Sparse Random Structures: Analysis and Computation (10w5033)

Arriving in Banff, Alberta Sunday, January 24 and departing Friday January 29, 2010


Emmanuel Candes (Stanford University)

Alan Edelman (Massachusetts Institute of Technology)

(University of California, Santa Barbara)

(University of Michigan)

(Queen's University)

Balint Virag (University of Toronto)


Our objectives:

We wish to extend to sparse and combinatorial structures the benefits
that random matrix theory has had on continuous and dense systems.
Topics will concentrate on applications to random structures
including random graphs, networks,compressed sensing, sparse matrices,
and low
rank approximation theory.

Our approach:

There are a number of outstanding references on random structures available to practitioners wanting to address a new problem they might
encounter. The goal of this workshop is Making this rich body of knowledge accessible to the non-specialist will jump-start the discovery of new applications of this theory. Introducing
mathematicians to a set of related applied problems can lead to the
development of newer and more powerful techniques.

We believe that bringing together the practitioners and the
mathematicians will jumpstart research in the area of sparse random
structures. By incorporating an explicit emphasis on the dialogue
between practitioners and theorists, we hope that an important
contribution of the workshop will be the development of sparse random
matrix models that adequately capture the essential complexities of
the real-world problems without being so complicated that theorists
cannot get answers for them. The initial conversations will surely
have to be on an application-by-application basis with moderated
discussion sessions between theorists and practitioners in the field
to ensure that aspects of the problem that could affect the solution
are not missed.

We wish to involve and envision a heavy role for young researchers
who can benefit from the interplay of disciplines represented so as to
obtain breakthroughs that are so valuable in this area.

Progress on this front is likely to be deliberate because there is an
art to model building which makes it difficult to rush. We believe
that a five day workshop will allow us to schedule just the right
amount of structured time for such informal conversations.

Longer term objective:
If the past is any indicator of the future than it seems as though
every time a new scientific or engineering community has re-discovered
random matrices then a whole new set of applications and possibilities
seems to open up. The fortuitous meeting of Montgomery and Dyson over
tea at Princeton was one such instance that led to the remarkable
connection between the Riemann-Zeta hypothesis and random matrix
theory A workshop that at once builds community among those actively
involved in the field and reaches out to a broader audience can
hopefully help forge another such connection.