🤖 AI Summary
This work investigates a geometric variant of the Zarankiewicz problem for polygon visibility graphs—specifically, the maximum number of edges in a visibility graph on $n$ vertices that contains no $K_{t,t}$ subgraph. Employing tools from combinatorial geometry, Davenport–Schinzel sequence theory, and divide-and-conquer analysis, we establish the first quasi-linear upper bound $ ilde{O}(n)$ for general simple polygons. For star-shaped and monotone polygons, we obtain tight linear bounds $Theta(n)$. Furthermore, in the more general setting of pseudo-segment visibility graphs induced by $n$ points on a simple closed curve, we prove an $O(n log n)$ upper bound and an $Omega(n alpha(n))$ lower bound, where $alpha(n)$ denotes the inverse Ackermann function. These results determine the asymptotic growth rate of the Zarankiewicz function for visibility graphs: $ ilde{O}(n)$ in the general case, $Theta(n)$ for special polygon classes, and—crucially—reveal its intrinsic connection to $alpha(n)$ for pseudo-segment arrangements.
📝 Abstract
We prove a quasi-linear upper bound on the size of $K_{t,t}$-free polygon visibility graphs. For visibility graphs of star-shaped and monotone polygons we show a linear bound. In the more general setting of $n$ points on a simple closed curve and visibility pseudo-segments, we provide an $O(n log n)$ upper bound and an $Omega(nalpha(n))$ lower bound.