OpenSpiel state methods: resample_from_infostate
resample_from_infostate(player_id: int, rng: callable)
Resamples a new game history that is consistent with the specified player’s
information state. The resampled state keeps player_id’s private information
and all public information unchanged, but resamples the private information of
other players. Chance outcomes are sampled uniformly from consistent
possibilities.
The rng parameter should be a callable that returns a float in [0, 1), used
to sample from chance actions.
For perfect information games, this returns a clone of the current state. For imperfect information games, the game must implement this method.
Examples:
import pyspiel
import random
game = pyspiel.load_game("kuhn_poker")
state = game.new_initial_state()
state.apply_action(0) # Deal card 0 to player 0
state.apply_action(1) # Deal card 1 to player 1
# Resample from player 0's perspective
resampled = state.resample_from_infostate(
0, lambda: random.random())
# Player 0 still has card 0, but player 1's card may differ