Title:Accelerating the design of soft matter materials using machine learning and statistical physics.
Time: May 14, 2024 (Tuesday) 10:00-11:00
Venue: Guangzhou International Campus C3-c204
All students and faculties are welcome to join this lecture.
Abstract:
Predicting the emergence of materials from a microscopic scale still faces significant challenges. Machine learning and reverse design have opened up new paradigms for understanding and preparing novel materials. Nevertheless, due to the significant contributions of entropy, multiscale and many-body interactions, thermodynamic non-equilibrium states, and the widespread presence of active systems, utilizing machine learning and reverse design to prepare soft matter materials still faces many difficulties. Soft matter systems with (partial) orientational and positional order, such as liquid crystals, quasicrystals, plastic crystals, and the omnipresent thermal noise in the study systems, make it extremely difficult to use machine learning to classify the states of matter for these materials. In this presentation, Professor Dijkstra will discuss and address the following scientific questions: Can we use machine learning to automatically identify the local structure of materials, study phase transitions in materials, classify phases, and find corresponding order parameters? Can we identify the dynamic pathways of phase transitions, and can we use machine learning for coarse-graining experimental models? Finally, Professor Dijkstra will demonstrate how machine learning can be used to reverse-design particle interactions to stabilize a material with special symmetry found in nature, namely quasicrystals.
Brief Biography：
Marjolein Dijkstra, a tenured professor in the Department of Physics at Utrecht University in the Netherlands, is a Fellow of the Dutch Royal Academy of Arts and Sciences. Professor Dijkstra has made remarkable research contributions in the field of reverse design of soft matter materials using computer simulations, providing strong theoretical