Import a run's solution and make it stick

Reads the solution of one run into modOut and saves it, so the scenario keeps it after the session ends. This is the step between a finished solve and a scenario that still has its results tomorrow: solve_scenario() never saves, and read_solution() alone leaves the solution in memory.

Usage

import_solution(
  object,
  run,
  ...,
  restore_solver = TRUE,
  save = TRUE,
  promote = FALSE,
  cleanup = FALSE,
  verbose = TRUE
)

Arguments

object

a scenario object.

run

character, the run to import ("<solve>" or "<variant>/<solve>").

...

passed to read_solution().

restore_solver

logical; take the run's solver identity (name, lang, solver, method, formats) from its solver.csv. Never the command line.

save

logical; write the result with save_scenario(). FALSE returns the scenario with the solution in memory only — which is the situation this function exists to end, so it is on by default.

promote

logical; after importing, make this the scenario's own solution with promote_solution() — copied to <scenario>/modOut/ with its provenance, after which the run tree is scratch. Off by default, since which attempt is final is a choice.

cleanup

logical; remove the run's regenerable solver files afterwards. Off by default.

verbose

logical.

Value

the scenario, invisibly, with modOut populated and run active.

Details

Pick the run with scenario_solutions(), which shows each attempt's status, objective, solver and backend. Nothing is chosen for you.

The solver's output/ is left alone. Importing makes it removable — it is no longer the only copy — but removing it is drop_solver_outputs()'s job, and cleanup = FALSE here means this function never does it.

Why this is a verb rather than two calls. When read_solution() cannot read output/variable_list.csv it returns the scenario unchanged, discarding the run switch. Saving that would write an empty modOut/, which flips the run to "imported" and would let the solution's only copy be cleaned up. This refuses to save a read that produced no variables.

See also

scenario_solutions(), read_solution(), drop_solver_outputs()

Examples

if (FALSE) { # \dontrun{
open_project("~/projects/mymodel")
scen <- getScenario("BASE")

# read a solve that ran elsewhere (a cluster or cloud run) back in
scenario_runs(scen)
scen <- import_solution(scen, run = "glpk")

# `promote = TRUE` makes it the scenario's active solution
scen <- import_solution(scen, run = "highs", promote = TRUE)
} # }