Michael Luvalle

Background

About Michael Luvalle

I will start with my philosophy of statistics. This started before my Doctorate, when I was applying to grad school I thought that one could make significant progress in physical problems by incorporating the physics into the logic of the statistical investigation. In graduate school this took a back seat to learning the mathematical statistics of the day, but resurfaced with a vengeance when I started working at bell labs.The way I approach a problem is to start with the physical (or more generally substantive) knowledge around the problem, trying to understand enough to make sure the questions and approach to data are consistent with underlying substantive framework.In the framework of predicting the life of material systems, this resulted in studying time transfomations induced in degradation paths by kinetic theory as a way to approach accelerated testing, leading to step stress experiments as a way to differentiate between kinetic models which extrapolate differently, ultimately leading to demarcation and activity maps. In the framework of medium range prediction of weather, the stock market and perhaps societal dynamics (if you like science fiction) this leads through chaos theory to the use of delay map embedding, multiview embedding, and then to looking at the residuals of multiview embedding as a way to characterize the underlying dynamics, and the differences between real and model systems.Historically, post PhD at UC Davis, I started as an mathematical statistician, teaching at Kansas State University.At Bell Labs I started migrating first to applied statistics, then to using physical theory to generate statistical ideas. Major work has been in accelerated testing, First applied to Printed wiring boards, then generally across AT & T, then specifically to optical fiber issues. All of this focussed on learning how to extrapolate. In the last 12 years have added work on modeling and predicting chaotic systems, first in the context of environments devices sit in, then because it is interesting.Currently I'd define myself as a data scientist. Working on using statistics, data science, and physics to explore and model material systems (e.g. optical amplifier fibers under the influence of ionizing radiation, light and heat), geophysical systems (earthquakes, seasonal climate) and other chaotic systems.Since 2016 I've been at Rutgers, as teaching faculty, with my main time spent on trying to learn to teach well. I am striving still to reach all my students with the wonder of the subject. Research is necessarily slower.