Solver: EAX — run.eax • salesperson

Inexact TSP solvers based on a genetic approach.

# S3 method for eax
run(solver, instance, max.trials = 1L, pop.size = 100L,
  off.size = 30L, cutoff.time = 10, opt.tour.length = NULL,
  seed = as.integer(ceiling(runif(1L) * 2^15)), with.restarts = FALSE,
  with.GPX = FALSE, snapshot.step = 0L, full.matrix = FALSE,
  verbose = FALSE, log.trajectory = TRUE, work.dir = NULL,
  output.files.prefix = NULL, keep.output.files = FALSE,
  init.pop = NULL, ...)

Arguments

solver	[`TSPSolver`] TSPSolver object.
instance	[`character(1)` \| `Network`] Path to TSP instance in TSPlib format or `Network` object.
max.trials	[`integer(1)`] Number of independent runs. At the moment this is fixed to 1.
pop.size	[`integer(1)`] Population size. Default is 100.
off.size	[`integer(1)`] Number of offspring generated in each generation. Default is 30.
cutoff.time	[`numeric(1)`] Maximal running time in seconds. Default is 10.
opt.tour.length	[`integer(1) \| NULL`] Length of an optimal TSP tour. This is only relevant for benchmarking. Keep in mind, that in case of a given optimal tour length most internal stopping criteria of EAX are deacitvated. The algorithm stops if an optimal tour is found or the cutoff time is reached.
seed	[`integer(1)`] Seed for the random numbers generator. Defaults to a random integer between 1 and \(2^15\).
with.restarts	[`logical(1)`] Should EAX restart if a plateau is reached? Default is `FALSE`.
with.GPX	[`logical(1)`] Activate GPX2 crossover? Default is `FALSE`.
snapshot.step	[`integer(1)`] Possibility to log the entire population each `snapshot.step` times. Default is `0`, i.e., do not log at all.
full.matrix	[`logical(1)`] Only relevant if `instance` is a `Network`. If `TRUE`, the network is exported to TSPlib format with explicit edge weight definition. Default is `FALSE`.
verbose	[`logical(1)`] Should the output of the solver be printed? Default is `FALSE`.
log.trajectory	[`logical(1)`] Should the optimization progress be logged? Default is `FALSE`.
work.dir	[`character(1)`] Path to working directory of the solver. Defaults to temporary directory via `tempfile()`.
output.files.prefix	[`character(1)`] Prefix for file names used by solver for storing output. Default is a randomly generated temporary prefix.
keep.output.files	[`logical(1)`] Flag indicating whether output files shall be kept or deleted. Default is `FALSE`, i.e., output files are processed and output is loaded into R object. Afterwards the output files are deleted. This flag is particularly useful in large benchmark studies if `work.dir` and `output.files.prefix` is set.
init.pop	[`list`] List of lists. Each sublist needs to contains three components: n [`integer(1)`] Number of nodes of the TSP problem. tour [`integer(n)`] The actual tour. tour.length [`integer(1)`] Length of the tour. Default is `NULL`, i.e., the initial population is generated randomly and 2-Opt is applied to each solution before the evolutionary loop starts.
...	[any] Not used at the moment.

Value

[TSPSolverResult]

Note

This solver requires integer inter-city distances.

References

Nagata, Y. and Kobayashi, S. (2013). A powerful genetic algorithm using edge assembly crossover for the travelling salesman problem. INFORMS Journal on Computing, 25(2):346-363.

Nagata, Y. and Kobayashi, S. (1997). Edge assembly crossover: A high-power genetic algorithm for the travelling salesman problem. In Baeck, T., editor, Proceedings of the Seventh International Conference on Genetic Algorithms (ICGA97), pages 450-457, San Francisco, CA. Morgan Kaufmann.