arXiv subjects

Lionel Pournin

Publications and source records attributed to Lionel Pournin.

At least 19 recordsLinked to original sources

arXiv2026

Strong convexity in flip-graphs

The set of triangulations of a surface $Σ$ with a prescribed set $X$ of vertices can be endowed with a graph structure $\mathcal{F}(Σ,X)$ called a flip-graph, whose edges connect two triangulations that differ by a single arc. It is known that when $X$ is the vertex set of a convex Euclidean polygon $\mathrm{P}$, the subgraph $\mathcal{F}_\varepsilon(\mathrm{P},X)$ induced in $\mathcal{F}(\mathrm{P},X)$ by the triangulations that contain a given arc $\varepsilon$ is strongly convex in the sense that all the geodesic paths in $\mathcal{F}(\mathrm{P},X)$ between two such triangulations remain in that subgraph. Here, we provide a related result that involves a triangle instead of an arc: we show that if the three edges of a triangle $τ$ appear in (possibly distinct) triangulations along a geodesic path in $\mathcal{F}(\mathrm{P},X)$, then $τ$ must belong to a triangulation in that path. More generally, we prove that certain $3$\nobreakdash-dimensional simplicial complexes related to the geodesics in $\mathcal{F}(\mathrm{P},X)$ are flag and provide two consequences. The first consequence is that $\mathcal{F}_\varepsilon(\mathrm{P},X)$ is not always strongly convex when $X$ is obtained from the vertex set of $P$ by adding just two points. The second, in the case when $Σ$ is a topological surface, is that the number of arc crossings between two triangulations does not allow to approximate their distance in $\mathcal{F}(Σ,X)$ by a factor of less than $3/2$.

math.GT↗

arXiv2026

Flip-graph non-convexity for once-punctured polygons

The set of the triangulations with vertex set $X$ of a simple polygon $\mathrm{P}$ can be structured into a flip-graph $\mathcal{F}(\mathrm{P},X)$ whose edges connect two triangulations that differ by a single arc. The geometry of flip-graphs has been thoroughly studied and it is known that the subgraph $\mathcal{F}_\varepsilon(\mathrm{P},X)$ induced by the triangulations that contain a given arc $\varepsilon$ is strongly convex in $\mathcal{F}(\mathrm{P},X)$ when $\mathrm{P}$ is convex and $X$ contains no puncture (points in the interior of $\mathrm{P}$) and at most one flat vertex (points in the interior of an edge). When $X$ contains at least two punctures or flat vertices, it is also known that this strong convexity property fails. Here, we close the last open case by showing that, for any convex polygon with sufficiently many vertices, one can always place a single puncture in $X$ in such a way that $\mathcal{F}_\varepsilon(\mathrm{P},X)$ is not strongly convex in $\mathcal{F}(\mathrm{P},X)$. We prove a similar result for simple polygons with a single reflex vertex. The main ingredients in our proofs are a decomposition lemma for a class of $3$-dimensional triangulations and a hyperbolic volume argument regarding their embedding into $\mathbb{H}^3$.

math.MG↗

arXiv2026

Geometry of Sparsity-Inducing Norms

Sparse optimization seeks an optimal solution with few nonzero entries. To achieve this, it is common to add to the criterion a penalty term proportional to the $\ell_1$-norm, which is recognized as the archetype of sparsity-inducing norms. In this approach, the number of nonzero entries is not controlled a priori. By contrast, in this paper, our motivation is to find an optimal solution with at most~$k$ nonzero coordinates (or for short, $k$-sparse vectors), where $k$ is a given sparsity threshold (or ``sparsity budget''). For this purpose, we study the class of generalized $k$-support dual~norms that arise from any given so-called source norm. When added as a penalty term, we provide conditions under which such generalized $k$-support dual~norms promote $k$-sparse solutions. The result follows from an analysis of the exposed faces of closed convex sets generated by $k$-sparse vectors, and of how primal support identification can be deduced from dual information. Finally, we study some of the geometric properties of the unit balls for the $k$-support dual~norms and their dual norms when the source norm belongs to the family of $\ell_p$-norms. In particular, we show a striking structural property: every proper face of the unit balls for the $k$-support dual~norms is a hypersimplex, i.e., the convex hull of $0/1$-valued points with the same $\ell_0$-norm.

math.OC↗

arXiv2026

Flat simplices and kissing polytopes

We consider how flat a lattice simplex contained in the hypercube $[0,k]^d$ can be. This question is related to the notion of kissing polytopes: two lattice polytopes contained in the hypercube $[0,k]^d$ are kissing when they are disjoint but their distance is as small as possible. We show that the smallest possible distance of a lattice point $P$ contained in the cube $[0,k]^3$ to a lattice triangle in the same cube that does not contain $P$ is $$ \frac{1}{\sqrt{3k^4-4k^3+4k^2-2k+1}} $$ when $k$ is at least $2$. We also improve the known lower bounds on the distance of kissing polytopes for $d$ at least $4$ and $k$ at least $2$.

math.MG↗

arXiv2025

Flip-graphs of non-orientable filling surfaces

Consider a surface $Σ$ with punctures that serve as marked points and at least one marked point on each boundary component. We build a filling surface $Σ_n$ by singling out one of the boundary components and denoting by $n$ the number of marked points it contains. We consider the triangulations of $Σ_n$ whose vertices are the marked points and the associated flip-graph $\mathcal{F}(Σ_n)$. Quotienting $\mathcal{F}(Σ_n)$ by the homeomorphisms of $Σ$ that fix the privileged boundary component results in a finite graph $\mathcal{MF}(Σ_n)$. Bounds on the diameter of $\mathcal{MF}(Σ_n)$ are available when $Σ$ is orientable and we provide corresponding bounds when $Σ$ is non-orientable. We show that the diameter of this graph grows at least like $5n/2$ and at most like $4n$ as $n$ goes to infinity. If $Σ$ is an unpunctured Möbius strip, $\mathcal{MF}(Σ_n)$ coincides with $\mathcal{F}(Σ_n)$ and we prove that the diameter of this graph grows exactly like $5n/2$ as $n$ goes to infinity.

math.GT↗

arXiv2025

Deep sections of the hypercube

Consider a non-negative number $t$ and a hyperplane $H$ of $\mathbb{R}^d$ whose distance to the center of the hypercube $[0,1]^d$ is $t$. If $t$ is equal to $0$ and $H$ is orthogonal to a diagonal of $[0,1]^d$, it is known that the $(d-1)$-dimensional volume of $H\cap[0,1]^d$ is a strictly increasing function of $d$ when $d$ is at least $3$. The study of the monotonicity of this volume is extended for $t$ up to above $1/2$ and, when $d$ is large enough, for every non-negative $t$. In particular, a range for $t$ is identified such that this volume is a strictly decreasing function of $d$ over the positive integers. The local extremality of the $(d-1)$-dimensional volume of $H\cap[0,1]^d$ when $H$ is orthogonal to a diagonal of either $[0,1]^d$ or a lower dimensional face is also determined for the same values of $t$. It is shown for instance that when $t$ is above an explicit constant and $d$ is large enough, this volume is always strictly locally maximal when $H$ is orthogonal to a diagonal of $[0,1]^d$. A precise estimate for the convergence rate of the Eulerian numbers to their limit Gaussian behavior is provided along the way.

math.MG↗

arXiv2025

Kissing polytopes in dimension 3

It is shown that the smallest possible distance between two disjoint lattice polytopes contained in the cube $[0,k]^3$ is exactly $$ \frac{1}{\sqrt{2(2k^2-4k+5)(2k^2-2k+1)}} $$ for every integer $k$ at least $4$. The proof relies on modeling this as a minimization problem over a subset of the lattice points in the hypercube $[-k,k]^9$. A precise characterization of this subset allows to reduce the problem to computing the roots of a finite number of degree at most $4$ polynomials, which is done using symbolic computation.

math.MG↗

arXiv2024

Small kissing polytopes

A lattice $(d,k)$-polytope is the convex hull of a set of points in $\mathbb{R}^d$ whose coordinates are integers ranging between $0$ and $k$. We consider the smallest possible distance $\varepsilon(d,k)$ between two disjoint lattice $(d,k)$-polytopes. We propose an algebraic model for this distance and derive from it an explicit formula for $\varepsilon(2,k)$. Our model also allows for the computation of previously intractable values of $\varepsilon(d,k)$. In particular, we compute $\varepsilon(3,k)$ when $4\leq{k}\leq8$, $\varepsilon(4,k)$ when $2\leq{k}\leq3$, and $\varepsilon(6,1)$.

math.CO↗

arXiv2024

Many equiprojective polytopes

A $3$-dimensional polytope $P$ is $k$-equiprojective when the projection of $P$ along any line that is not parallel to a facet of $P$ is a polygon with $k$ vertices. In 1968, Geoffrey Shephard asked for a description of all equiprojective polytopes. It has been shown recently that the number of combinatorial types of $k$-equiprojective polytopes is at least linear as a function of $k$. Here, it is shown that there are at least $k^{3k/2+o(k)}$ such combinatorial types as $k$ goes to infinity. This relies on the Goodman--Pollack lower bound on the number of order types and on new constructions of equiprojective polytopes via Minkowski sums.

math.MG↗

arXiv2024

The expansion of half-integral polytopes

The expansion of a polytope is an important parameter for the analysis of the random walks on its graph. A conjecture of Mihai and Vazirani states that all $0/1$-polytopes have expansion at least 1. We show that the generalization to half-integral polytopes does not hold by constructing $d$-dimensional half-integral polytopes whose expansion decreases exponentially fast with $d$. We also prove that the expansion of half-integral zonotopes is uniformly bounded away from $0$. As an intermediate result, we show that half-integral zonotopes are always graphical.

math.CO↗

arXiv2023

Kissing polytopes

We investigate the following question: how close can two disjoint lattice polytopes contained in a fixed hypercube be? This question stems from various contexts where the minimal distance between such polytopes appears in complexity bounds of optimization algorithms. We provide nearly matching lower and upper bounds on this distance and discuss its exact computation. We also give similar bounds in the case of disjoint rational polytopes whose binary encoding length is prescribed.

math.MG↗

arXiv2023

The rotation distance of brooms

The associahedron $\mathcal{A}(G)$ of a graph $G$ has the property that its vertices can be thought of as the search trees on $G$ and its edges as the rotations between two search trees. If $G$ is a simple path, then $\mathcal{A}(G)$ is the usual associahedron and the search trees on $G$ are binary search trees. Computing distances in the graph of $\mathcal{A}(G)$, or equivalently, the rotation distance between two binary search trees, is a major open problem. Here, we consider the different case when $G$ is a complete split graph. In that case, $\mathcal{A}(G)$ interpolates between the stellohedron and the permutohedron, and all the search trees on $G$ are brooms. We show that the rotation distance between any two such brooms and therefore the distance between any two vertices in the graph of the associahedron of $G$ can be computed in quasi-quadratic time in the number of vertices of $G$.

math.CO↗

arXiv2023

Local extrema for hypercube sections

Consider the hyperplanes at a fixed distance $t$ from the center of the hypercube $[0,1]^d$. Significant attention has been given to determining the hyperplanes $H$ among these such that the $(d-1)$-dimensional volume of $H\cap[0,1]^d$ is maximal or minimal. In the spirit of a question by Vitali Milman, the corresponding local problem is considered here when $H$ is orthogonal to a diagonal or a sub-diagonal of the hypercube. It is proven in particular that this volume is strictly locally maximal at the diagonals in all dimensions greater than $3$ within a range for $t$ that is asymptotic to $\sqrt{d}/\!\log d$. At lower order sub-diagonals, this volume is shown to be strictly locally maximal when $t$ is close to $0$ and not locally extremal when $t$ is large. This relies on a characterisation of local extremality at the diagonals and sub-diagonals that allows to solve the problem over the whole possible range for $t$ in any fixed, reasonably low dimension.

math.MG↗

arXiv2023

Counting geodesics between surface triangulations

Given a surface $Σ$ equipped with a set $P$ of marked points, we consider the triangulations of $Σ$ with vertex set $P$. The flip-graph of $Σ$ whose vertices are these triangulations, and whose edges correspond to flipping arcs appears in the study of moduli spaces and mapping class groups. We consider the number of geodesics in the flip-graph of $Σ$ between two triangulations as a function of their distance. We show that this number grows exponentially provided the surface has enough topology, and that in the remaining cases the growth is polynomial.

math.GT↗

arXiv2023

The complexity of geometric scaling

Geometric scaling, introduced by Schulz and Weismantel in 2002, solves the integer optimization problem $\max \{c\mathord{\cdot}x: x \in P \cap \mathbb Z^n\}$ by means of primal augmentations, where $P \subset \mathbb R^n$ is a polytope. We restrict ourselves to the important case when $P$ is a $0/1$-polytope. Schulz and Weismantel showed that no more than $O(n \log n \|c\|_\infty)$ calls to an augmentation oracle are required. This upper bound can be improved to $O(n \log \|c\|_\infty)$ using the early-stopping policy proposed in 2018 by Le Bodic, Pavelka, Pfetsch, and Pokutta. Considering both the maximum ratio augmentation variant of the method as well as its approximate version, we show that these upper bounds are essentially tight by maximizing over a $n$-dimensional simplex with vectors $c$ such that $\|c\|_\infty$ is either $n$ or $2^n$.

math.OC↗

arXiv2022

Sizing the White Whale

We propose a computational, convex hull free framework that takes advantage of the combinatorial structure of a zonotope, as for example its symmetry group, to orbitwise generate all canonical representatives of its vertices. We illustrate the proposed framework by generating all the 1 955 230 985 997 140 vertices of the $9$-dimensional White Whale. We also compute the number of edges of this zonotope up to dimension $9$ and exhibit a family of vertices whose degree is exponential in the dimension. The White Whale is the Minkowski sum of all the $2^d-1$ non-zero $0/1$-valued $d$-dimensional vectors. The central hyperplane arrangement dual to the White Whale, made up of the hyperplanes normal to these vectors, is called the resonance arrangement and has been studied in various contexts including algebraic geometry, mathematical physics, economics, psychometrics, and representation theory.

math.CO↗

arXiv2021

Diameter estimates for graph associahedra

Graph associahedra are generalized permutohedra arising as special cases of nestohedra and hypergraphic polytopes. The graph associahedron of a graph $G$ encodes the combinatorics of search trees on $G$, defined recursively by a root $r$ together with search trees on each of the connected components of $G-r$. In particular, the skeleton of the graph associahedron is the rotation graph of those search trees. We investigate the diameter of graph associahedra as a function of some graph parameters. We give a tight bound of $Θ(m)$ on the diameter of trivially perfect graph associahedra on $m$ edges. We consider the maximum diameter of associahedra of graphs on $n$ vertices and of given tree-depth, treewidth, or pathwidth, and give lower and upper bounds as a function of these parameters. We also prove that the maximum diameter of associahedra of graphs of pathwidth two is $Θ(n\log n)$. Finally, we give the exact diameter of the associahedra of complete split and of unbalanced complete bipartite graphs.

math.CO↗

arXiv2021

Shallow sections of the hypercube

Consider a $d$-dimensional closed ball $B$ whose center coincides with that of the hypercube $[0,1]^d$. Pick the radius of $B$ in such a way that the vertices of the hypercube are outside of $B$ and the midpoints of its edges in the interior of $B$. It is known that, when $d\geq3$, the $(d-1)$-dimensional volume of $H\cap[0,1]^d$, where $H$ is a hyperplane of $\mathbb{R}^d$ tangent to $B$, is largest possible if and only if $H$ is orthogonal to a diagonal of the hypercube. It is shown here that the same holds when $d\geq5$ but the interior of $B$ is only required to contain the centers of the square faces of the hypercube.

math.MG↗