Construct a BREAD prior for the summary-mode model

Default prior for fit_bread_summary(): weakly-informative Normal-Inverse-Gamma. mu0 defaults to a zero vector (length inferred from the design) and Lambda0 to lambda0 * I (diagonal prior precision, inferred at fit time). The inverse-gamma hyperparameters default to a0 = b0 = 0.001, which is approximately Jeffreys.

Usage

bread_prior(mu0 = NULL, Lambda0 = NULL, lambda0 = 0.01, a0 = 0.001, b0 = 0.001)

Arguments

mu0

Optional prior mean vector for the coefficients. If NULL, a zero vector of the correct dimension is used at fit time.

Lambda0

Optional prior precision matrix. If NULL, lambda0 * I is used at fit time.

lambda0

Scalar prior precision used when Lambda0 is NULL. Default 0.01 (weak).

a0, b0

Inverse-gamma hyperparameters for the residual variance. Defaults a0 = b0 = 0.001.

Value

A list with class "bread_prior".