As stated by Banach fixed point theorem, a contraction mapping has only one fixed point. In plain words it means that the contraction mapping T has only one solution that satisfy $Tx = x$.

A stupid question came to my mind : is it just another way to say that eigenvalue is equal to one for some square matrix T ? If so, why would people build the concept of "contraction mapping" rather than just say so.

  • $\begingroup$ Notice that contraction does not need to be a matrix in general. Also, some matrices have a fix point but aren't contraction. $\endgroup$ – user251257 Aug 16 '15 at 21:04
  • $\begingroup$ If the fixed point is unique, the eigenvalue cannot be $1$, or you will get uncountably many fixed points as solutions to $Ax=x$. And I assume you want a nonzero solution for $Ax=x$. $\endgroup$ – Gary. Aug 16 '15 at 21:07
  • $\begingroup$ Any linear mapping has a fixed point. It is $x=0$. $\endgroup$ – A.Γ. Aug 16 '15 at 21:52

Sorry, I am not sure I understood the question, but for a linear map $T$ ,having a unique point is not equivalent to $1$ being an eigenvalue: If $\lambda=1$ , then the associated eigenspace of points {$x: T(x)=x$} is a subspace, and so it will be of dimension one or higher, so it will contain uncountably -many points that will map to themselves, i.e., infinitely -many fixed points, so if you want a unique fixed point then $1$ cannot be an eigenvalue. And, as far as contractions, you need to have the structure of a normed/metric space to talk about contractions. Notice too, as someone wrote in the comments,that contraction maps may not be linear, so they cannot necessarily be represented by a matrix.

Maybe the bottom of this link http://mathworld.wolfram.com/LinearTransformation.html

Will help answer your question.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.