Borsuk-Ulam and Fixed Point Theorems
At any given moment on the surface of the Earth there are always two antipodal points with exactly the same temperature and barometric pressure. We can go even further: on each longitude (the North and South lines running from pole to pole) there will also be two antipodal points sharing exactly the same temperature. How is this possible? The answer is given by the Borsuk-Ulam Theorem: a powerful tool in Topology with a great deal of applications to every branch of science.
A continuous function from a -dimensional sphere into the Euclidean space of dimension maps some pair of antipodal points to the same point
The implications of this Theorem are huge. Some of the most useful are cited below:
- No subset of can be homeomorphic to a -dimensional sphere
- If we cover the sphere with a family of open sets, then at least one of these sets must contain a pair of antipodal points
But the most useful application of Borsuk-Ulam is without a doubt the Brouwer Fixed Point Theorem.
Every continuous function from a convex compact subset of a Euclidean space to itself has a fixed point.
The proof of Brouwer Fixed Point from Borsuk-Ulam is immediate, and I urge the readers to find it by themselves as a nice exercise. There are many different proofs of Brouwer Fixed Point without Borsuk-Ulam, the simplest of them all, done by C.A.Rogers as a simplification of a previous proof by J.W.Milnor: it uses the most basic of mathematics, accesible to students with knowledge of elementary integral Calculus.
A version of this proof starts by showing that there are no differentiable maps with continuous derivative, from a closed ball to its border, that fixes all points in that sphere:
Let denote the unit ball in the Euclidean -space, whose border is the -dimensional sphere There is no map (once-differentiable with continuous derivative) such that for all
The use of the previous result together with the Stone-Weierstrass Theorem guarantees the existence of fixed points for any continuous function
Indeed, the Stone-Weierstrass Theorem gives the existence, for each of a polynomial satisfying Consider the sequence of functions which converges uniformly to
We claim that each function has a fixed point If this were not true, we would be able to construct functions satisfying for all Such a construction could be as follows:
If has no fixed point, then it is possible to draw, for each , a ray through emanating from . This ray intersects at a single point, which we denote The function thus defined is and satisfies for all contradicting the previous result. |
This implies, as we announced before, the existence of a sequence of points satisfying The compactness of guarantees a convergent subsequence . Let be its limit. It must be then
and we have just found a fixed point for
The proof of the Theorem for continuous functions for a given convex compact set is done through an appropriate homeomorphism
Miscellaneous
Note that the Brouwer Fixed Point Theorem is not constructive: although the proof indicates how to come up with a fixed point, there is no explicit construction that finds such point. The key here is the Stone-Weierstrass Theorem, that claims the existence of a sequence of polynomials, but does not compute them.
We can find nonetheless a constructive version of a Fixed Point Theorem in the field of Analysis: the Banach Fixed Point Theorem. This result gives a general criterion that guaranties that, if satisfied, the procedure of iterating a function will yield a fixed point.
Let be a non-empty complete metric space. Let be a contraction mapping on , i.e.: there is a nonnegative real number such that
for all
for all Then the map admits one and only one fixed point in Furthermore, this fixed point can be found as follows: start with an arbitrary element in and define an iterative sequence by for This sequence converges, and its limit is the fixed point of
Other versions also allow us to count how many fixed points will be for a given function. That is the case of both the Lefschetz and the Nielsen Fixed Point Theorems.
References
Topology (2nd Edition) (See all Topology Books)
What about the converse? Does the Brouwer’s fixed point theorem imply the Boursuk-ulam Theorem?