Apr 23, 2024
3:45pm - 4:00pm
Room 320, Level 3, Summit
Chrisy Xiyu Du1,2,Ella King3,2,Qian-Ze Zhu2,Samuel Schoenholz4,Michael Brenner2
University of Hawai'i at Mānoa1,Harvard University2,Simons Foundation3,OpenAI4
Chrisy Xiyu Du1,2,Ella King3,2,Qian-Ze Zhu2,Samuel Schoenholz4,Michael Brenner2
University of Hawai'i at Mānoa1,Harvard University2,Simons Foundation3,OpenAI4
Soft Materials are ubiquitous in everyday life and are crucial in many different forms of revolutionary technologies. One property of Soft Materials is their ability to self-assemble into intricate structures from a finite set of building blocks with continuously tunable parameters. This giant design space of building blocks is a double-edged sword: on one side it provides researchers infinite possibilities to design building blocks for targeted functions, while on the other side it might take forever to search the design space. Here, we propose a new inverse design method that captures both interaction and geometry of building blocks. By enabling rigid body functionalities in JAX-MD, an end-to-end differentiable molecular dynamics engine, we can create soft materials model with components that are simple enough to design yet powerful enough to capture complex materials properties. In this talk, I will discuss the implementation of the methods alongside examples to showcase its potential applications.