## Section: New Results

### Data Structures and Robust Geometric Computation

#### A probabilistic approach to reducing the algebraic complexity of computing Delaunay triangulations

Participant : Jean-Daniel Boissonnat.

In collaboration with Ramsay Dyer (Johann Bernoulli Institute, University of Groningen, Netherlands) and Arijit Ghosh (Max-Planck-Institut für Informatik, Saarbrücken, Germany).

Computing Delaunay triangulations in ${\mathbb{R}}^{d}$ involves evaluating the so-called in_sphere predicate that determines if a point $x$ lies inside, on or outside the sphere circumscribing $d+1$ points ${p}_{0},...,{p}_{d}$. This predicate reduces to evaluating the sign of a multivariate polynomial of degree $d+2$ in the coordinates of the points $x,{p}_{0},...,{p}_{d}$. Despite much progress on exact geometric computing, the fact that the degree of the polynomial increases with $d$ makes the evaluation of the sign of such a polynomial problematic except in very low dimensions. In this paper, we propose a new approach that is based on the witness complex, a weak form of the Delaunay complex introduced by Carlsson and de Silva. The witness complex $\mathrm{Wit}(L,W)$ is defined from two sets $L$ and $W$ in some metric space $X$: a finite set of points $L$ on which the complex is built, and a set $W$ of witnesses that serves as an approximation of $X$. A fundamental result of de Silva states that $\mathrm{Wit}(L,W)=\mathrm{Del}\left(L\right)$ if $W=X={\mathbb{R}}^{d}$. In [25] , [41] , we give conditions on $L$ that ensure that the witness complex and the Delaunay triangulation coincide when $W$ is a finite set, and we introduce a new perturbation scheme to compute a perturbed set ${L}^{\text{'}}$ close to $L$ such that $\mathrm{Del}\left({L}^{\text{'}}\right)=\mathrm{Wit}({L}^{\text{'}},W)$. Our perturbation algorithm is a geometric application of the Moser-Tardos constructive proof of the Lovász local lemma. The only numerical operations we use are (squared) distance comparisons (i.e., predicates of degree 2). The time-complexity of the algorithm is sublinear in $\left|W\right|$. Interestingly, although the algorithm does not compute any measure of simplex quality, a lower bound on the thickness of the output simplices can be guaranteed.

#### Smoothed complexity of convex hulls

Participants : Marc Glisse, Rémy Thomasse.

In collaboration with O. Devillers (VEGAS Project-team) and X. Goaoc (Université Marne-la-Vallée)

We establish an upper bound on the smoothed complexity of convex hulls in ${\mathbb{R}}^{d}$ under uniform Euclidean (${\ell}^{2}$) noise. Specifically, let $\{{p}_{1}^{*},{p}_{2}^{*},...,{p}_{n}^{*}\}$ be an arbitrary set of $n$ points in the unit ball in ${\mathbb{R}}^{d}$ and let ${p}_{i}={p}_{i}^{*}+{x}_{i}$, where ${x}_{1},{x}_{2},...,{x}_{n}$ are chosen independently from the unit ball of radius $\delta $. We show that the expected complexity, measured as the number of faces of all dimensions, of the convex hull of $\{{p}_{1},{p}_{2},...,{p}_{n}\}$ is $O\left({n}^{2-\frac{4}{d+1}}{(1+1/\delta )}^{d-1}\right)$; the magnitude $\delta $ of the noise may vary with $n$. For $d=2$ this bound improves to $O\left({n}^{\frac{2}{3}}(1+{\delta}^{-\frac{2}{3}})\right)$.

We also analyze the expected complexity of the convex hull of ${\ell}^{2}$ and Gaussian perturbations of a nice sample of a sphere, giving a lower-bound for the smoothed complexity. We identify the different regimes in terms of the scale, as a function of $n$, and show that as the magnitude of the noise increases, that complexity varies monotonically for Gaussian noise but non-monotonically for ${\ell}^{2}$ noise [31] , [38] .

#### Realization Spaces of Arrangements of Convex Bodies

Participant : Alfredo Hubard.

In collaboration with M. Dobbins (PosTech, South Korea) and A. Holmsen (KAIST, South Korea)

In [23] , we introduce combinatorial types of arrangements of convex bodies, extending order types of point sets to arrangements of convex bodies, and study their realization spaces. Our main results witness a trade-off between the combinatorial complexity of the bodies and the topological complexity of their realization space. On one hand, we show that every combinatorial type can be realized by an arrangement of convex bodies and (under mild assumptions) its realization space is contractible. On the other hand, we prove a universality theorem that says that the restriction of the realization space to arrangements of convex polygons with a bounded number of vertices can have the homotopy type of any primary semialgebraic set.

#### Limits of order types

Participant : Alfredo Hubard.

In collaboration with X. Goaoc (Institut G. Monge), R. de Joannis de Verclos (CNRS-INPG), J-S. Sereni (LORIA), and J. Volec (ETH)

The notion of limits of dense graphs was invented, among other reasons, to attack problems in extremal graph theory. It is straightforward to define limits of order types in analogy with limits of graphs, and in [24] we examine how to adapt to this setting two approaches developed to study limits of dense graphs. We first consider flag algebras, which were used to open various questions on graphs to mechanical solving via semidefinite programming. We define flag algebras of order types, and use them to obtain, via the semidefinite method, new lower bounds on the density of 5- or 6-tuples in convex position in arbitrary point sets, as well as some inequalities expressing the difficulty of sampling order types uniformly. We next consider graphons, a representation of limits of dense graphs that enable their study by continuous probabilistic or analytic methods. We investigate how planar measures fare as a candidate analogue of graphons for limits of order types. We show that the map sending a measure to its associated limit is continuous and, if restricted to uniform measures on compact convex sets, a homeomorphism. We prove, however, that this map is not surjective. Finally, we examine a limit of order types similar to classical constructions in combinatorial geometry (Erdös-Szekeres, Horton...) and show that it cannot be represented by any somewhere regular measure; we analyze this example via an analogue of Sylvester's problem on the probability that k random points are in convex position.