PR #130 introduced breaking changes which effect the explicit calibration strategy. As it exists now, the explicit strategy is not very optimal, as it runs the entire calibration look (including model execution) for each calibration_catchment explicitly defined by the realization. This can be done more efficiently with a many-to-many connection of calibration features with evaluation features and a smarter implementation of the search space semantics, similar to the independent strategy.
The calibration_catchment needs to be deprecated and removed and the explicit strategy refactored to implement the intended semantics via, likely, a list of CalibrationSets of size one.
PR #130 introduced breaking changes which effect the
explicit
calibration strategy. As it exists now, the explicit strategy is not very optimal, as it runs the entire calibration look (including model execution) for eachcalibration_catchment
explicitly defined by the realization. This can be done more efficiently with a many-to-many connection of calibration features with evaluation features and a smarter implementation of the search space semantics, similar to theindependent
strategy.The
calibration_catchment
needs to be deprecated and removed and the explicit strategy refactored to implement the intended semantics via, likely, a list ofCalibrationSet
s of size one.