
Single forward filter step (scalar right-covariance, sigma-squared times R)
Source:R/func_FF.R
FF_1step_R_sigma2R.RdPerforms one step of the forward filter under the model where the right
covariance of \(Y\) is \(\sigma^2 R\) with \(\sigma^2 \sim IG(n, d)\).
The inverse of \(R\) (Rinv) must be precomputed and passed in.
Arguments
- Yt
Numeric matrix. Data matrix at time \(t\) (\(N \times S\)).
- Ft
Numeric matrix. Covariate matrix at time \(t\) (\(N \times p\)).
- Gt
Numeric matrix. State transition matrix at time \(t\) (\(p \times p\)).
- Wt
Numeric matrix. State noise left-covariance at time \(t\) (\(p \times p\)).
- Vt
Numeric matrix. Observation noise left-covariance at time \(t\) (\(N \times N\)).
- mt_1
Numeric matrix. Filtered state mean at \(t-1\) (\(p \times S\)).
- Mt_1
Numeric matrix. Filtered state left-covariance at \(t-1\) (\(p \times p\)).
- nt_1
Numeric scalar. Shape parameter of \(\sigma^2 | D_{t-1}\).
- Dt_1
Numeric scalar. Rate parameter of \(\sigma^2 | D_{t-1}\).
- Rinv
Numeric matrix. Precomputed inverse of the spatial correlation matrix \(R\) (\(S \times S\)).
- delta
Discount factor. Not implemented: values other than 1 give a warning and are ignored. Defaults to
1.
Value
A named list with updated filtering parameters:
- nt
Updated shape parameter.
- Dt
Updated rate parameter.
- at
One-step-ahead state mean (\(p \times S\)).
- At
One-step-ahead state left-covariance (\(p \times p\)).
- mt
Filtered state mean (\(p \times S\)).
- Mt
Filtered state left-covariance (\(p \times p\)).
- delta
Always 1; discounting is not implemented.
Examples
set.seed(1)
dat <- gen_ffbs_data(N = 5, S = 3, p = 2, nT = 1)
par <- dat$para
R <- 0.5^abs(outer(1:3, 1:3, "-"))
one <- FF_1step_R_sigma2R(dat$Y[[1]], par$F0, par$G0, par$W0, par$V0,
par$m0, par$M0, nt_1 = 3, Dt_1 = 1,
Rinv = solve(R))
c(nt = one$nt, Dt = one$Dt)
#> nt Dt
#> 10.500000 1.523951
one$mt
#> [,1] [,2] [,3]
#> [1,] -0.8310570 -0.3986266 0.1869505
#> [2,] -0.1131768 1.0532239 -0.5331634