Stopping

Every heuristic needs a stop. It is how you say how much work to spend.

stop is required, and it is a predicate over a State snapshot (iteration, evaluations, elapsed, best_value, direction). Use the helpers and combine them with any_of:

from pymetaheuristics.core import (
    any_of, max_evaluations, max_iterations, max_time, target_value)

stop = any_of(max_iterations(1000), max_evaluations(5000),
              max_time(2.0), target_value(300))

Note

target_value compares in the problem's direction, so the target 300 above means "value at least 300" for this maximization problem.

Any Callable[[State], bool] works as a stop. See Extending for one.

Recap

• stop is required: a predicate over a State snapshot.
• Combine max_iterations, max_evaluations, max_time and target_value with any_of.

Next: how operators plug in.