Past Combinatorial Theory Seminar

Zarankiewicz showed that no $K_{r+1}$-free graph with minimum degree exceeding $(r-1)n/r$ can exist. This was generalised by Erdös and Stone, who showed that $K_{r+1}$ may be replaced by any graph $H$ with chromatic number $r+1$ at the cost of a $o(n)$ term added to the minimum degree.

Andr\'asfai, Erdös and S\'os proved a stability result for Zarankiewicz' theorem: $K_{r+1}$-free graphs with minimum degree exceeding $(3r-4)n/(3r-1)$ are forced to be $r$-partite. Recently, Alon and Sudakov used the Szemer\'edi Regularity Lemma to obtain a corresponding stability result for the Erdös-Stone theorem; however this result was not best possible. I will describe a simpler proof (avoiding the Regularity Lemma) of a stronger result which is conjectured to be best possible.

HTML clipboard /*-->*/ /*-->*/ We introduce a "Polya choice" urn model combining elements of the well known "power of two choices" model and the "rich get richer" model. From a set of $k$ urns, randomly choose $c$ distinct urns with probability proportional to the product of a power $\gamma>0$ of their occupancies, and increment one with the smallest occupancy. The model has an interesting phase transition. If $\gamma \leq 1$, the urn occupancies are asymptotically equal with probability 1. For $\gamma>1$, this still occurs with positive probability, but there is also positive probability that some urns get only finitely many balls while others get infinitely many.

HTML clipboard /*-->*/ /*-->*/

Fix a positive integer $k$, and consider the class of all graphs which do not have $k+1$  vertex-disjoint cycles.  A classical result of Erdos and P\'{o}sa says that each such graph $G$ contains a blocker of size at most $f(k)$.  Here a {\em blocker} is a set $B$ of vertices such that $G-B$ has no cycles.

 

We give a minor extension of this result, and deduce that almost all such labelled graphs on vertex set $1,\ldots,n$ have a blocker of size $k$.  This yields an asymptotic counting formula for such graphs; and allows us to deduce further properties of a graph $R_n$ taken uniformly at random from the class: we see for example that the probability that $R_n$ is connected tends to a specified limit as $n \to \infty$.

 

There are corresponding results when we consider unlabelled graphs with few disjoint cycles. We consider also variants of the problem involving for example disjoint long cycles.

 

This is joint work with Valentas Kurauskas and Mihyun Kang.

HTML clipboard /*-->*/ /*-->*/

Gupta et al. introduced a very general multi-commodity flow problem in which the cost of a given flow solution on a graph $G=(V,E)$ is calculated by first computing the link loads via a load-function l, that describes the load of a link as a function of the flow traversing the link, and then aggregating the individual link loads into a single number via an aggregation function.

 

We show the existence of an oblivious routing scheme with competitive ratio $O(\log n)$ and a lower bound of $\Omega(\log n/\logl\og n)$ for this model when the aggregation function agg is an $L_p$-norm.

 

Our results can also be viewed as a generalization of the work on approximating metrics by a distribution over dominating tree metrics and the work on minimum congestion oblivious. We provide a convex combination of trees such that routing according to the tree distribution approximately minimizes the $L_p$-norm of the link loads. The embedding techniques of Bartal and Fakcharoenphol et al. [FRT03] can be viewed as solving this problem in the $L_1$-norm while the result on congestion minmizing oblivious routing solves it for $L_\infty$. We give a single proof that shows the existence of a good tree-based oblivious routing for any $L_p$-norm.

One of the main aims in the theory of percolation is to find the `critical probability' above which long range connections emerge from random local connections with a given pattern and certain individual probabilities. The quintessential example is Kesten's result from 1980 that if the edges of the square lattice are selected independently with probability $p$, then long range connections appear if and only if $p>1/2$.  The starting point is a certain self-duality property, observed already in the early 60s; the difficulty is not in this observation, but in proving that self-duality does imply criticality in this setting.

Since Kesten's result, more complicated duality properties have been used to determine a variety of other critical probabilities. Recently, Scullard and Ziff have described a very general class of self-dual percolation models; we show that for the entire class (in fact, a larger class), self-duality does imply criticality.

Let $G=(V,E)$ be a graph with weights $\{w_e : e \in E\}$, and assume all weights are distinct. If $G$ is finite, then the well-known Prim's algorithm constructs its minimum spanning tree in the following manner. Starting from a single vertex $v$, add the smallest weight edge connecting $v$ to any other vertex. More generally, at each step add the smallest weight edge joining some vertex that has already been "explored" (connected by an edge) to some unexplored vertex.

If $G$ is infinite, however, Prim's algorithm does not necessarily construct a spanning tree (consider, for example, the case when the underlying graph is the two-dimensional lattice ${\mathbb Z}^2$, all weights on horizontal edges are strictly less than $1/2$ and all weights on vertical edges are strictly greater than $1/2$).

The behavior of Prim's algorithm for *random* edge weights is an interesting and challenging object of study, even
when the underlying graph is extremely simple. This line of research was initiated by McDiarmid, Johnson and Stone (1996), in the case when the underlying graph is the complete graph $K_n$. Recently Angel et. al. (2006) have studied Prim's algorithm on regular trees with uniform random edge weights. We study Prim's algorithm on $K_n$ and on its infinitary analogue Aldous' Poisson-weighted infinite tree. Along the way, we uncover two new descriptions of the Poisson IIC, the critical Poisson Galton-Watson tree conditioned to survive forever.

Joint work with Simon Griffiths and Ross Kang.

Let $F$ be a uniformly distributed random $k$-SAT formula with $n$ variables and $m$ clauses. We present a polynomial time algorithm that finds a satisfying assignment of $F$ with high probability for constraint densities $m/n

Let $A$ be a finite set in some ambient group. We say that $A$ is a $K$-approximate group if $A$ is symmetric and if the set $A.A$ (the set of all $xy$, where $x$, $y$ lie in $A$) is covered by $K$ translates of $A$. I will illustrate this notion by example, and will go on to discuss progress on the "rough classification" of approximate groups in various settings: abelian groups, nilpotent groups and matrix groups of fixed dimension. Joint work with E. Breuillard.

The Gilbert model of a random geometric graph is the following: place points at random in a (two-dimensional) square box and join two if they are within distance $r$ of each other. For any standard graph property (e.g.  connectedness) we can ask whether the graph is likely to have this property.  If the property is monotone we can view the model as a process where we place our points and then increase $r$ until the property appears.  In this talk we consider the property that the graph has a Hamilton cycle.  It is obvious that a necessary condition for the existence of a Hamilton cycle is that the graph be 2-connected. We prove that, for asymptotically almost all collections of points, this is a sufficient condition: that is, the smallest $r$ for which the graph has a Hamilton cycle is exactly the smallest $r$ for which the graph is 2-connected.  This work is joint work with Jozsef Balogh and B\'ela Bollob\'as

For graphs $L_1,\dots,L_k$, the Ramsey number $R(L_1,\ldots,L_k)$ is the minimum integer $N$ such that for any edge-colouring of the complete graph $K_N$ by $k$ colours there exists a colour $i$ for which the corresponding colour class contains $L_i$ as a subgraph.

In this talk, we shall discuss recent developments in the case when the graphs $L_1,\dots,L_k$ are all cycles and $k\ge2$.

There are many theorems concerning cycles in graphs for which it is natural to seek analogous results for directed graphs. I will survey

recent progress on certain questions of this type. New results include

(i) a solution to a question of Thomassen on an analogue of Dirac’s theorem

for oriented graphs,

(ii) a theorem on packing cyclic triangles in tournaments that “almost” answers a question of Cuckler and Yuster, and

(iii) a bound for the smallest feedback arc set in a digraph with no short directed cycles, which is optimal up to a constant factor and extends a result of Chudnovsky, Seymour and Sullivan.

These are joint work respectively with (i) Kuhn and Osthus, (ii) Sudakov, and (iii) Fox and Sudakov.

We consider Markovian models on graphs with local dynamics. We show that, under suitable conditions, such Markov chains exhibit both rapid convergence to equilibrium and strong concentration of measure in the stationary distribution. We illustrate our results with applications to some known chains from computer science and statistical mechanics.

Consider the Erdos-Renyi random graph $G(n,p)$ inside the critical window, so that $p = n^{-1} + \lambda n^{-4/3}$ for some real \lambda. In

this regime, the largest components are of size $n^{2/3}$ and have finite surpluses (where the surplus of a component is the number of edges more than a tree that it has). Using a bijective correspondence between graphs and certain "marked random walks", we are able to give a (surprisingly simple) metric space description of the scaling limit of the ordered sequence of components, where edges in the original graph are re-scaled by $n^{-1/3}$. A limit component, given its size and surplus, is obtained by taking a continuum random tree (which is not a Brownian continuum random tree, but one whose distribution has been exponentially tilted) and making certain natural vertex identifications, which correspond to the surplus edges. This gives a metric space in which distances are calculated using paths in the original tree and the "shortcuts" induced by the vertex identifications. The limit of the whole critical random graph is then a collection of such

metric spaces. The convergence holds in a sufficiently strong sense (an appropriate version of the Gromov-Hausdorff distance) that we are able to deduce the convergence in distribution of the diameter of $G(n,p)$, re-scaled by $n^{-1/3}$, to a non-degenerate random variable, for $p$ in the critical window.

This is joint work (in progress!) with Louigi Addario-Berry (Universite de Montreal) and Nicolas Broutin (INRIA Rocquencourt).

Random partial orders and random linear extensions

Several interesting models of random partial orders can be described via a

process that builds the partial order one step at a time, at each point

adding a new maximal element. This process therefore generates a linear

extension of the partial order in tandem with the partial order itself. A

natural condition to demand of such processes is that, if we condition on

the occurrence of some finite partial order after a given number of steps,

then each linear extension of that partial order is equally likely. This

condition is called "order-invariance".

The class of order-invariant processes includes processes generating a

random infinite partial order, as well as those that amount to taking a

random linear extension of a fixed infinite poset.

Our goal is to study order-invariant processes in general. In this talk, I

shall explain some of the problems that need to be resolved, and discuss

some of the combinatorial problems that arise.

(joint work with Malwina Luczak)

Regions of a link diagram can be colored in black and white in a checkerboard manner. Putting a vertex in each black region and connecting two vertices by an edge if the corresponding regions share a crossing yields a planar graph. In 1987 Thistlethwaite proved that the Jones polynomial of the link can be obtained by a specialization of the Tutte polynomial of this planar graph. The goal of my talk will be an explanation of a generalization of Thistlethwaite's theorem to virtual links. In this case graphs will be embedded into a (higher genus, possibly non-oriented) surface. For such graphs we used a generalization of the Tutte polynomial called the Bollobas-Riordan polynomial. For graphs on
surfaces the natural duality can be generalized to a duality with respect to a subset of edges. The generalized dual graph might be embedded into a different surface. I will explain a relation between the Bollobas-Riordan polynomials of dual graphs. This relation unifies various Thistlethwaite type theorems.

Parity games are simple two-player, infinite-move games particularly useful in Computer Science for modelling non-terminating reactive systems and recursive processes.  A longstanding open problem related to these games is whether the winner of a parity game can be decided in polynomial time.  One of the most promising algorithms to date is a strategy improvement algorithm of Voege and Jurdzinski, however no good bounds are known on its running time.

In this talk I will discuss how the problem of finding a winner in a parity game can be reduced to the problem of locally finding a global sink on an acyclic unique sink oriented hypercube.  As a consequence, we can improve (albeit only marginally) the bounds of the strategy improvement algorithm.

This talk is similar to one I presented at the InfoSys seminar in the Computing Laboratory in October.

In the first part of the talk we will introduce the notion of property testing and briefly discuss some results in testing graph properties in the framework of property testing.

Then, we will discuss a recent result about testing expansion in bounded degree graphs. We focus on the notion of vertex-expansion:   \newline an $a$-expander is a graph $G = (V,E)$ in which every subset $U$ of $V$ of at most $|V|/2$ vertices has a neighborhood of size at least $a|U|$. Our main result is that one can distinguish good expanders from graphs that are far from being weak expanders in time approximately $O(n^{1/2})$.

We design a property testing algorithm that accepts every $a$-expander with probability at least 2/3 and rejects every graph that is $\epsilon$-far from an $a^*$-expander with probability at least 2/3, where $a^* = O(a^2/(d^2 log(n/\epsilon)))$, $d$ is the maximum degree of the graphs, and a graph is called $\epsilon$-far from an $a^*$-expander if one has to modify (add or delete) at least $\epsilon d n$ of its edges to obtain an $a^*$-expander. The algorithm assumes the bounded-degree graphs model with adjacency list graph representation and its running time is $O(d^2 n^{1/2} log(n/\epsilon)/(a^2 \epsilon^3))$.

This is a joint work with Christian Sohler.