pcode_lkh {pCODE}

pcode_lkh (likelihood and multiple dimension version)

Description

Obtain estimates of both structural and nuisance parameters of an ODE model by parameter cascade method.

Usage

pcode_lkh(data, likelihood.fun, time, ode.model, par.names,
                 state.names, par.initial, basis.list, lambda, controls)

Arguments

data

A data frame or a matrix contain observations from each dimension of the ODE model.

likelihood.fun

A function computes the likelihood or the loglikelihood of the errors.

time

A vector contains observation ties or a matrix if time points are different between dimesion.

ode.model

An R function that computes the time derivative of the ODE model given observations of states variable and structural parameters.

par.names

The names of structural parameters defined in the 'ode.model'.

state.names

The names of state variables defined in the 'ode.model'.

par.initial

Initial value of structural parameters to be optimized.

basis.list

A list of basis objects for smoothing each dimension's observations. Can be the same or different across dimensions.

lambda

Penalty parameter.

controls

A list of control parameters. See ‘Details’.

Details

The controls argument is a list providing addition inputs for the nonlinear least square optimizer:

• nquadpts Determine the number of quadrature points for approximating an integral. Default is 101.

• smooth.lambda Determine the smoothness penalty for obtaining initial value of nuisance parameters.

• tau Initial value of Marquardt parameter. Small values indicate good initial values for structural parameters.

• tolx Tolerance for parameters of objective functions. Default is set at 1e-6.

• tolg Tolerance for the gradient of parameters of objective functions. Default is set at 1e-6.

• maxeval The maximum number of evaluation of the optimizer. Default is set at 20.

Value

structural.par

The structural parameters of the ODE model.

nuisance.par

The nuisance parameters or the basis coefficients for interpolating observations.