Simulate an approximate LNA path using a non-centered parameterization for the log-transformed counting process LNA. Resample the initial path in place, then update with elliptical slice sampling.

propose_lna_approx(
  lna_times,
  lna_draws,
  lna_pars,
  lna_param_inds,
  lna_tcovar_inds,
  init_start,
  param_update_inds,
  stoich_matrix,
  forcing_inds,
  forcing_tcov_inds,
  forcings_out,
  forcing_transfers,
  max_attempts,
  ess_updates,
  ess_warmup,
  lna_bracket_width,
  step_size,
  lna_pointer,
  set_pars_pointer
)

Arguments

lna_times: vector of interval endpoint times
lna_pars: numeric matrix of parameters, constants, and time-varying covariates at each of the lna_times
init_start: index in the parameter vector where the initial compartment volumes start
param_update_inds: logical vector indicating at which of the times the LNA parameters need to be updated.
stoich_matrix: stoichiometry matrix giving the changes to compartments from each reaction
forcing_inds: logical vector of indicating at which times in the time-varying covariance matrix a forcing is applied.
max_attempts: maximum number of tries to restart negative increments if reject_negatives is false.
ess_updates: number of elliptical slice sampling updates between samples.
ess_warmup: number of elliptical slice sampling updates before the first sample.
step_size: initial step size for the ODE solver (adapted internally, but too large of an initial step can lead to failure in stiff systems).
lna_pointer: external pointer to the compiled LNA integration function.
set_pars_pointer: external pointer to the function for setting LNA pars.
forcing_matrix: matrix containing the forcings.

Value

list containing the stochastic perturbations (i.i.d. N(0,1) draws) and the LNA path on its natural scale which is determined by the perturbations.