Computing to Predict and to Understand
Petros Koumoutsakos - ETH Zürich, Switzerland
Session chair: Valeria Krzhizhanovskaya

Abstract : Computing is a domain of knowledge emerging from the fusion of models, algorithms, data and of course computers. From the prediction of weather to the analysis of massive amounts of data, Computing has become omnipresent in Science. Computing can be used to probe and understand complex systems in ways that may not be possible by observation or experiments. Computing can also be used to make predictions and take corresponding decisions. In this talk I wish to celebrate some successes of this mode of scientific inquiry and take time to discuss some questions. Do we understand what is it that we are computing ? How certain can we be on our predictions ? What is the relevance of the computer output to the problem we wanted to solve ? Should we even care about such issues ? I will argue that our times present a golden opportunity to answer such questions by the proper integration of Computing and Data Sciences. Computing is essential for the progress of humankind and we must understand its powers and limitations. I will argue that in times characterized by “automation” and “easy” data we must nurture human thinking (or Computing with and without Computers) for the present and future generations.

http://www.cse-lab.ethz.ch/member/petros-koumoutsakos/