dreamCalibrate: Utility to calibrate a function using dream

In dream: DiffeRential Evolution Adaptive Metropolis

Description

Calibrate a function using dream, a specified likelihood function lik.fun and observed values obs

Usage

dreamCalibrate(FUN, pars, obs, lik.fun=calc.rmse,lik.control=NULL,FUN.pars = list(), ...)

Arguments

FUN	model function with first argument a vector of parameter values of length ndim.
pars	a list of variable ranges. Any names will be propagated to output.
obs	a numeric vector of observed values, corresponding to the output of FUN
lik.fun	A function that returns the log likelihood of model predictions matching observed values. log.lik=f(predicted,observed,control). See included functions listed below. Default uses RMSE, as implemented by calc.rmse.
lik.control	A list of any extra arguments to be passed to lik.fun
FUN.pars	A list of any extra arguments to be passed to FUN.
...	Extra arguments to be passed to dream, e.g. control

Details

There are S3 methods for: predict, coef.

Log likelihood functions included are: calc.rmse, calc.weighted.rmse, calc.loglik

Value

An object inheriting from dream, i.e. with the same elements and:

FUN	The function calibrated
FUN.pars	The extra arguments originally passed to that function
lik.fun	A convenience function f(pars) that returns the log likelihood for a parameter set, as would be calculated using the given calibration dataset and chosen likelihood function.

See Also

See dream for details on the calibration method, visualisation of its results and diagnostics.

See snow.chains for parallelised calibration of external models (expert users only).

• demo(FME.nonlinear.model)Calibrating the non-linear model shown in the FME vignette

dream documentation built on May 2, 2019, 5:21 p.m.