The three differ in the estimator they fit, the class they carry and a field
or two naming the architecture. Everything else – the fitted estimator, the
prior, the two standardizers, the dimensions, the names, the simulation
counts, the device – is the same list in the same order, and every consumer
downstream (save_npe(), check_fit_alive(), cat_fit_common(),
summary()) reads it by name from all three.
Usage
new_nsbi_fit(de, prior, prep, class, extra = list())Arguments
- de
The fitted estimator. A ratio estimator travels in the same slot as a density estimator, since nothing downstream cares which it is until it comes time to score a
(theta, x)pair.- prior
The prior the simulations were drawn from.
- prep
The list from
prepare_simulations().- class
The class to stamp on the result.
- extra
Named list of fields specific to this fit type, inserted before
device.