Periods of rational maps modulo primes

Let $K$ be a number field, let $phi in K(t)$ be a rational map of degree at least 2, and let $alpha, beta in K$. We show that if $alpha$ is not in the forward orbit of $beta$, then there is a positive proportion of primes ${mathfrak p}$ of $K$ such that $alpha mod {mathfrak p}$ is not in the forward orbit of $beta mod {mathfrak p}$. Moreover, we show that a similar result holds for several maps and several points. We also present heuristic and numerical evidence that a higher dimensional analog of this result is unlikely to be true if we replace $alpha$ by a hypersurface, such as the ramification locus of a morphism $phi : {mathbb P}^{n} to {mathbb P}^{n}$.

The Dynamical Mordell-Lang Conjecture

We prove a special case of a dynamical analogue of the classical Mordell-Lang conjecture. In particular, let $phi$ be a rational function with no superattracting periodic points other than exceptional points. If the coefficients of $phi$ are algebraic, we show that the orbit of a point outside the union of proper preperiodic subvarieties of $(bP^1)^g$ has only finite intersection with any curve contained in $(bP^1)^g$. We also show that our result holds for indecomposable polynomials $phi$ with coefficients in $bC$. Our proof uses results from $p$-adic dynamics together with an integrality argument. The extension to polynomials defined over $bC$ uses the method of specializations coupled with some new results of Medvedev and Scanlon for describing the periodic plane curves under the action of $(phi,phi)$ on $bA^2$.

A gap principle for dynamics

Let $f_1,...,f_gin {mathbb C}(z)$ be rational functions, let $Phi=(f_1,...,f_g)$ denote their coordinatewise action on $({mathbb P}^1)^g$, let $Vsubset ({mathbb P}^1)^g$ be a proper subvariety, and let $P=(x_1,...,x_g)in ({mathbb P}^1)^g({mathbb C})$ be a nonpreperiodic point for $Phi$. We show that if $V$ does not contain any periodic subvarieties of positive dimension, then the set of $n$ such that $Phi^n(P) in V({mathbb C})$ must be very sparse. In particular, for any $k$ and any sufficiently large $N$, the number of $n leq N$ such that $Phi^n(P) in V({mathbb C})$ is less than $log^k N$, where $log^k$ denotes the $k$-th iterate of the $log$ function. This can be interpreted as an analog of the gap principle of Davenport-Roth and Mumford.