What solutions does this scenario have, and which is worth importing?

One row per run, joining what the run record holds with what can be read back from a run's raw output/ without importing it. Read-only: it writes nothing, not even the objectives it recovers.

Usage

scenario_solutions(object, peek = TRUE)

Arguments

object: a scenario object.
peek: logical; FALSE skips reading output/ and leaves solver_status / peeked objectives as NA. Faster on a scenario with many large runs.

Value

a tibble: run, variant, solve, status, solver_status, objective, objective_src, solver_name, lang, solver, duration_sec, imported, has_output, output_mb, active.

Details

run.yml's objective is computed from the in-memory solution when the solve finishes, so it is NA for any run whose solution was never imported — which is exactly the set you need to compare after several attempts. This recovers it per run from the single vObjective file in output/, and objective_src says which reading you are looking at: "record" from run.yml, "output" peeked from the dump, NA when neither is available.

solver_status likewise comes from the run's output/log.csv and is the solver's own verdict, independent of status, which is what the driver recorded. The two disagree exactly when a solve finished but was never read back.

Nothing here ranks the runs or picks one. Objectives are comparable only within one problem: a "<variant>/<solve>" run solves a different problem (its own modInp), so compare within a variant, not across.

Examples

if (FALSE) { # \dontrun{
scenario_solutions(scen)                      # what is there
import_solution(scen, "julia_highs_barrier")  # make one of them stick
} # }