Generally speaking, I am looking for a generalization of the Schauder fixed point theorem, which applies to the situation described briefly below.

All references I read (e.g. E. Zeider 'Nonlinear Functional Analysis and its Applications I Fixed Point theorems') provide only standard version of the theorem.

My situation is as follows , $X$ is a Banach space (a subspace of continuous functions on the interval $[0,1]$), $f\colon X\to X$ is a continuous map. There exist a set $A$ (I describe it below) such that $$ f(A)\subset A, $$ where $A$ is NON convex. STILL I know a lot about $A$, it is a set of continuous functions on the interval $[0,1]$, close, bounded. And e.g. $A$ contains a convex set $C$, and there is a homotopy between any function in $A$ and a function in $C$. My intuition is that there exists a fixed point of $f$ within $A$.

I cannot apply the more general Lefshetz theorem, as I do not have direct access to $f$.

  • 1
    $\begingroup$ Do you mean perhaps that there is a homotopy of $A$ that simultaneously retracts all of its points it to a subset of $C$? $\endgroup$ Sep 1, 2016 at 7:23
  • $\begingroup$ @IgorKhavkine Yes, I think I can show this for my sets, but I think it is even more than contractible, I know contractible is not enough $\endgroup$
    – jaco
    Sep 1, 2016 at 14:19

1 Answer 1


Your guess is false. One way to see this is to note that each infinite-dimensional Banach space $B$ contains a closed bounded subset $A$ which is homeomorphic to $[0,\infty)$. Now, take a map $f: A\to A$ which corresponds to the shift map $t\mapsto t+1$ in $[0,\infty)$ and use the Tietze extension theorem to extend $f$ to a continuous map $B\to A$. Clearly, $f$ has no fixed points. Take $C$ to be a singleton in $A$. Clearly, $C$ a strong deformation retract of $A$, providing homotopies that you want to have.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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