latin_hypercube_input function

latin_hypercube_input draws a maximin Latin Hypercube sample of n_spectra parameter sets over the parameter ranges given in tab, and writes them to a CSV file (lh_ts.csv) for use as SCOPE simulation input (e.g. to generate a synthetic training/verification dataset). If LIDFa and LIDFb are among the sampled variables, the sampled pair is transformed so that abs(LIDFa + LIDFb) <= 1 is respected.

Usage

latin_hypercube_input(
  tab = read.table(file.path("input", "dataset for_verification", "input_borders.csv"),
    header = TRUE),
  n_spectra = 30,
  outdir = file.path("input", "dataset for_verification")
)

Arguments

tab

data.frame. Parameter bounds table with columns include (logical, whether the variable is sampled), lower/upper (parameter bounds), and variable (parameter name). Defaults to reading input/dataset for_verification/input_borders.csv.

n_spectra

integer. Number of parameter sets (LHS samples) to draw. Default 30.

outdir

character. Output directory in which lh_ts.csv is written (and, if tab is missing, from which input_borders.csv is read). Default input/dataset for_verification.

Value

No return value; called for its side effect of writing the sampled parameter table to file.path(outdir, 'lh_ts.csv') and printing a summary message. Errors if that file already exists.

Author

Christiaan van der Tol(Original version in Matlab)

Carlos Camino (Ported version into R)

Examples

if (FALSE) { # \dontrun{
latin_hypercube_input(tab, n_spectra = 30, outdir = "input/dataset for_verification")
} # }