Hst.sumup: Create Covariance Matrix

In widals: Weighting by Inverse Distance with Adaptive Least Squares

Description

Calculate the covariance matrix of all model covariates

Usage

Hst.sumup(Hst.ls, Hs = NULL, Ht = NULL)

Arguments

Hst.ls	Space-time covariates. A list of length τ, each element containing a n x p_st numeric matrix.
Hs	Spacial covariates. An n x p_s numeric matrix.
Ht	Temporal covariates. An τ x p_t numeric matrix.

Details

Important: The order of the arguments in this function is NOT the same as in the returned covariance matrix. The order in the covariance matrix is the same as in other functions in this package: Hs, Ht, Hst.ls.

Value

A (p_s+p_t+p_st) x (p_s+p_t+p_st) numeric, symmetrix, non-negative definite matrix.

Examples

	
tau <- 20
n <- 10
Ht <- cbind(sin(1:tau), cos(1:tau))

Hs <- cbind(rnorm(10), rnorm(n, 5, 49))

Hst.ls <- list()
for(tt in 1:tau) {
Hst.ls[[tt]] <- cbind(rnorm(n, 1, 0.1), rnorm(n, -200, 21))
}


Hst.sumup(Hst.ls, Hs, Ht)



########### standardize all covariates

x1 <- stnd.Hst.ls(Hst.ls, NULL)$sHst.ls
x2 <- stnd.Hs(Hs, NULL, FALSE)$sHs
x3 <- stnd.Ht(Ht, n)


Hst.sumup(x1, x2, x3)



## The function is currently defined as
function (Hst.ls, Hs = NULL, Ht = NULL) 
{
    tau <- length(Hst.ls)
    if(tau < 1) { tau <- nrow(Ht) }
    if(is.null(tau)) { tau <- 10 ; cat("tau assumed to be 10.", "\n") }
    n <- nrow(Hst.ls[[1]])
    if(is.null(n)) { n <- nrow(Hs) }
    big.sum <- 0
    for (i in 1:tau) {
        if (!is.null(Ht)) {
            Ht.mx <- matrix(Ht[i, ], n, ncol(Ht), byrow = TRUE)
        }
        else {
            Ht.mx <- NULL
        }
        big.sum <- big.sum + crossprod(cbind(Hs, Ht.mx, Hst.ls[[i]]))
    }
    return(big.sum)
  }

widals documentation built on Dec. 8, 2019, 1:07 a.m.