Computational Thinking: Building Bridges between Physical Models and Statistical Inference

Seminar | May 9 | 12-1:30 p.m. | 1011 Evans Hall

 Fernando Perez, University of California, Berkeley

 Department of Statistics

The scientific traditions of physics and applied mathematics have focused mainly on defining simplified models of the world amenable to analytical and numerical approximation. Today, the flood of real-world data at unprecedented levels of resolution and diversity creates opportunity for building richer scientific descriptions that combine statistical inference with "classical" models. I will discuss what tools and systems we must build to support scientific reasoning and inference on real systems informed by heterogeneous, multimodal data and only partial analytical and numerical solutions. What properties should our languages, libraries and environments have, to help us think computationally with fluidity across these boundaries? I will explore these questions from my own perspective building both numerical algorithms and environments for exploratory scientific computing.