GMIG has recently completed an analysis of coseismic deformation through the inverse-dislocation and fault-shape problems. The analysis uses variable elastic parameters and GPS or InSAR data to enhance our understanding of failure at faults.  

GMIG has produced a novel formulation simulating the dynamic evolution of spontaneous ruptures, governed by rate- and state-dependent friction laws. The analysis also considers interaction with seismic waves in a prestressed elastically deforming body. Researchers proved well-posedness with a multi-rate iterative coupling scheme based on the variational form of elastic-gravitational equations.

GMIG uses a state-of-the-art deep-learning to forecast fault-time-to-failure intervals. The approach uses an attention network, as well as unsupervised classification and selection of acoustic emissions. This physics-informed deep-neural-network architecture has been successfully tested in laboratory experiments. In the field, the research program is developing causal inference of slow-slip events, and researchers characterize tremors using hierarchical clustering and manifold embeddings.