CGMTorch: A Framework for Gradient-based Design of Computational Granular Metamaterials

Atoosa Parsa, Corey OHern, Rebecca Kramer-Bottiglio, Josh Bongard

Keywords: Unconventional Computing, Granular Metamaterials, Differentiable Simulation, Inverse Design

Abstract: Unconventional computing devices leverage the intrinsic dynamics of a physical substrate to perform fast, energy-efficient, and special-purpose computations. Granular metamaterials have great potential for creating such computing devices. However, there is no general framework for the inverse design of large-scale granular materials. Here, we develop a gradient-based optimization framework for harmonically driven granular materials to obtain a target wave response. Using this framework, we design basic logic gates in which mechanical vibrations carry the information at predetermined frequencies. Our findings show that a gradient-based optimization method can greatly expand the design space of computational metamaterials and provide the opportunity to systematically traverse their parameter space to find materials with the desired functionalities.

Submission Number: 53