Mean-field variational inference for a normal linear model.
Usage
vb_lm(
X,
y,
mu0,
Sigma0,
a0 = 0.01,
b0 = 0.01,
prior = 1L,
tol = 1e-08,
maxiter = 100L,
verbose = FALSE
)Arguments
- X
The design matrix
- y
The response vector
- mu0
The prior mean for beta
- Sigma0
The prior covariance for beta
- a0
The scale hyper-parameter
- b0
The shape hyper-parameter
- prior
The prior to be used, `1` - inverse-gamma(a0,b0), `2`- half-t(a0,b0)
- tol
Tolerance for convergence of the elbo
- maxiter
Maximum number of iterations allowed
- verbose
Print trace of the lower bound to console. Default is
FALSE.