One-dimensional multifacility location

Place k new (variable) points among m existing (fixed) points on a line so that the weighted interaction costs are minimal – TDA's gloc. Each variable point ends up at one of the fixed points; only the order of the fixed points on the line matters, not their spacing (verified against exhaustive search in the test suite).

Usage

tda_locate_line(
  fixed_weights,
  variable_weights = NULL,
  options = list(),
  dir = tempfile("tda"),
  ...
)

Arguments

fixed_weights

a k x m matrix: interaction weight between each variable point (row) and each fixed point (column, in line order).

variable_weights

optional symmetric k x k matrix of weights between the variable points themselves (default: none).

options

a named list of further TDA options for gloc.

dir

working directory.

...

passed to tda_run. The graph-setup options opt=, gt=, perm= and sc= pass through to the underlying edge load, as in tda_g.

Value

An integer vector of length k: for each variable point, the index of the fixed point it is placed at.

See also

Other graph analysis: plot.tda_graph(), tda_dmet(), tda_g(), tda_g_analyses, tda_graph(), tda_ptree(), tda_ucl()

Examples

# two facilities, four sites; facility 1 mostly serves site 2,
# facility 2 site 3, and the two facilities interact
vf <- rbind(c(8, 1, 1, 2), c(1, 1, 7, 3))
tda_locate_line(vf, variable_weights = rbind(c(0, 5), c(5, 0)))
#> [1] 2 3