Symplectic billiards in a Penthouse]{Every symplectic billiard orbit in a Penthouse is periodic
The symplectic billiard map of a convex polygon takes an oriented chord $xy$ to the chord $yz$ for which $xz$ is parallel to the side that contains $y$. A Penthouse of modulus $m=\lfloor(a+1)/a\rfloor$ is a unit square with a triangle of height $a>0$ on its top side. Albers, Banhatti, Sadlo, Schwartz and Tabachnikov proved that every orbit is periodic when $m=1$. They conjectured that for every $m$ and every $a$ with $1/a\notin\Z$, every orbit is periodic with the period $16m-4$, $16m+4$ or $16m+12$. They also conjectured that the tiles of period $16m+12$ survive the change from $m$ to $m+1$, where a tile is the set of the chords whose orbits visit the sides in one periodic sequence. The first conjecture would add a two-parameter family to the polygons in which every orbit is periodic, a class that Albers and Tabachnikov asked to describe. We prove the first conjecture for every orbit that avoids the vertices, every $m$ and every position of the apex. In suitable coordinates, the first-return map to the chords from the base to the right side of the roof is a quarter turn of the torus followed by a translation, so its fourth power is the identity. The chords of each period form a set whose phase area has a closed form that does not depend on the position of the apex. The second conjecture holds if a tile of period $P$ means the set of the chords of period $P$, whereas for each $m\ge2$ it fails at $m-1$ or $m$ positions of the apex if a tile has the meaning above. For $m=1$ and the apex above the midpoint of the top side, three tile orbits exist, whereas the theorem for $m=1$ lists six.
Authors
- Sungwoo Kang
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24422652
- Primary Topic
- Mathematical Dynamics and Fractals
- Type
- preprint