Complexity of trajectories in rectangular billiards

From MaRDI portal



Abstract: Revised version: some minor errors and typos fixed; exposition watered. Abstract: To a trajectory of a billiard in parallelogram we assign its symbolic trajectory - the sequence of numbers of coordinate plane, to which the faces met by the trajectory are parallel. The complexity of the trajectory is the number of different words of length n occurring in it. We prove that for generic trajectories the complexity is well defined and calculate it, confirming the conjecture of Arnoux, Mauduit, Shiokawa and Tamura [AMST].


The Sturmian sequences are the binary sequences that are a coding of a billiard trajectory in a \((2D)\) square, where the vertical sides are coded by 1 and the horizontal sides by 0. In particular, the (block) complexity of a Sturmian sequence is given by \(\rho (n) = n + 1\), where \(\rho (n)\) is the number of factors (subblocks) of the sequence with length \(n\). What happens if one plays billiard in a cube or hypercube? A conjecture of Rauzy stated that the complexity of the trajectories for the cubic billiards is given by \(\rho (n) = n^2 + n + 1\). This conjecture has been proved by \textit{P. Arnoux}, \textit{C. Mauduit}, \textit{I. Shiokawa} and \textit{J.-I. Tamura} who published two papers [Bull. Soc. Math. Fr. 122, No. 1, 1-12 (1994; Zbl 0791.58034) and Tokyo J. Math. 17, No. 1, 211-218 (1994; Zbl 0814.11014)]. These four authors also conjectured a general formula for the hypercube, the formula presenting a mysterious symmetry in \(n\) (the length of blocks) and \(d-1\) (where \(d\) is the dimension). The author of the paper under review solves the question completely stating in particular that, for reasonable starting angles, one has in dimension \(d\) \[ \rho_d (n) = \sum^{\min (d - 1,n)}_{k = 0} k! {d - 1 \choose k} {n \choose k}. \]




Cited in
(38)








This page was built for publication: Complexity of trajectories in rectangular billiards

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1903332)