Load parameters for fitting.
spim_fit_pars_load(path, region, assumptions, kernel_scaling)The path to load from. This will be a directory containing files "info.csv", "prior.csv" and "proposal.csv" (for spim_pars_pmcmc_load) but also the baseline parameters ("base.rds") and a transformation function ("transform.R")
A single region, or vector of regions, to load.
The name of any assumptions to apply to filter the baseline parmeters (e.g., "central")
The scaling coefficient for loading the proposal variance-covariance matrix.
A list.
There are lots of types of parameters here, depending on the perspective
This function deals with loading in the "logical" parameters; things like "vaccine efficacy" or "vaccination doses". These can be any R object.
Our aim is to create sircovid parameters, the result of
sircovid::lancelot_parameters. This will be a list with
structured parmeters that correspond directly to dust/odin
inputs. Most likely these will be
mcstate::multistage_parameters objects that cover multiple
epochs with different model parameters, and these might be
grouped together in a list so that a multiparameter (nested)
filter can be run.
We'll end up working with small(ish) numeric vectors of parameters from the pmcmc process. Our transform function will convert this vector of parameters, using the logical parameters, into baseline parameters.
The mcmc parameters object that this function will return (as
$mcmc) is a mcstate::pmcmc_parameters or
mcstate::pmcmc_parameters_nested object, which contains all
the information for the above.