Michael McKerns
Florida International University
Lecture Information:
- February 3, 2023
- 2:00 PM
- Zoom (Registartion Link)
Speaker Bio
Mike has a PhD in Physics from the University of Alabama at Birmingham. He started at Caltech as a postdoc in Materials Science, and soon transitioned to a Senior Scientist in Materials Science and Advanced Computing. Mike has been a Scientist in Information Sciences at Los Alamos National Laboratory since 2018, and has been developing open source scientific software for over twenty years. He is the primary author of over a dozen python packages, including mystic, pathos, klepto, and dill. His tools have been downloaded over .5B times.
Description
The mystic software provides tools that facilitate the creation of complex workflows for artificial intelligence, machine learning, optimization, and uncertainty quantification. The pathos software enables users to effortlessly convert python code to run in asynchronous parallel across hierarchical networks of threads, processors, nodes, and machines. Over a decade ago, both codes began as a summer project to build optimization software that could leverage parallel computing. mystic and pathos have evolved through the years through funding from the NSF, NNSA, DOE, and other sources, and today are leveraged by some of the largest companies, national laboratories, and research institutions in cutting- edge research and product development. This talk will introduce tools within mystic and pathos that provide workflows for the automated discovery of optimal physics-informed models, the automated control of instruments, and the robust design of experiments under uncertainty. We will also discuss the use of these tools in the development of better error mitigation algorithms in quantum computing, the design of more efficient quantum sensing experiments, the automated control of particle accelerators, the learning of robust flux limiters in fluid dynamics simulations, and the development of highly-accurate surrogates for the nuclear equation of state.
This event requires advance registration to receive Zoom link. Please register here.