non-empty, complete metric space, and let $T$ be a Let $(X,d)$ be a non-empty, complete metric space, and let $T$ be a
contraction mapping on $(X,d)$ with constant $q$. Pick an arbitrary contraction mapping on $(X,d)$ with constant $q$. Pick an arbitrary
$x_0 \in X$, and define the sequence $(x_n)_{n=0}^{\infty}$ by $x_0 \in X$, and define the sequence $(x_n)_{n=0}^{\infty}$ by
$x_n:=T^nx_0$. Let $a:=d(Tx_0,x_0)$. We first show by induction that $x_n:=T^nx_0$. Let $a:=d(Tx_0,x_0)$. We first show by induction that
for any $n\ge 0$, for any $n\ge 0$,
$$ $$
d(T^nx_0,x_0)\le\frac{1-q^n}{1-q} a. d(T^nx_0,x_0)\le\frac{1-q^n}{1-q} a.
$$ $$
For $n=0$, this is obvious. For any $n\ge 1$, suppose that For $n=0$, this is obvious. For any $n\ge 1$, suppose that
$d(T^{n-1}x_0,x_0)\le\frac{1-q^{n-1}}{1-q}a$. Then $d(T^{n-1}x_0,x_0)\le\frac{1-q^{n-1}}{1-q}a$. Then
\begin{eqnarray*} \begin{eqnarray*}
d(T^nx_0,x_0)&\le&d(T^nx_0,T^{n-1}x_0)+d(x_0,T^{n-1}x_0)\\ d(T^nx_0,x_0)&\le&d(T^nx_0,T^{n-1}x_0)+d(x_0,T^{n-1}x_0)\\
&\le&q^{n-1}d(Tx_0,x_0)+\frac{1-q^{n-1}}{1-q}a\\ &\le&q^{n-1}d(Tx_0,x_0)+\frac{1-q^{n-1}}{1-q}a\\
&=&\frac{q^{n-1}-q^n}{1-q}a+\frac{1-q^{n-1}}{1-q}a\\ &=&\frac{q^{n-1}-q^n}{1-q}a+\frac{1-q^{n-1}}{1-q}a\\
&=&\frac{1-q^n}{1-q}a &=&\frac{1-q^n}{1-q}a
\end{eqnarray*} \end{eqnarray*}
by the triangle inequality and repeated application of the property by the triangle inequality and repeated application of the property
$d(Tx,Ty)\le qd(x,y)$ of $T$. By induction, the inequality holds for $d(Tx,Ty)\le qd(x,y)$ of $T$. By induction, the inequality holds for
all $n \ge 0$.\\ all $n \ge 0$.\\
\\ \\
Given any $\epsilon>0$, it is possible to choose a natural number $N$ Given any $\epsilon>0$, it is possible to choose a natural number $N$
such that $\frac{q^n}{1-q}a<\epsilon$ for all $n\ge N$, because such that $\frac{q^n}{1-q}a<\epsilon$ for all $n\ge N$, because
$\frac{q^n}{1-q}a\to 0$ as $n\to\infty$. Now, for any $m,n\ge N$ (we $\frac{q^n}{1-q}a\to 0$ as $n\to\infty$. Now, for any $m,n\ge N$ (we
may assume that $m\ge n$), may assume that $m\ge n$),
\begin{eqnarray*} \begin{eqnarray*}
d(x_m,x_n)&=&d(T^mx_0,T^nx_0)\\ d(x_m,x_n)&=&d(T^mx_0,T^nx_0)\\
&\le&q^nd(T^{m-n}x_0,x_0)\\ &\le&q^nd(T^{m-n}x_0,x_0)\\
&\le&q^n\frac{1-q^{m-n}}{1-q}a\\ &\le&q^n\frac{1-q^{m-n}}{1-q}a\\
&<&\frac{q^n}{1-q}a<\epsilon, &<&\frac{q^n}{1-q}a<\epsilon,
\end{eqnarray*} \end{eqnarray*}
so the sequence $(x_n)$ is a Cauchy sequence. Because $(X,d)$ is so the sequence $(x_n)$ is a Cauchy sequence. Because $(X,d)$ is
complete, this implies that the sequence has a limit in $(X,d)$; complete, this implies that the sequence has a limit in $(X,d)$;
define $x^*$ to be this limit. We now prove that $x^*$ is a fixed define $x^*$ to be this limit. We now prove that $x^*$ is a fixed
point of $T$. Suppose it is not, then $\delta:=d(Tx^*,x^*)>0$. point of $T$. Suppose it is not, then $\delta:=d(Tx^*,x^*)>0$.
However, because $(x_n)$ converges to $x^*$, there is a natural However, because $(x_n)$ converges to $x^*$, there is a natural
number $N$ such that $d(x_n,x^*)<\delta/2$ for all $n\ge N$. Then number $N$ such that $d(x_n,x^*)<\delta/2$ for all $n\ge N$. Then
\begin{eqnarray*} \begin{eqnarray*}
d(Tx^*,x^*)&\le&d(Tx^*,x_{N+1})+d(x^*,x_{N+1})\\ d(Tx^*,x^*)&\le&d(Tx^*,x_{N+1})+d(x^*,x_{N+1})\\
&\le&qd(x^*,x_N)+d(x^*,x_{N+1})\\ &\le&qd(x^*,x_N)+d(x^*,x_{N+1})\\
&<&\delta/2+\delta/2=\delta, &<&\delta/2+\delta/2=\delta,
\end{eqnarray*} \end{eqnarray*}
contradiction. So $x^*$ is a fixed point of $T$. It is also unique. contradiction. So $x^*$ is a fixed point of $T$. It is also unique.
Suppose there is another fixed point $x'$ of $T$; because $x'\neq Suppose there is another fixed point $x'$ of $T$; because $x'\neq
x^*$, $d(x',x^*)>0$. But then x^*$, $d(x',x^*)>0$. But then
$$ $$
d(x',x^*)=d(Tx',Tx^*)\le qd(x',x^*)<d(x',x^*), d(x',x^*)=d(Tx',Tx^*)\le qd(x',x^*)<d(x',x^*),
$$ $$
contradiction. Therefore, $x^*$ is the unique fixed point of $T$. contradiction. Therefore, $x^*$ is the unique fixed point of $T$.