continue_pmcmc: Run more iterations of PMCMC
In stewid/SimInf: A Framework for Data-Driven Stochastic Disease Spread Simulations

continue_pmcmc

R Documentation

Run more iterations of PMCMC

Description

Run more iterations of PMCMC

Usage

continue_pmcmc(
  object,
  obs_process,
  n_iterations,
  post_proposal = NULL,
  init_model = NULL,
  post_particle = NULL,
  verbose = getOption("verbose", FALSE)
)

## S4 method for signature 'SimInf_pmcmc'
continue_pmcmc(
  object,
  obs_process,
  n_iterations,
  post_proposal = NULL,
  init_model = NULL,
  post_particle = NULL,
  verbose = getOption("verbose", FALSE)
)

Arguments

`object`	The `SimInf_pmcmc` object to continue from.
`obs_process`	Specification of the stochastic observation process. The `obs_process` can be specified as a `formula` if the model contains only one node and there is only one data point for each `time` in `data`. The left hand side of the formula must match a column name in the `data` data.frame and the right hand side of the formula is a character specifying the distribution of the observation process, for example, `Iobs ~ poisson(I)`. The following distributions are supported: `x ~ binomial(size, prob)`, `x ~ poisson(rate)` and `x ~ uniform(min, max)`. The observation process can also be a function to evaluate the probability density of the observations given the simulated states. The first argument passed to the `obs_process` function is the result from a run of the model and it contains one trajectory with simulated data for a time-point. The second argument to the `obs_process` function is a `data.frame` containing the rows for the specific time-point that the function is called for. Note that the function must return the log of the density.
`n_iterations`	An integer specifying the number of iterations to run the PMCMC.
`post_proposal`	An optional function that, if non-`NULL`, is applied on the model after the proposal has been set for the model, but before running the particle filter. The function must accept one argument of type `SimInf_model` with the current model of the fitting process. This function can be useful to update, for example, `ldata` of the model before running a trajectory with proposed parameters. The function must return the model object which is then used in the particle filter.
`init_model`	An optional function that, if non-NULL, is applied in the particle filter before running each proposal. The function must accept one argument of type `SimInf_model` with the current model of the fitting process. This function can be useful to specify the initial state of `u0` or `v0` of the model before running a trajectory with proposed parameters.
`post_particle`	An optional function that, if non-NULL, is applied after each completed particle. The function must accept three arguments: 1) an object of `SimInf_pmcmc` with the current state of the fitting process, 2) an object `SimInf_pfilter` with the last particle and one filtered trajectory attached, and 3) an integer with the iteration in the fitting process. This function can be useful to, for example, monitor, save and inspect intermediate results. Note that the second `SimInf_pfilter` argument, is non-NULL only for the first particle in the chain, and for accepted particles.
`verbose`	prints diagnostic messages when `TRUE`. The default is to retrieve the global option `verbose` and use `FALSE` if it is not set. When `verbose=TRUE`, information is printed every 100 iterations. For pmcmc, it is possible to get information every nth information by specifying `verbose=n`, for example, `verbose=1` or `verbose=10`.