Skip to contents

Initial-value factory for Model 2a chains

Source: R/make_inits_2a.R
make_inits_2a.Rd

Returns an inits function (of chain) for runjags::run.jags(). It reuses initsfunction() for the per-chain RNG seed/name and adds modest starting values for the new Model 2a nodes:

  • lambda: small positive starts (the first biomarker's loadings are constrained > 0 in the model, so starts must be positive);

  • mu.par: started at the hyperprior means. The within-biomarker random effects w and the factors f are left for JAGS to initialise from their priors (N(0, prec.par) and N(0,1)), which is robust.

Usage

make_inits_2a(n_antigen_isos, n_params, mu_hyp)

Arguments

n_antigen_isos

integer number of biomarkers.

n_params

integer number of kinetic parameters (5).

mu_hyp

A n_antigen_isos x n_params matrix of hyperprior means (the mu.hyp element of the priors).

Value

A function f(chain) returning a list of initial values.