Dr. Giuseppe Romano

MIT-IBM Watson AI Laboratory

Inverse Design of Thermal Transport Across Scales: From Energy Harvesting to Heat-Based Computing 

Heat is usually something we try to get rid of, but it can also be designed, steered, and even used to compute. Gradient-based optimization allows efficient design of materials and devices with even billions of degrees of freedom. The key enabler is the "adjoint method," which computes the gradient of a PDE solution with just one additional solve of a system comparable in complexity to the original. In this talk I will report on our recent efforts on inverse design of materials with tailored thermal transport properties, beginning with thermal metamaterials that cloak objects from thermal detection and control the transient dynamics of heat flow. I will then turn to the nanoscale, where nondiffusive effects, such as hydrodynamic and ballistic transport, are captured by the Boltzmann transport equation. Within this regime, I will show topology-optimized nanoporous membranes for thermal energy harvesting that obey minimum lengthscale constraints [1]. Finally, I will present inverse-designed structures that perform matrix-vector multiplication using heat itself as the signal carrier [2]. I will close with an overview of the software ecosystem supporting this work.
[1] G. Romano and S. G. Johnson, Struct. Multidisc. Optim. 65, 297 (2022).
[2] C. Silva and G. Romano, Phys. Rev. Appl. 25, 014073 (2026).