GenerateInitialSample: Generates the initial sampling of the space of configurations

In fcampelo/MetaTuner: MetaTuner: Metaheuristics Tuner

Description

Generates an initial sample within the space of configurations using a space-covering design. The initial sample is scaled to the interval [0,1] for all parameters

Usage

GenerateInitialSample(m0, dim, method = c("lhs", "sobol"), ndigits = 4, ...)

Arguments

m0	number of points to be generated
dim	dimension for generating the sample (i.e., number of parameters being tuned)
method	type of method to be used in the generation of the sample (see section Methods for details).
ndigits	number of decimal places to use for each parameter.
...	further parameters to be passed down to the specific methods (see section Methods for details).

Value

a list vector where each object is a configuration list, containing the following fields: - config, a named list containing parameter values - Yij, a data frame with two columns: instance.ID and y (the performance value of the configuration on the instance). This data frame is generated empty. - perf, a numeric scalar containing the summary performance value of the configuration

Methods

Currently there are two methods implemented for the generation of the initial sample:

• Latin hypercube sampling ("lhs"): a space-covering design based on Latin Hypercube Sampling. Needs no additional parameters. Needs no additional parameters.

• Low-discrepancy sequences of points ("sobol"): a space-covering design using Sobol's low-discrepancy sequences of points. Based on the implementation of package SobolSequence. Needs no additional parameters.

Author(s)

Felipe Campelo ([e-mail]), Athila Trindade ([e-mail])

fcampelo/MetaTuner documentation built on May 29, 2019, 9:14 a.m.