Computational controls for TDMM fits — tdmm.control • TDMM

Construct a named computational control list for tdmm(). tdmm.control() is retained as an alias of tdmm_control(). A partial named list can also be passed directly.

Usage

tdmm.control(chains = 3L, iterations = 10000L, burn_in = 3000L,
  thin = 5L, adapt = 5000L, seed = NULL, cores = NULL,
  parallel = FALSE)
tdmm_control(chains = 3L, iterations = 10000L, burn_in = 3000L,
  thin = 5L, adapt = 5000L, seed = NULL, cores = NULL,
  parallel = FALSE)

Arguments

chains: Number of independent MCMC chains.
iterations: Total iterations per chain, including burn-in.
burn_in: Burn-in iterations per chain. Must be less than iterations.
thin: Posterior thinning interval.
adapt: JAGS adaptation iterations.
seed: Optional nonnegative integer seed for the fitter.
cores: Optional positive integer core count when parallel execution is enabled.
parallel: Use separate chains on forked workers when TRUE.

Details

Individual MCMC arguments remain available. Supplying the same setting in both places is an error. An omitted control list retains the serial defaults; the tdmm.parallel() compatibility wrapper retains its existing adaptation default. Parallel execution uses macOS/Linux forked workers. Evaluation and plotting grids are selected by plot() or plot.tdmm(), rather than by fitting control. Setting a seed affects the R random-number generator.

Value

A named list of computational settings.

Examples

tdmm.control(chains = 3, iterations = 10000, burn_in = 3000,
  seed = 218)
#> $chains
#> [1] 3
#> 
#> $iterations
#> [1] 10000
#> 
#> $burn_in
#> [1] 3000
#> 
#> $thin
#> [1] 5
#> 
#> $adapt
#> [1] 5000
#> 
#> $seed
#> [1] 218
#> 
#> $cores
#> NULL
#> 
#> $parallel
#> [1] FALSE
#>