R/p_beta.R

In aashen12/BARS: Bayesian Adaptive Regression Splines

library(mvtnorm)

p_beta <- function(sig_sq, tau_sq = 1000, X, y) {
  p = ncol(X) - 1
  sig <- solve( (1/sig_sq) * (t(X) %*% X) + (1/tau_sq) * diag(p+1) )
  mu <- (1/sig_sq) * sig %*% t(X) %*% y
  rmvnorm(1, mean = mu, sigma = sig)
} # full conditional for the regression coefficients