In the first talk we gave a brief outline of the contents of the course. In the rest of the semester we will get deeper into some topics. In the coming lecture ( and the next one) we will discuss Kazhdan property T and its connections with expanders and with first cohomology groups. No prior knowledge will be assumed.
The expander Chernoff bound states that random walks over expanders are good samplers, at least for a certain range of parameters. In this talk we will be interested in “Parity Samplers” that have the property that for any test set, about half of the sample sets see the test set an *even* number of times, and we will check whether random walks over expanders are good parity samplers. We will see that:
1. Random walks over expanders fare quite well with the challenge, but, 2. A sparse Random complex does much better.
The "PCP theorem" says that problems in NP are hard in a robust or stable way.
I will give a brief intro to PCPs (and explain the acronym) and then try to outline a proof of the PCP theorem based on "agreement expansion" which is a form of high dimensional expansion.
My aim is to show how high dimensional expansion is inherently present in PCP type questions.
Abstract:
I will introduce the notion of (PCP)-agreement expansion which is an important building block in PCPs constructions.
I will then show that a high dimensional expanders imply PCP-agreement expanders.
based on Joint work with Irit Dinur
Speaker : Tatiana Nagnibeda (University of Geneva)
Abstract: The definition of a Ramanujan graph extends naturally to infinite graphs: an infinite graph is Ramanujan if its spectral radius is not larger than (and hence equal to) the spectral radius of its universal covering tree. As with infinite families of finite graphs, it is interesting and non-trivial to understand, how much Ramanujan graphs resemble trees. I will discuss some results in this direction obtained in a joint work with Vadim Kaimanovich, by investigating ergodic properties of boundary actions of free groups.
Speaker: Steven Damelin (The American Mathematical Society)
Abstract:
A classical problem in geometry goes as follows. Suppose we are given two sets of $D$ dimensional data, that is, sets of points in $R^D$. The data sets are indexed
by the same set, and we know that pairwise distances between corresponding points are equal in the two data sets. In other words, the sets are isometric. Can this correspondence be extended
to an isometry of the ambient Euclidean space?
In this form the question is not terribly interesting; the answer has long known
In the recent theory of high dimensional expanders, the following open problem was raised by Gromov: Are there bounded degree high dimensional expanders?
For the definition of high dimensional expanders, we shall follow the pioneers of this field, and consider the notions of coboundary expanders (Linial-Meshulam) and topological expanders (Gromov).
In a recent work, building on an earlier work of Kaufman-Kazhdan-Lubotzky in dimension 2, we were able to prove the existence of bounded degree expanders according to Gromov, in every dimension.
There are several prominent computational problems for which
simple iterative methods are widely preferred in practice despite
an absence of runtime or performance analysis (or "worse", actual
evidence that more sophisticated methods have superior
performance according to the usual criteria). These situations
raise interesting challenges for the analysis of algorithms.
We are concerned in this work with one such simple method: a
classical iterative algorithm for balancing matrices via scaling
transformations. This algorithm, which goes back to Osborne and
The family of high rank arithmetic groups is a class of groups playing an important role in various areas of mathematics. It includes SL(n,Z), for n>2 , SL(n, Z[1/p] ) for n>1, their finite index subgroups and many more.
A number of remarkable results about them have been proven including; Mostow rigidity, Margulis Super rigidity and the Quasi-isometric rigidity.
