Saturday, May 22, 2010

"Every mathematics talk should have at least a little bit of fruit fly sex."

The 20th and 21st of May in the year 2010 saw a conglomeration of network scientists and statisticians in Oxford's Mathematical Institute for the "Oxford/Harvard Workshop on Networks and Statistics" (for which I was the primary organizer). We had three distinguished visitors from Harvard's statistics department: Edo Airoldi discussed dynamic network tomography, Joe Blitzstein discussed estimation in exponential random graph models, and Patrick Wolfe discussed point process models. Speakers from Oxford came from a wide variety of departments and research groups, including the Department of Statistics, the Nuffield Network of Network Researchers, the Oxford Internet Institute, the Neuroscience group, the Department of Physics, the Department of Engineering Science, and (of course) OCIAM. In addition to members of these research groups, participants also included faculty, postdocs, and graduate students from plant sciences, the Systems Biology Doctoral Training Centre, political science, the Oxford-Man Institute of Quantitative Finance, and even one statistician from Cambridge.

The format was perhaps a bit unusual, with plenty of time left open for discussions that allowed for much more discourse among participants than what typically occurs at workshops. To facilitate this, all speakers were encouraged to discuss open problems and to make a point of stressing not only what they did understand but more importantly what they did not understand and what problems they wanted to see solved. The discussion periods including extensive brainstorming on which problems were most desirable to solve and for which problems the situation was perhaps ripe to try to do it now. For example, sociologist Bernie Hogan stressed the need for mathematicians, statisticians, and physicists to develop good methods to study overlapping communities (so-called "soft partitioning") in networks, and several participants had interesting ideas regarding how one might try to do this.

In his plenary talk, which was held jointly with OCIAM's "Differential Equations and Applications" seminar (this was a talk on applications), Gero Miesenböck introduced us to the emerging field of "optogenetics", which seeks to develop genetic strategies for observing and controlling the function of brain circuits with light. He talk was fascinating, and it even contained fruit fly sex in it. (I think that all mathematics talks should contain at least a little bit of fruit fly sex in it.) It also included some fascinating time series and nonlinear oscillations. Gero stressed his desire for mathematical modelling to accompany his work---his field is currently entirely bereft of it, and it is crucially needed---and we hope to follow up his talk with a Friday morning workshop in the fall.

The Oxford/Harvard Workshop on Networks and Statistics seemed to be a great success, though there is a higher bar for success that I hope that this conference will achieve---namely, I want published papers that ultimately result from interactions and discussions that began at this workshop. Ask me again in 2 years to see if this more genuine measure of success has been achieved.

No comments: