Fellows' Presentation

Imprecise Computational Science: Closing Gaps, Forming Alloys

Photo by Tony RinaldoPhoto by Tony Rinaldo

Part of the 2016–2017 Fellows' Presentation Series

Lecture by Petros Koumoutsakos RI '17

Computing is revolutionizing our intellectual capacity to tackle complex problems. At the same time, the energy demands of computers are soaring while vast amounts of data are challenging the classical methods of computational model building. Petros Koumoutsakos's work addresses these challenges by focusing on fundamental computing patterns: multiscale modeling, uncertainty quantification, and their interfaces.

During his stay at Radcliffe, he is exploring how these interfaces can be translated into algorithms for computers equipped with a range of processor accuracies. He is interested in such questions as: How much energy must we consume for answering a particular scientific problem? Is it possible to design algorithms with a specific tradeoff between precision in computing and communication? Do these questions intersect with the social sciences and the arts?

Free and open to the public.