The Robert S. Pepper Distinguished Professor
Electrical Engineering and Computer Sciences (EECS) department at U.C. Berkeley.
In this talk, he will argue that deterministic models have historicallyproven extremely valuable in engineering, despite fundamental limits. I examine the role that models play in engineering and contrast it with the role that they play in science, and I argue that determinism is an
extraordinarily valuable property in engineering, even more than science. I will then show that deterministic models for cyber-physical systems, which combine computation with physical dynamics, remain elusive in practice. I will argue that the next big advance in engineering methods must include deterministic models for CPS, and I
will show that such models are both possible and practical. I will then examine some fundamental limits of determinism, showing that chaos limits its utility for prediction, and that incompleteness means that at least for CPS, nondeterminism is inevitable.