Recently, the research group led by Dr. Liang Chao at the Institute for Disaster Management and Reconstruction (IDMR) of Sichuan University has published a new scientific peer-reviewed article in the interdisciplinary field of earthquake dynamic rupture and artificial intelligence (AI).

They published a research paper titled "RuptureNet2D, a Deep Neural Network Based Surrogate for Dynamic Earthquake Rupture Simulation in Two Dimensions" in the internationally renowned Journal of Geophysical Research: Solid Earth (Nature Index). It is accessible via: https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024JB030069?af=R

Zekang Gong and Ziyi Wang, undergraduate students from the Innovation Class of 2021 at the College are the co-first authors and Dr. Liang Chao is the corresponding author. Other co-authors included Dr. Andreas Nienkötter and Associate Professor Peng Xinyan from IDMR, Dr. Ren Chunmei from Peking University, and Associate Professor Wang Jianze from Sichuan University's College of Architecture and Environment.

While dynamic earthquake rupture simulation is crucial for hazard assessment, it is computationally expensive. As such, Dr. Liang and his team have tried to address this challenge by developing an end-to-end deep learning model, RuptureNet2D, which can efficiently predict earthquake rupture time and final slip, achieving a significant speedup (by factors of 1,000 to 100,000) compared to traditional simulations. This model provides a novel approach for the rapid analysis of complex earthquake processes and dynamic inversion, which, as a result, will help facilitate improved efficiency in dynamic inversion and seismic hazard assessment.