Examples#
Worked examples of the inverse problems fips solves, as Jupyter notebooks. They run when the documentation is built, so the output you see is what the current code produces.
- Gravity: density below the surface
The density of buried prisms from the gravity measured above them: a small, self-contained problem with a correlated prior kept as Kronecker factors, and a sweep over the regularization factor.
- A regional flux inversion
A year of monthly methane fluxes on a grid from daily concentrations at three sites: a sparse forward operator, a background as the offset, maps from
to_xarray, and the posterior of monthly totals.- Does the inversion recover a known flux?
Does the inversion recover a known flux? A world makes the data and the inversion solves them: coverage of the monthly totals, a mismatch assumed too small or too large, a bias on one site, and a truth of 1.5 times the prior.
- Which observations constrain a regional total?
Which observations constrain a regional total? The prior and posterior uncertainty of a region’s annual total and of a seasonal difference, what each site and each month of observations contributes, and which candidate site would help most.