createInitialSample: Create Initial Sample

In kerschke/flacco: Feature-Based Landscape Analysis of Continuous and Constrained Optimization Problems

Create Initial Sample

Description

Convenient helper function, which creates an initial sample - either based on random (uniform) sampling or using latin hypercube sampling.

Usage

createInitialSample(n.obs, dim, control)

Arguments

n.obs	[integer(1)] Number of observations.
dim	[integer(1)] Number of dimensions.
control	[list] Control argument. For further information refer to the details.

Details

Per default, this function will produce n.obs observations of size dim in the range from 0 to 1. If you want to create a more specific initial sample, the following control arguments might be helpful:

• init_sample.type: Should the initial sample be created based on random uniform sampling ("random") or on a latin hypercube sample ("lhs")? The default is "random".

• init_sample.lower: The lower bounds of the initial sample. Either a vector of size dim or a scalar (if all lower bounds are identical). The default is 0.

• init_sample.upper: The upper bounds of the initial sample. Either a vector of size dim or a scalar (if all upper bounds are identical). The default is 1.

Value

[matrix].
A matrix, consisting of n.obs rows of dim-dimensional observations.

Examples

# (1) create a simple initial sample:
X = createInitialSample(300, 5)
summary(X)

# (2) create a more specific initial sample:
ctrl = list(init_sample.type = "lhs",
  init_sample.lower = c(-5, 2, 0),
  init_sample.upper = 10)
X = createInitialSample(200, 3, control = ctrl)
summary(X)

kerschke/flacco documentation built on Dec. 5, 2022, 12:56 a.m.