This PR contains the DataAssimilator class, which implements an EnKF update. Currently this class is free-floating and isn't used anywhere other than the tests.
This is a fairly vanilla EnKF implementation, it contains all of the basic functionality but not everything that we'll eventually want. The focus was just on getting something plausible that works. Specifically, it doesn't currently have:
Localization
Inflation
An enriched state vector that includes uncertain parameters
Any sense of distributed memory on different MPI processes
The max row sizes of the sparse matrices are also very conservative (i.e. large), which doesn't matter for the small test problems, but could be an issue for a real problem.
This PR contains the DataAssimilator class, which implements an EnKF update. Currently this class is free-floating and isn't used anywhere other than the tests.
This is a fairly vanilla EnKF implementation, it contains all of the basic functionality but not everything that we'll eventually want. The focus was just on getting something plausible that works. Specifically, it doesn't currently have:
The max row sizes of the sparse matrices are also very conservative (i.e. large), which doesn't matter for the small test problems, but could be an issue for a real problem.