Skip to contents

Constrained maximum-likelihood estimation of the b0, b1 and phi parameters from incidence data.

Source: R/calibration_functions.R
RSVsim_max_likelihood_dust.Rd

Wrapper function for the stats::nlminb function (non-linear box-constrained quasi-Newton optimisation). Used to maximise the RSVsim_log_likelihood_dust function with data and parameters formatted by the RSVsim_calibration_parameters function.

Usage

RSVsim_max_likelihood_dust(
  data,
  fitted_parameter_names,
  fixed_parameters,
  times,
  cohort_step_size = 10,
  init_conds = NULL,
  warm_up = NULL,
  scale_parameters = FALSE,
  lower_ll,
  upper_ll
)

Arguments

fitted_parameter_names

Vector of parameter names to exclude from the list because they are fitted. Default: NULL.

times

Simulation times. Default: 0 - 3650 days with intervals of 0.25 days.

cohort_step_size

Time steps to run the model over before adjusting the ages of all cohorts. Default: 10 days.

init_conds

Initial conditions to run the model. List. Default: NULL. If NULL: 1% RSV prevalence is assumed for people during the primary infection. All other people are assumed to be susceptible to their primary infection.

warm_up

Length of time-points to exclude before calculating the likelihood. Default: NULL.

scale_parameters

Boolean operator indicating whether to scale the parameters between 0 and 1 to help fitting multiple parameters on difference scales. Default: FALSE.

lower_ll

Vector of lower and boundaries on the fitted parameters.

upper_ll

Vector of upper boundaries on the fitted parameters.

Value

List of fitted parameter values.