This talk will discuss recent results in computational physics, computer graphics, and learning/data science. The first section will present a method for accurately simulating two-way solid-fluid coupling even when solids may be smaller than the size of a computational grid cell. Next, a numerical method for simulating materials with large surface energies (such as liquid metals) will be discussed. The third part of the talk will introduce techniques for obtaining sparse semantic solutions to inverse and optimization problems, with a case study of inferring facial expressions from RGB images. I will conclude with future research directions at the interfaces of simulation, graphics, learning, and data.
David Hyde is an Assistant Professor of Computer Science at Vanderbilt University. He was first a Regents Scholar at UCSB, earning a B.S. in Mathematics with highest honors at age 19. Hyde then earned a Ph.D. in Computer Science (with Distinction in Teaching) from Stanford, where he was a DoD NDSEG Fellow and a Gerald J. Lieberman Fellow. He also earned M.S. degrees in computer science and applied math. Most recently, David was a PIC Assistant Adjunct Professor in the Department of Mathematics at UCLA. Hyde's research has been supported by the Army Research Lab, the Department of Energy, and BHP Billiton. In an earlier life he helped build successful technology companies in quantum computing, databases, and data science.