T1: Partial Inner Product w.r.t. the First Argument

In TMasak/surfcov: Random Surfaces' Covariance Estimation beyond Separability

Partial Inner Product w.r.t. the First Argument

Description

Calculates T_1(X x X, B_r), i.e. the partial inner product of the covariance of X with the current weight B_r resulting in a temporal kernel (proxy for A_r), which is then orthogonalized w.r.t A_1,...,A_r-1.

Usage

T1(X, A, B, sigma, r, N, K1, K2)

Arguments

X	data set, array of size N x K1 x K2
A	temporal kernels, array of size (>= r) x K1 x K1
B	spatial kernels, array of size (>= r) x K2 x K2
sigma	separable component scores, vector of size >= r
r	current sought separable rank
N	sample size
K1	temporal grid size
K2	spatial grid size

Value

r-th temporal kernel A_r, array of size K1 x K1

Examples

X <- array(rnorm(20*3*4),c(20,3,4))
A <- array(runif(2*3^2),c(2,3,3))
B <- array(runif(2*4^2),c(2,4,4))
T1(X, A, B, c(1,1), 2, 20, 3, 4)

TMasak/surfcov documentation built on April 25, 2022, 12:15 a.m.