An End-to-End Differentiable Platform for Integrated Subsurface Modeling

GitHub | Documentation

GeoBrain is an open, modular, and extensible platform for Geoscientific Bayesian Reasoning with Artificial Intelligence, designed specifically for integrated subsurface modeling.

By combining differentiable physics, Bayesian inference, and deep learning, GeoBrain enables end-to-end workflows for subsurface characterization -- from geomodeling and rock physics to geophysical simulation and inversion.

• Differentiable Multiphysics Modeling -- Geostatistics, rock physics, wave propagation, and reservoir simulation in a unified computational graph
• 70+ Rock Physics Models -- Effective medium theories, granular media, fluid substitution, empirical relations, anisotropy, and resistivity
• Wave Physics -- 2D/3D acoustic & elastic FDTD solvers, AVO modeling, full waveform inversion
• Flow Simulation -- Differentiable two-phase reservoir simulation with dynamic well control
• Bayesian Samplers -- SVGD, HMC, NUTS, Langevin Dynamics for rigorous uncertainty quantification
• Deep Learning Integration -- Deep Image Prior, autoencoders, GANs, VAEs, and diffusion models
• Real-World Applications -- CO₂ storage characterization (Illinois Basin, Sleipner CCS site)

git clone https://github.com/GeoBrain-Project/GeoBrain.git
cd GeoBrain
pip install -e ".[all]"

See the documentation for tutorials and API reference.