likelis: Linearised likelihood functions.

In naolsen/simm.fda: Simultaneous inference for Misaligned Functional Data

Linearised likelihood functions.

Description

Linearised likelihood functions.

Usage

like.nowarp(param, r, amp_cov, t, sig = FALSE, pr = FALSE, ...)

likelihood(param, r, amp_cov, t, param.w, Zis, warp_cov, tw, sig = FALSE,
  pr = FALSE, w = NULL, parallel = FALSE)

likelihood.lap(param, r, amp_cov, t, param.w, Zis, warp_cov, tw, sig = FALSE,
  pr = FALSE, w, parallel = FALSE)

Arguments

param	Amplitude parameters
r, Zis	Residual vectors/matrices
amp_cov	Amplitude covariance function
t, tw	Observation time points, warp time points
param.w	Warp parameters
warp_cov	Warp covariance function
sig	Return sigma or likelihood?
pr	Print option. Only used for debugging.
parallel	Calculate likelhood in parallel? Not available for like.nowarp

Details

likelihood is the default option (linearised likehood). like.par performs parallelized likelihood. likelihood.lap uses Laplace approximation. If w is the optimal posterior, then Laplace approximation and linearisation are the same, in general lik_laplace <= lik_lin. like.nowarp is used when there is no warping used (and thus no linearisation).

Value

-2*logL(parameters)

See Also

ppMulti

naolsen/simm.fda documentation built on June 28, 2022, 2:41 a.m.