repfdr (version 1.0)

em.control: Control Parameters for the EM algorithm

Description

Input parameters for the EM algorithm.

Usage

em.control(pi.initial = NULL, max.iter = 10000, tol = 1e-12,
           nr.threads = 0, verbose = TRUE)

Arguments

pi.initial

Initial guess for the probabilities of the vectors of associations status. If NULL then 0.9 is assigned for the c(0,...,0) configuration and 0.1 is distributed uniformly for all other configurations.

max.iter

Maximum number of EM iterations.

tol

Tolerance (in maximum absolute difference between two EM iterations in estimated probabilities) before declaring convergence and stopping.

nr.threads

Number of processing threads to use. If zero (the default), will automatically detect the number of compute cores available and spawn one thread per core.

verbose

An indicator of whether to report progress (running iteration number) during computation.

Value

• A list with the input values.

Details

The function is used inside the control argument in repfdr and piem.

See Also

repfdr piem

Examples

data(zmat)
input3 <- ztobins(zmat, 3, plot = FALSE)
pbz    <- input3$pdf.binned.z
bz     <- input3$binned.z.mat

out <- repfdr(pbz,bz,"replication",
              control = em.control(pi.initial = c(0.48,rep(0.02,26)),
              verbose = TRUE, nr.threads = 1))
# convergence after 194 iterations; iterations are printed; run bit slower (1 thread)