gp_logLikelihood: Compute log likelihood function for Gaussian Process model.
In svdataman/gin: Gaussian Processes for time series inference

Description Usage Arguments Value Notes See Also

View source: R/gp_functions.R

gp_logLikelihood returns the log likelihood for a GP model.

1 2	gp_logLikelihood(theta, acv.model = NULL, tau = NULL, dat = NULL, PDcheck = TRUE, chatter = 0)

`theta`	(vector) parameters for covariance function the first element is the mean value mu
`acv.model`	(name) name of the function to compute ACV(tau \| theta)
`tau`	(matrix) N*N matrix of lags at which to compute ACF
`dat`	(matrix) an N * 3 matrix of data: 3 columns
`PDcheck`	(logical) use Matrix::nearPD to coerse the matrix
`chatter`	(integer) higher values give more run-time feedback

scalar value of log[likelihood(theta)]

Compute the log likelihood for Gaussian Process model with parameters theta given data \{t, y, dy\} and an (optional) N*N matrix of lags, tau. See algorithm 2.1 of Rasmussen & Williams (2006). The input data matrix dat should contain three columns: t, y, dy. t[i] and y[i] give the times and the measured values at those times. dy gives the 'error' on the measurements y, assumed to be independent Gaussian errors wih standard deviation dy. If dy is not present we assumine dy[i] = 0 for all i. The columns t, y, and dy are all n-element vectors.