StrategyRanker

Source:

src/ometeotl_core/game/utility.py

Local role: Ranks strategies by evaluating projected terminal nodes with a utility function.

Big-picture role: Game-layer bridge from projected strategy structures to comparable utility outcomes.

Inheritance:

standard class

Constructor:

StrategyRanker(utility_function)

Methods:

evaluate_strategy(strategy, actor, context=None) -> RankedStrategy
rank_strategies(strategies, actor, context=None) -> list[RankedStrategy]

Important behavior:

validates strategy trees before ranking
aggregates branch probabilities per child node
supports directed acyclic strategies with duplicate terminal paths by accumulating terminal path probabilities per node id
supports scalar and multi-criteria utility outputs

Example:

from ometeotl_core.game.utility import StrategyRanker, WeightedSumUtility

utility = WeightedSumUtility("fw", {"score": 1.0})
ranker = StrategyRanker(utility)

# Rank a candidate set of strategies for an actor
ranked = ranker.rank_strategies([strategy_a, strategy_b, strategy_c], actor)
for rs in ranked:
    print(rs.strategy.id, rs.utility_frame.scalar_value)

best = ranked[0]
print("Best strategy:", best.strategy.id)

Ometeotl