fitode {fitode} | R Documentation |
Fit ordinary differential equations model
Description
This function fits ordinary differential equations models to a uni- or multi-variate time series by maximum likelihood. It relies on sensitivity equations to compute gradients of the estimated trajectory with respect to model parameters. This allows one to use gradient-based optimization algorithms, which can provide more robust estimation.
Usage
fitode(
model,
data,
start,
tcol = "times",
method = "BFGS",
optimizer = "optim",
link,
fixed = list(),
prior = list(),
prior.density = TRUE,
control = list(maxit = 1e+05),
solver.opts = list(method = "rk4"),
solver = ode,
skip.hessian = FALSE,
force.hessian = FALSE,
use.ginv = TRUE,
quietly = FALSE,
...
)
Arguments
model |
odemodel object |
data |
data frame with a time column and observation columns |
start |
named vector of starting parameter values |
tcol |
(character) time column |
method |
optimization method |
optimizer |
optimizer |
link |
named vector or list of link functions for model parameters |
fixed |
named vector or list of model parameters to fix and their values |
prior |
list of formulas specifying prior distributions |
prior.density |
(logical) should priors represent probability distributions? |
control |
see |
solver.opts |
options for ode integration. See |
solver |
ode solver |
skip.hessian |
skip hessian calculation |
force.hessian |
(logical) calculate the hessian numerically instead of taking the jacobian of the gradients based on sensitivity equations |
use.ginv |
(logical) use generalized inverse ( |
quietly |
suppress progress messages? |
... |
mle2 arguments |
Value
An object of class “fitode” as described in fitode-class
.