Variational approximation for linear mixed model assuming Gaussian distribution priors on fixed and grouped effects. Priors on the variance parameters are Scaled-Inverse-Wishart distributions.
Usage
vb_lmm(
y,
X,
Zlist,
J,
R,
mu_beta0,
Sigma_beta0,
xi_sigma,
Lambda_sigma,
xi_k,
Lambda_k,
tol = 1e-08,
maxiter = 100L,
verbose = FALSE,
trace = FALSE
)Arguments
- y
The response vector
- X
The design matrix
- Zlist
Collection of group design matrices
- J
First dimension of each Z in Zlist (e.g. number of subjects/sites)
- R
Second dimension of each Z in Zlist (e.g. number of variables, intercept and slope would be R = 2)
- mu_beta0
The prior mean for beta
- Sigma_beta0
The prior covariance for beta
- xi_sigma
The first prior parameter for covariance Sigma
- Lambda_sigma
The second prior parameter for covariance Sigma
- xi_k
A vector of first covariance parameters for hierarchical covariance
- Lambda_k
A list of second covariance parameters for hierarchical covariance
- tol
Tolerance level
- maxiter
Maximum iterations
- verbose
Print trace of the lower bound to console. Default is
FALSE.- trace
Print a trace of `mu` to console.