inactivation_OED_penalty: Optimum Experimental Design of Microbial Inactivation with...

In bioOED: Sensitivity Analysis and Optimum Experiment Design for Microbial Inactivation

Description

Performs an optimum experimental design for the settings selected including a function which penalties points too close. The OED is based on the FIM, estimated using the local sensitivity functions provided by sensitivity_inactivation.

Usage

inactivation_OED_penalty(inactivation_model, parms, temp_profile,
  parms_fix, n_points, time_min, criteria = "D", n_times = 100,
  sensvar = "logN", optim_algorithm = "global", opts_global = NULL,
  ...)

Arguments

inactivation_model	Character string defining the inacivation model.
parms	Named numeric vector defining the model parameters. They must be named according to the needs of predict_inactivation.
temp_profile	Data frame defining the temperature profile. It must contain a column named time and a column named temperature.
parms_fix	Named numeric vector defining the model parameters to be omitted during the calculation of the local sensitivities.
n_points	Number of measurements which will be taken during the experiment.
time_min	Numeric value indicating the minimum space between measurements.
criteria	Character defining the criteria for the OED. Either D (default) or E-mod.
n_times	Integer defining th enumber of discrete time points used for the interpolation of the local sensitivities.
sensvar	Character defining the variable to use for the OED. Either logN (default) or N.
optim_algorithm	Character defining the type of algorithm to use for the optimization. Either global (default) or local.
opts_global	List defining the options for the global optimization algorithm (see MEIGO). By default, global solver with a maximum of 50000 function evaluations and printout on every step.
...	Additional arguments passed to penalty_function.

Value

A list of class OEDinactivation with the following items:

• optim: Objetc returned by the optimization function.

• model: Inactivation model used for the calculations.

• parms: Nominal model parameters.

• parms_fix: Model parameters not considered for the sensitivity calculation.

• criteria: Criteria used for the OED.

• sensvar: Variable used for the OED.

• optim_algorithm: Type of optimization algorithm.

• optim_times: Optimum measurement times calculated.

• penalty: Logical indicating whether penalty function was used.

• temp_profile: Temperature profile of the experiment.

Examples

## Definition of input variables

parms_fix <- c(temp_ref = 57.5)
parms <- c(delta_ref = 3.9,
           z = 4.2,
           p = 1,
           N0 = 1e6
)

temp_profile <- data.frame(time = c(0, 60), temperature = c(30, 60))

n_points <- 5
time_min <- 10

## Not run: 

## OED with local optimization

set.seed(0123182)

local_OED <- inactivation_OED_penalty("Mafart", parms, temp_profile, parms_fix,
                              n_points, criteria = "E-mod", sensvar = "logN",
                              optim_algorithm = "local", time_min = time_min)

print(local_OED$optim_times)
plot(local_OED)

## OED with global optimization

opts_global <- list(maxeval=500,  local_solver=0,
                    local_finish="DHC", local_iterprint=1)

global_OED <- inactivation_OED_penalty("Mafart", parms, temp_profile, parms_fix,
                               n_points, criteria = "E-mod", opts_global = opts_global,
                               time_min = time_min)

print(global_OED$optim_times)
plot(global_OED)


## End(Not run)

bioOED documentation built on Aug. 7, 2019, 5:03 p.m.