ECE Seminar: Giuseppe Romano

ECE Seminar: Giuseppe Romano

Title: Accelerated Materials Discovery with Differentiable Programming: From Thermal Energy Harvesters to Chiplet Design

Abstract: Automatic Differentiation (AD) software initially developed for machine learning may benefit applications beyond neural networks. A prominent example is given by physics solvers, where language-wide differentiability enables large-scale inverse design. In this talk, I will report on our recent materials and systems optimization efforts, along with the implemented open-source software. The first application is thermal energy harvesting, for which we identify an optimal nanostructured Si membrane. This system, which has a minimum feature of about 100 nm, has been fabricated with the Focused Ion Beam (FIB) at MIT.nano. The second application focuses on the inverse design of spatiotemporal modulated materials, a class of systems that can potentially unlock new physics, such as time interfaces [2]. The latest application pertains to chiplets floorplan design, where, in collaboration with the MIT-IBM Watson AI Lab, we designed a framework for minimizing the maximum temperature during operation. This thermally aware approach, combined with the maximum wire length constraint, enables a faster design compared to common gradient-free approaches. In the latest part of the talk, I will briefly describe MatInverse, the in-house JAX-based software for multiphysics inverse design.

[1] G. Romano and S. G. Johnson. "Inverse design in nanoscale heat transport via interpolating interfacial phonon transmission." Structural and Multidisciplinary Optimization 65.10 (2022): 297.

[2] H. Moussa et al. "Observation of temporal reflection and broadband frequency translation at photonic time interfaces." Nature Physics (2023): 1-6.

Bio: Giuseppe Romano is a research scientist at MIT. His research integrates differentiable programming and multiscale modeling to accelerate materials discovery for energy and computing applications. He maintains several scientific open-source packages, including OpenBTE, a tool for nanoscale heat transfer. In 2018, he was a visiting scientist at the NASA Jet Propulsion Lab. He joined MIT in 2010 after receiving his Ph.D. in Electrical Engineering from the University of Rome Tor Vergata.