m2LL_smw: Minus 2 time the log-likelihood for reduced rank spatial...
In ericwfox/slmrf:

m2LL_smw

R Documentation

Minus 2 time the log-likelihood for reduced rank spatial linear model.

Description

Minus 2 time the log-likelihood for reduced rank spatial linear model.

Usage

m2LL_smw(theta, y, X, dist_K, dist_S, est_meth, cov_type = "exponential")

Arguments

`theta`	Numeric vector of covariance parameters (use log transformed nugget, partial sill, and range, in that order).
`y`	Numeric response vector.
`X`	Design matrix.
`dist_K`	distance matrix for knot locations.
`dist_S`	distance matric between data locations and knots (`nrow` is number of data points and `ncol` is number of knots).
`est_meth`	character: Estimation method: 'REML' for restricted maximum likelihood (default), or 'ML' for maximum likelihood.
`cov_type`	character; type of covariance type of covariance model: `exponential`, `gaussian`, or `spherical`; default is `exponential`.

Value

Minus 2 times the log-likelihood for reduced rank spatial linear model using Sherman-Morrison-Woodbury formula. See Cressie FRK paper equation 2.16 for decomposition.