# Does a birational involution of C^n always have a fixed point?

The title describes completely the question.

For n=1 it is an easy exercise. For n=2 the statement is still true, and depends on the classification of birational involutions of P^2 (see e.g. http://arxiv.org/abs/math.AG/9907028). For n=3 and above I could not find any useful hint in the literature.

-
In the specific case I'm interested in an additional condition is true: the involution is homotopic (through birational automorphism) to an automorphism with a fixed point. I don't know if it helps. –  Andrea Altomani Mar 17 '10 at 12:49
Maybe I am missing something, but it seems to me that $(x,y)\mapsto(x+1/y,-y)$ is a counterexample. –  user2035 Mar 31 '10 at 8:31
The answer is no. There are plenty of counterexamples, for example the map given by "a-fortiori", which is $(x,y)\mapsto (x+1/y,-y)$, which can be generalised to any dimension.
The good question is probably to look at birational involutions of $\mathbb{P}^n$.