Package-level declarations
Types
Options governing concurrent solver execution (parallel random restarts, solver portfolios).
The engine for concurrent solver execution: runs the member solvers of a concurrent run (parallel random restarts, a solver portfolio) on a bounded dispatcher, each with its own factory-provisioned evaluation resources, and hands their results back in deterministic member order.
Options for the optional confirmation stage run after all members of a concurrent solver run complete. Members can return "best" solutions estimated with different precision (different replication counts, different luck), so picking the winner from point estimates alone favors noise. The confirmation stage re-evaluates the top candidates with common random numbers and picks the winner from the confirmed estimates — standard ranking-and-selection hygiene.
Provisions the per-member evaluation resources for concurrent solver execution.
The lifecycle outcome of one member of a concurrent solver run.
The standard member-evaluator provisioner: each member gets its own sequential SimulationProvider over a pooled, silenced model, with a private solution cache and a stream tape starting at the member's block offset.
The confirmation stage for concurrent solver runs: re-evaluates the top candidate solutions under common random numbers and picks the winner from the confirmed estimates. Members of a concurrent run can return bests with different statistical precision, so a winner picked from raw point estimates favors noise; a short CRN comparison of the finalists is the standard ranking-and-selection remedy.
Creates a fresh solver instance bound to the supplied per-member evaluation resources.
The result of one member of a concurrent solver run, captured on the member's worker after its solver finished (or failed). Results are safe to read once obtained from the runner: the await establishes the necessary happens-before ordering.
One member of a concurrent solver run: the factory that creates its solver instance plus the member-specific configuration applied at creation time.
A solver portfolio: N solver instances — typically of different algorithms — race on the same problem concurrently, each on its own worker with its own factory-created solver and private evaluation resources, and the portfolio reports the best solution found across all members.