Initial sample type
Three different options are available for generation of the initial sample in the SCE algorithm:
· Monte Carlo sampling. In this case the initial parameter sets are randomly generated within the feasible parameter range specified on the Model Parameters page assuming a uniform distribution.
· Latin hypercube sampling. In this case the individual parameters are sampled according to a stratified sampling scheme where the feasible parameter interval is divided into s equal intervals (s being the sample size) and a point is then randomly selected within each interval.
· Initial sample from previous optimisation run. This option allows continuing the optimisation from the last iteration loop of a previous optimisation run.
Random seed
Random seed used in the optimisation. Can be set to any positive integer value. Since the SCE method is a probabilistic search procedure, different optimisation results will be obtained by using different random seeds.
Use initial parameter values
Option to specify if the initial parameter values specified on the Model Parameters page should be included in the initial sample.
File name
File name of the file containing the optimisation results from a previous optimisation run to be used as initial conditions. The file to be specified is the AutoCal SCE optimisation output file (see Section Optimisation output).