A countable group is said to be homogeneous if whenever tuples of elements u, v satisfy the same first-order formulas there is an automorphism of the group sending one to the other. We had previously proved with Rizos Sklinos that free groups are homogeneous, while most surface groups aren't. In a joint work with Ayala Dente-Byron, we extend this to give a complete characterization of torsion-free hyperbolic groups that are homogeneous.
A countable group is said to be homogeneous if whenever tuples of elements u, v satisfy the same first-order formulas there is an automorphism of the group sending one to the other. We had previously proved with Rizos Sklinos that free groups are homogeneous, while most surface groups aren't. In a joint work with Ayala Dente-Byron, we extend this to give a complete characterization of torsion-free hyperbolic groups that are homogeneous.
Let Γ be a discrete group. A group Γ is called unitarisable if for any Hilbert space H and
any uniformly bounded representation π : Γ → B(H) of Γ on H there exists a bounded operator
S : H → H such that S^{−1}π(g)S is a unitary representation for any g ∈ Γ. It is well known that
amenable groups are unitarisable. It has been open ever since whether amenability characterises unitarisability of groups.
Dixmier: Are all unitarisable groups amenable?
One of the approaches to study unitarisability is related to the space of the Littlewood functions
An L^2-function on a finite volume hyperbolic surface is called non-autocorrelated if it is perpendicular to its image under A_r, the operator of averaging over the circle of radius r, where r is fixed. We show that the support of such a function is small, namely, it takes not more than (r+1) / exp(r/2) of the volume of the surface. In my talk, I'll prove this result, and show its connection to the equidistribution of the circle on a surface (proved by Nevo).
Abstract: When are two elements in a given group conjugate? We solve this problem for the group of tree almost-automorphisms. These are homeomorphisms of the tree boundary which locally look like tree automorphisms. The solution is tied with the dynamics of the group action on the tree boundary. This work is joint with W. Lederle.
Abstract: A subgroup is said to be almost normal if it is commensurable
to all of its conjugates. Even though there may not be a well-defined
quotient group, there is still a well-defined quotient space that admits
an isometric action by the ambient group. We can deduce many geometric
and algebraic properties of the ambient group by examining this action.
In particular, we will use quotient spaces to prove a relative version
of Stallings-Swan theorem on groups of cohomological dimension one. We
Abstract:
Measure equivalence of countable groups is a measure theoretic analogue
of quasi-isometry.
For example, any two lattices in the same Lie group are by definition
measure equivalent.
We prove that any countable group that is measure equivalent to Out(Fn)
is virtually isomorphic to Out(Fn). This is a joint work with Camille
Horbez.
Abstract:
A group G is stable in permutations if every almost-action of G on a finite set is close to some actual action. Part of the interest in this notion comes from the observation that a non-residually finite stable group cannot be sofic.
I will show that surface groups are stable in a flexible sense, that is if one is allowed to "add a few extra points" to the action. This is the first non-trivial stability result for a non-amenable group.
Combinatorial group theory began with Dehn's study of surface
groups, where he used arguments from hyperbolic geometry to solve the
word/conjugacy problems. In 1984, Cannon generalized those ideas to all
"hyperbolic groups", where he was able to give a solution to the
word/conjugacy problem, and to show that their growth function satisfies
a finite linear recursion. The key observation that led to his
discoveries is that the global geometry of a hyperbolic group is determined locally:
first, one discovers the local picture of G, then the recursive structure
