init_chains: Initialize Chains

In EMC2: Bayesian Hierarchical Analysis of Cognitive Models of Choice

Initialize Chains

Description

Adds a set of start points to each chain. These start points are sampled from a user-defined multivariate normal across subjects.

Usage

init_chains(
  emc,
  start_mu = NULL,
  start_var = NULL,
  particles = 1000,
  cores_per_chain = 1,
  cores_for_chains = length(emc)
)

Arguments

emc	An emc object made by make_emc()
start_mu	A vector. Mean of multivariate normal used in proposal distribution
start_var	A matrix. Variance covariance matrix of multivariate normal used in proposal distribution. Smaller values will lead to less deviation around the mean.
particles	An integer. Number of starting values
cores_per_chain	An integer. How many cores to use per chain. Parallelizes across participant calculations.
cores_for_chains	An integer. How many cores to use to parallelize across chains. Default is the number of chains.

Value

An emc object

Examples

# Make a design and an emc object
design_DDMaE <- design(data = forstmann,model=DDM,
                           formula =list(v~0+S,a~E, t0~1, s~1),
                           constants=c(s=log(1)))

DDMaE <- make_emc(forstmann, design_DDMaE)
# set up our mean starting points (same used across subjects).
mu <- c(v_Sleft=-2,v_Sright=2,a=log(1),a_Eneutral=log(1.5),a_Eaccuracy=log(2),
       t0=log(.2))
# Small variances to simulate start points from a tight range
var <- diag(0.05, length(mu))
# Initialize chains, 4 cores per chain, and parallelizing across our 3 chains as well
# so 4*3 cores used.
DDMaE <- init_chains(DDMaE, start_mu = mu, start_var = var,
                     cores_per_chain = 1, cores_for_chains = 1)
# Afterwards we can just use fit
# DDMaE <- fit(DDMaE, cores_per_chain = 4)

EMC2 documentation built on April 11, 2025, 5:50 p.m.