sommer (version 4.3.5)

atc: atc covariance structure

Description

atc creates a diagonal covariance structure for specific levels of the random effect to be used with the mmec solver.

Usage

atc(x, levs, thetaC, theta)

Value

$res

a list with the provided vector and the variance covariance structure expected.

Arguments

x

vector of observations for the random effect.

levs

levels of the random effect to use for building the incidence matrices.

thetaC

an optional symmetric matrix for constraints in the variance-covariance components. The symmetric matrix should have as many rows and columns as the number of levels in the factor 'x'. The values in the matrix define how the variance-covariance components should be estimated:

0: component will not be estimated

1: component will be estimated and constrained to be positive

2: component will be estimated and unconstrained

3: component will be fixed to the value provided in the theta argument

theta

an optional symmetric matrix for initial values of the variance-covariance components. The symmetric matrix should have as many rows and columns as the number of levels in the factor 'x'. The values in the matrix define the initial values of the variance-covariance components that will be subject to the constraints provided in thetaC. If not provided, initial values will be calculated as:

theta* = diag(ncol(mm))*.05 + matrix(.1,ncol(mm),ncol(mm))

where mm is the incidence matrix for the factor 'x'. The values provided should be scaled by the variance of the response variable.

theta = theta*/var(y)

Author

Giovanny Covarrubias-Pazaran

References

Covarrubias-Pazaran G (2016) Genome assisted prediction of quantitative traits using the R package sommer. PLoS ONE 11(6): doi:10.1371/journal.pone.0156744

See Also

The function vsc to know how to use atc in the mmec solver.

Examples

x <- as.factor(c(1:5,1:5,1:5));x
atc(x, c("1","2"))
## how to use the theta and thetaC arguments:
# data(DT_example)
# DT <- DT_example
# theta <- diag(2)*2; theta # initial VCs
# thetaC <- diag(2)*3; thetaC # fixed VCs
# ans1 <- mmec(Yield~Env,
#              random= ~ vsc( atc(Env, levs=c("CA.2013", "CA.2011"),
#                             theta = theta,thetaC = thetaC),isc(Name) ),
#              rcov= ~ units, nIters = 1,
#              data=DT)
# summary(ans1)$varcomp