cov.RQ.beta0: Computationally cheap estimate for beta0 for cov.RQ.

In JimSkinner/spca: Structured PCA

Description

Computationally cheap estimate for beta0 for cov.RQ.

Usage

cov.RQ.beta0(X, locations, k)

Arguments

X	The dataset being analysed with stpca
locations	Matrix containing the location of each feature in rows
k	Latent dimensionality used in stpca

Examples

# Construct some synthetic data, initialise beta for the RQ kernel from this dataset.
library(functional)
n = 10; k = 4; dim=c(10, 10)
kern = Curry(cov.noisy.SE, beta=log(c(2, 0.4, 0.1)))
synth = synthesize_data_kern(n, k, dim, kern, noisesd=0.2)
beta0 = cov.RQ.beta0(synth$X, synth$grid, k)
stopifnot(length(beta0) == 3)
stopifnot(all(is.finite(beta0)))

JimSkinner/spca documentation built on May 7, 2019, 10:52 a.m.