OpenSpiel state methods: distribution_support

Back to Core API reference

distribution_support()

Returns the support of the state distribution that needs to be updated at the current mean field node. States in the support are identified by their string representations. This should only be called when current_player() returns PlayerId.MEAN_FIELD. Can return an empty list if the distribution is not needed at this point.

This method is specific to mean field games.

Examples:

import pyspiel

game = pyspiel.load_game("mfg_crowd_modelling_2d")
state = game.new_initial_state()

# Advance to a mean field node
while not state.is_mean_field_node():
    if state.is_chance_node():
        action = state.chance_outcomes()[0][0]
        state.apply_action(action)
    else:
        state.apply_action(state.legal_actions()[0])

support = state.distribution_support()
print(type(support))    # Output: <class 'list'>