Simulate data for ABC inference using specified priors

This is a core function for ABC inference using demografr. It generates simulation replicates and computes summary statistic for the next step of an inference procedure, which is the ABC estimation itself.

simulate_abc(
  model,
  priors,
  functions,
  observed,
  iterations,
  sequence_length,
  recombination_rate,
  mutation_rate = 0,
  data = NULL,
  format = c("ts", "files"),
  file = NULL,
  packages = NULL,
  attempts = 1000,
  engine = NULL,
  model_args = NULL,
  engine_args = NULL
)

Arguments

model: Either a slendr model generating function (in which case engine must be either "msprime" or "slim", i.e. one of the two of slendr's simulation back ends), or a path to a custom user-defined SLiM or msprime script (in which case engine must be "custom").
priors: A list of prior distributions to use for sampling of model parameters
functions: A named list of summary statistic functions to apply on simulated tree sequences
observed: A named list of observed summary statistics
iterations: How many simulation replicates to run?
sequence_length: Amount of sequence to simulate using slendr (in numbers of basepairs)
recombination_rate: Recombination rate to use for the simulation
mutation_rate: Mutation rate to use for the simulation
file: If not NULL, a path where to save the data frame with simulated grid results
packages: A character vector with package names used by user-defined summary statistic functions. Only relevant when parallelization is set up using future::plan() to make sure that the parallelized tree-sequence summary statistic functions have all of their packages available.
attempts: Maximum number of attempts to generate prior values for a valid demographic model (default is 1000)
engine: Which simulation engine to use? Values "msprime" and "slim" will use one of the built-in slendr simulation back ends. Which engine will be used is determined by the nature of the model. If engine = NULL, then spatial slendr models will by default use the "slim" back end, non-spatial models will use the "msprime" back end, and custom user-defined model scripts will use the "custom" engine. Setting this argument explicitly will change the back ends (where appropriate). Setting this argument for custom simulation script has no effect.
model_args: Optional (non-prior) arguments for the slendr model generating function. Setting this argument for custom simulation script has no effect.
engine_args: Optional arguments for the slendr simulation back end. Setting this argument for custom simulation script has no effect.

Value

A list object of the class demografr_abc_sims containing the results of ABC simulations, sampled parameters, priors, and tree-sequence summary statistics