BZPEERSource author record
Explore this author’s publications, collaborators and research topics. This catalog record has not been claimed by its author.
Catalog footprint
Published work
Let Θ(G) denote the Shannon capacity of a graph G. We give an elementary proof of the equivalence, for any graphs G and H, of the inequalities Θ(G⊔H)>Θ(G)+Θ(H) and Θ(G⊠H)>Θ(G)Θ(H). This was shown independently by Wigderson and Zuiddam [2022] using Kadison-Dubois duality and the Axiom of choice.
We study relations between geometric embeddings of graphs and the spectrum of associated matrices, focusing on outerplanar embeddings of graphs. For a simple connected graph G=(V,E), we define a "good" G-matrix as a V×V matrix with negative entries corresponding to adjacent nodes, zero entries corresponding to distinct nonadjacent nodes, and exactly one negative eigenvalue. We give an algorithmic proof of the fact that it G is a 2-connected graph, then either the nullspace representation defined by any "good" G-matrix with corank 2 is an outerplanar embedding of G, or else there exists a "good" G-matrix with corank 3.
A {\em cyclic graph} is a graph with at each vertex a cyclic order of the edges incident with it specified. We characterize which real-valued functions on the collection of cubic cyclic graphs are partition functions of a real vertex model (P. de la Harpe, V.F.R. Jones, Graph invariants related to statistical mechanical models: examples and problems, Journal of Combinatorial Theory, Series B 57 (1993) 207--227). They are characterized by `weak reflection positivity', which amounts to the positive semidefiniteness of matrices based on the `k-join' of cubic cyclic graphs (for all $k\in\oZ_+$). Basic tools are the representation theory of the symmetric group and geometric invariant theory, in particular the Hanlon-Wales theorem on the decomposition of Brauer algebras and the Procesi-Schwarz theorem on inequalities defining orbit spaces.
We characterize which graph invariants are partition functions of an edge-coloring model over the complex numbers, in terms of the rank growth of associated `connection matrices'.
Let G be a group of orthogonal transformations of a real Hilbert space H. Let R and W be bounded G-stable subsets of H. Let ∥.∥R be the seminorm on H defined by ∥x∥R:=supr∈R∣⟨r,x⟩∣ for x∈H. We show that if W is weakly compact and the orbit space Rk/G is compact for each $k\in\oN$, then the orbit space W/G is compact when W is equiped with the norm topology induced by ∥.∥R. As a consequence we derive the existence of limits of edge-colouring models which answers a question posed by Lovász. It forms the edge-colouring counterpart of the graph limits of Lovász and Szegedy, which can be seen as limits of vertex-colouring models. In the terminology of de la Harpe and Jones, vertex- and edge-colouring models are called `spin models' and `vertex models' respectively.
The {\it partially disjoint paths problem} is: {\it given:} a directed graph, vertices r1,s1,…,rk,sk, and a set F of pairs {i,j} from {1,…,k}, {\it find:} for each i=1,…,k a directed ri−si path Pi such that if {i,j}∈F then Pi and Pj are disjoint. We show that for fixed k, this problem is solvable in polynomial time if the directed graph is planar. More generally, the problem is solvable in polynomial time for directed graphs embedded on a fixed compact surface. Moreover, one may specify for each edge a subset of {1,…,k} prescribing which of the ri−si paths are allowed to traverse this edge.
We characterize the virtual link invariants that can be described as partition function of a real-valued R-matrix, by being weakly reflection positive. Weak reflection positivity is defined in terms of joining virtual link diagrams, which is a specialization of joining virtual link diagram tangles. Basic techniques are the first fundamental theorem of invariant theory, the Hanlon-Wales theorem on the decomposition of Brauer algebras, and the Procesi-Schwarz theorem on inequalities for closed orbits.
Let T be a set, of {\em types}, and let $ι,o:T\to\oZ_+$. A {\em T-diagram} is a locally ordered directed graph G equipped with a function τ:V(G)→T such that each vertex v of G has indegree ι(τ(v)) and outdegree o(τ(v)). (A directed graph is {\em locally ordered} if at each vertex v, linear orders of the edges entering v and of the edges leaving v are specified.) Let V be a finite-dimensional $\oF$-linear space, where $\oF$ is an algebraically closed field of characteristic 0. A function R on T assigning to each t∈T a tensor R(t)∈V∗⊗ι(t)⊗V⊗o(t) is called a {\em tensor representation} of T. The {\em trace} (or {\em partition function}) of R is the $\oF$-valued function pR on the collection of T-diagrams obtained by `decorating' each vertex v of a T-diagram G with the tensor R(τ(v)), and contracting tensors along each edge of G, while respecting the order of the edges entering v and leaving v. In this way we obtain a {\em tensor network}. We characterize which functions on T-diagrams are traces, and show that each trace comes from a unique `strongly nondegenerate' tensor representation. The theorem applies to virtual knot diagrams, chord diagrams, and group representations.
We show that if G=(V,E) is a 4-connected flat graph, then any real symmetric V×V matrix M with exactly one negative eigenvalue and satisfying, for any two distinct vertices i and j, Mij<0 if i and j are adjacent, and Mij=0 if i and j are nonadjacent, has the Strong Arnold Property: there is no nonzero real symmetric V×V matrix X with MX=0 and Xij=0 whenever i and j are equal or adjacent. (A graph G is {\em flat} if it can be embedded injectively in 3-dimensional Euclidean space such that the image of any circuit is the boundary of some disk disjoint from the image of the remainder of the graph.) This applies to the Colin de Verdière graph parameter, and extends similar results for 2-connected outerplanar graphs and 3-connected planar graphs.
We give necessary and sufficient conditions for a weight system on multiloop chord diagrams to be obtainable from a metrized Lie algebra representation, in terms of a bound on the ranks of associated connection matrices. Here a multiloop chord diagram is a graph with directed and undirected edges so that at each vertex precisely one directed edge is entering and precisely one directed edge is leaving, and each vertex is incident with precisely one undirected edge. Weight systems on multiloop chord diagrams yield the Vassiliev invariants for knots and links. The k-th connection matrix of a function f on the collection of multiloop chord diagrams is the matrix with rows and columns indexed by k-labeled chord tangles, and with entries equal to the f-value on the join of the tangles.
A {\em 3-graph} is a connected cubic graph such that each vertex is is equipped with a cyclic order of the edges incident with it. A {\em weight system} is a function f on the collection of 3-graphs which is {\em antisymmetric}: f(H)=−f(G) if H arises from G by reversing the orientation at one of its vertices, and satisfies the IHX-equation. Key instances of weight systems are the functions φg obtained from a metric Lie algebra g by taking the structure tensor c of g with respect to some orthonormal basis, decorating each vertex of the 3-graph by c, and contracting along the edges. We give equations on values of any complex-valued weight system that characterize it as complex Lie algebra weight system. It also follows that if f=φg for some complex metric Lie algebra g, then f=φg′ for some unique complex reductive metric Lie algebra g′. Basic tool throughout is geometric invariant theory.
We give a short proof of Szemerédi's regularity lemma, based on elementary Euclidean geometry. The general line of the proof is that of the standard proof (in fact, of Szemerédi's original proof), but most technicalities are swallowed by applying Pythagoras' theorem.
We characterize which graph invariants are partition functions of a spin model over the complex numbers, in terms of the rank growth of associated `connection matrices'.
We show that for each fixed k, the problem of finding k pairwise vertex-disjoint directed paths between given source-sink pairs in a planar directed graph is solvable in polynomial time. In fact, it suffices to fix the number of faces needed to cover all sources and sinks. Moreover, the method can be extended to any fixed compact orientable surface (instead of the plane) and to rooted trees (instead of paths). Our approach is algebraic and is based on cohomology over graph (nonabelian) groups. More precisely, let D=(V,A) be a directed graph and let (G,⋅) be a group. Call two function ϕ,ψ:A→G {\em cohomologous} if there exists a function p:V→G such that p(u)⋅ϕ(a)⋅p(w)−1=ψ(a) for each arc a=(u,w). Now given a function ϕ:A→G we want to find a function ψ cohomologous to ϕ such that each ψ(a) belongs to a prescribed subset H(a) of G. We give a polynomial-time algorithm for this problem in case G is a graph group and each H(a) is closed (i.e., if word xyz belongs to H(a) then also word y belongs to H(a)). The method also implies that such a ψ exists, if and only if for each s∈V and each pair P,Q of (undirected) s−s paths there exists an x∈G such that x⋅ϕ(P)⋅x−1∈H(P) and x⋅ϕ(Q)⋅x−1∈H(P). (Here ϕ(P) is the product of the ϕ(a) over the arcs in P. Similarly, H(P) is the (group subset) product of the H(a).)
Let k≤n. Each polynomial $p\in\oR[x_1,...,x_n]$ can be uniquely written as p=∑μμpμ, where μ ranges over the set M of all monomials in $\oR[x_1,...,x_k]$ and where $p_μ\in\oR[x_{k+1},...,x_n]$. If p is d-homogeneous and ε>0, we say that p is {\em ε-concentrated on the first k variables} if $$\sum_{μ\in M\atop°(μ)<d}\max_{x\in\oR^{n-k}\atop\|x\|=1}p_μ(x)^2\leq\varepsilon\|p\|^2,$$ where ∥p∥ is the Bombieri norm of p. We show that for each $d\in\oN$ and ε>0 there exists kd,ε such that for each n and each d-homogeneous $p\in\oR[x_1,...,x_n]$ there exists k≤kd,ε such that p is ε-concentrated on the first k variables {\em after some orthogonal transformation of $\oR^n$}. (So kd,ε is independent of the number n of variables.) We derive this as a consequence of a more general theorem on low rank approximation of polynomials.
Virtual links were introduced by Kauffman in 1999. We characterize the virtual link invariants that are partition functions of vertex models (as considered by de la Harpe and Jones), both in the real and in the complex case. We show that for any fixed number of states, these invariants form an affine variety. Basic techniques are the first and second fundamental theorem of invariant theory for the orthogonal group (in the sense of Weyl) and some related methods from algebraic geometry.
Let X be an inner product space, let G be a group of orthogonal transformations of X, and let R be a bounded G-stable subset of X. We define very weak and very strong regularity for such pairs (R,G) (in the sense of Szemerédi's regularity lemma), and prove that these two properties are equivalent. Moreover, these properties are equivalent to the compactness of the space (B(H),dR)/G. Here H is the completion of X (a Hilbert space), B(H) is the unit ball in H, dR is the metric on H given by dR(x,y):=supr∈R∣<r,x−y>∣, and (B(H),dR)/G is the orbit space of (B(H),dR) (the quotient topological space with the G-orbits as quotient classes). As applications we give Szemerédi's regularity lemma, a related regularity lemma for partitions into intervals, and a low rank approximation theorem for homogeneous polynomials.
We characterize which graph parameters are partition functions of a vertex model over an algebraically closed field of characteristic 0 (in the sense of de la Harpe and Jones). We moreover characterize when the vertex model can be taken so that its moment matrix has finite rank.
Let A(n,d) be the maximum number of 0,1 words of length n, any two having Hamming distance at least d. We prove A(20,8)=256, which implies that the quadruply shortened Golay code is optimal. Moreover, we show A(18,6)≤673, A(19,6)≤1237, A(20,6)≤2279, A(23,6)≤13674, A(19,8)≤135, A(25,8)≤5421, A(26,8)≤9275, A(21,10)≤47, A(22,10)≤84, A(24,10)≤268, A(25,10)≤466, A(26,10)≤836, A(27,10)≤1585, A(25,12)≤55, and A(26,12)≤96. The method is based on the positive semidefiniteness of matrices derived from quadruples of words. This can be put as constraint in a semidefinite program, whose optimum value is an upper bound for A(n,d). The order of the matrices involved is huge. However, the semidefinite program is highly symmetric, by which its feasible region can be restricted to the algebra of matrices invariant under this symmetry. By block diagonalizing this algebra, the order of the matrices will be reduced so as to make the program solvable with semidefinite programming software in the above range of values of n and d.
Research graph
Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.
BZPEER is loading the nearby papers, people, topics and institutions for this page.