Skip to contents

Data assimilation smoother

Source: R/aci-assimilation.R
aci_smoother.Rd

Generic reconstructing the hidden state from the whole observed record. The closed-form method is used for a cgns_model; a general stochastic_model is out of scope in this release.

Usage

aci_smoother(model, obs, ...)

# S3 method for class 'cgns_model'
aci_smoother(
  model,
  obs,
  filter = NULL,
  conditional = NULL,
  init = NULL,
  stepper = c("explicit", "implicit"),
  nsub = 1L,
  regularize = NULL,
  force_validate = FALSE,
  ...
)

# S3 method for class 'stochastic_model'
aci_smoother(model, obs, ...)

Arguments

model

A cgns_model or stochastic_model object.

obs

An observed trajectory, or anything as_obs() accepts.

...

Arguments passed to methods.

filter

Optional precomputed filter path; recomputed when NULL.

conditional

Optional aci_conditional_spec selecting a conditional ACI reduction; see aci_conditional().

init

Optional list with the initial hidden mean and cov.

stepper

Either "explicit" or "implicit".

nsub

Positive whole number of sub-steps taken per observation.

regularize

Covariance policy for this call; see aci_filter(). One record covers the whole call, so a filter recomputed here and the backward recursion that consumes it share the meta$regularization on the returned smoother.

force_validate

FALSE (the default) lets a filter that aci_filter() produced for this same model, observations and conditional specification, and that has not been altered since, skip the per-step re-validation of its covariances. Any other supplied path, including one that has been through saveRDS(), is validated in full as before. TRUE validates unconditionally. The smoother result is the same either way.

Value

An assimilation path: da_path_gaussian for the closed-form engine.

Methods (by class)

  • aci_smoother(cgns_model): Closed-form backward-ODE smoother for a conditional-Gaussian model.

  • aci_smoother(stochastic_model): Classed not-implemented condition for a general (non-CGNS) stochastic model.

See also

aci_filter(), aci()

Examples

m <- aci_dyad_model()
sim <- simulate(m, seed = 1, t_end = 2, dt = 0.01)
ob <- as_obs(sim)
f <- aci_filter(m, ob)
#> Warning: No init$cov supplied; using a diffuse prior. Discard an initial burn-in window when interpreting results.
aci_smoother(m, ob, filter = f)
#> <da_path_gaussian> kind = smoother, l = 1, N+1 = 201