I build physics-based and AI-accelerated simulators for the solid Earth, with a focus on AI for physical system simulation: user-friendly, reproducible, reliable and parallel open-source scientific software in geoscience and beyond.
A seismologist by training, I model earthquake rupture, multicycle fault behavior and ground shaking with parallel finite-element methods, and integrate those models with paleoseismic, geodetic and ground-motion observations. I joined the University of Texas Institute for Geophysics as its first hire dedicated to computational science, bridging geophysical research and high-performance computing. My work also spans geodynamics, InSAR, planetary interiors and climate modeling.
I believe software development should be driven by scientific questions, and that clean, tested, continuously refactored code frees scientists to do science.
Scenario earthquakes and ground shaking with EQdyna; observation-constrained multicycle dynamics for the southern San Andreas and Altyn Tagh faults; rate-and-state earthquake cycles on complex fault geometries with EQquasi; ground shaking over many cycles with the EQsimu coupling infrastructure; community code-verification benchmarks (SCEC).
EQdyna · EQquasi · EQsimu · EQdyna.2DcycleGraph neural network simulators that learn earthquake rupture dynamics from finite-element runs (EQGNS), and nonlinearly coupled permafrost dynamics from graph networks.
EQGNS · GNN surrogatesAn analytical and numerical toolset to constrain viscous anisotropy; mixed forward/inverse frameworks for faulting and adjoint-based joint inversion for material properties, including volcanic-arc rigidity from coseismic deformation.
FEniCS · adjoint methodsDevelopment of GMTSAR for next-generation InSAR satellites, and coupling InSAR with finite-element models to resolve compliant fault damage zones.
GMTSARInterior models of Mercury and the conditions for iron snow in its core, with the PIE inversion code.
PIECESM simulations of past and future El Niño behavior and aerosol–climate interactions.
CESMOpen-source codes I develop or co-develop.
Open Jupyter notebooks and MATLAB numerical tutorials developed for the Spring 2023 course at UT Austin. Course materials
Spring 2023Clean Code (R. C. Martin), Refactoring, 2nd ed. (M. Fowler), and Poor Charlie's Almanack (C. T. Munger), for clean scientific software and a multidisciplinary approach to research.