The \(\beta \)-Delaunay tessellation. II: the Gaussian limit tessellation (Q2144345)

In a previous paper (presently available from [``The $\beta$-Delaunay tessellation. I: Description of the model and geometry of typical cells'', Preprint, \url{arXiv:2005.13875}]), the same authors have introduced two new classes of stationary random triangulations. The first construction starts with a Poisson point process in \({\mathbb R}^{d-1}\times(0,\infty)\) whose intensity measure has a density (with respect to Lebesgue measure) which at \((v,h)\) is proportional to \(h^\beta\), with \(\beta>-1\). Then a paraboloid hull construction leads to a stationary simplicial tessellation of \({\mathbb R}^{d-1}\), the \(\beta\)-Delaunay tessellation. (Alternatively, dual Laguerre mosaics can be used.) The definition of the \(\beta'\)-Delaunay tessellation is similar, starting with a density on \({\mathbb R}^{d-1}\times (-\infty,0)\) proportional to \((-h)^{-\beta}\), for \(\beta>(d+1)/2\). The present paper is interested in limit tessellations as \(\beta\to\infty\). A notion of weak limit of tessellations is introduced via use of the skeletons of random tessellations as random closed sets, the Fell topology (and even a stronger topology), and the capacity functional. (This is the most intricate part of the investigation.) The weak limit, as \(\beta\to\infty\), of the (suitably rescaled) \(\beta\)-Delaunay and \(\beta'\)-Delaunay tessellation is then the so-called Gaussian--Delaunay tessellation. It can be derived, in a similar way, from a Poisson point process in \({\mathbb R}^{d-1}\times{\mathbb R}\) whose intensity measure is translation invariant in the \({\mathbb R}^{d-1}\) coordinate and has density proportional to \(e^{h/2}\) in the height coordinate \(h\). For the Gaussian-Delaunay tessellation (which is a stationary random triangulation, too), a volume-power weighted typical cell is defined. Its distribution can be given explicitly. From this representation, volume moments and expected angle sums of the weighted typical cells can be determined (for certain powers of the volume, the latter coincide with the angle sums of a regular \((d-1)\)-simplex). A further application is to cell intensities of the Gaussian-Delaunay tessellation. Remarkably, the volume-power weighted typical cell of the Gaussian-Delaunay tessellation has the same distribution as a suitably volume-power weighted Gaussian random simplex (whence the name of the Gaussian-Delaunay tessellation).