Proofs of the conjectures by Mecke for mixed line-generated tessellations (Q2439853)

The STIT tessellations comprise a class of random tessellations that have been studied in detail in recent years. \textit{J. Mecke} [``Inhomogeneous random planar tessellations generated by lines'', J. Contemp. Math. Anal., Armen. Acad. Sci. 45, No. 6, 357--367 (2010; \url{doi:10.3103/S1068362310060075})], introduced a large class of mixed line-generated tessellations in the plane. Starting from a fixed window, a line is chosen at random (according to some distribution) which divides the window into two cells. Then, at each time step, another random line \(g\) is generated and one of the cells \(C\) is chosen at random. If \(g\) hits \(C\), it generates two new cells (otherwise an empty ``quasi-cell'' arises). Although introduced in discrete time, Mecke proposed a transfer of this construction to a continuous time process and formulated three conjectures. The first conjecture says, roughly, that the tessellation generated in a window \(W\) and then intersected with a subwindow \(\hat W\subset W\) yields the corresponding tessellation constructed in \(\hat W\), where the line measure and the time parameter are adjusted appropriately. The second conjecture concerns the iteration (or nesting) of mixed line-generated tessellations. If the cells of one tessellation in this class are divided by i.i.d.\ copies of another tessellation, the outcome is again a tessellation of mixed line-generated type. Finally, the third conjecture says that in the case, where the driving line measure in \(W\) comes from a translation invariant measure in the plane, the mixed line-generated tessellation and the STIT model are the same (in distribution). In this paper, the author gives a proof for all three conjectures. The key tool is a formula for the waiting time until a convex set within a cell is hit by a line.