Overview¶
diff-soup provides three major functionalities: the modeling module is used to specify physical phenomena of interest and discretize them on a given domain, the simulation model provides tools for integration of large-scale, stiff dynamical systems, and inference provides end-to-end workflows and tools for PDE-constrained optimization problems, statistical inference, and uncertainty quantification.
-
Modeling
Domains, discretizations, the Recipe API, and the problem interface the engines consume.
-
Simulation
Running a solve: adaptive implicit time stepping, the Newton corrector, and the sparse linear backends.
-
Inference
Fitting parameters to data and quantifying uncertainty.