decision_pt += qs[ opt_qid_pair[1] ]
if ran_pt <= decision_pt:
return (opt_qid_pair[0], opt_qid_pair[1])
- return (None, None)
Q = np.ones([states_dim, actions_dim])
g.dice()
state_qid = find_state_qid(g.get_shutable(), g.get_diced())
while not g.is_over():
- opt, opt_qid = select_option( g.get_options(), Q[state_qid, :] )
- if opt:
+ options = g.get_options()
+ if len(options) > 0:
+ opt, opt_qid = select_option( options, Q[state_qid, :] )
old_score = g.get_score()
g.shut(opt)
g.dice()