ch7.py 文件源码

def run_single_regret(bandit_list,bandit_params,plays):
    sum_probs_chosen=0
    opt=np.zeros(plays)
    chosen=np.zeros(plays)
    bandit_probs = [x.get_prob() for x in bandit_list]
    opt_solution = max(bandit_probs)
    for i in range(0,plays):
        index = sample_distributions_and_choose(bandit_params)
        sum_probs_chosen+=bandit_probs[index]
        if(bandit_list[index].pull_handle()):
            bandit_params[index]=\
        (bandit_params[index][0]+1,bandit_params[index][1])
        else:
            bandit_params[index]=\
        (bandit_params[index][0],bandit_params[index][1]+1)
        opt[i] = (i+1)*opt_solution
        chosen[i] = sum_probs_chosen
    regret_total = map(sub,opt,chosen)
    return regret_total

#7.9
#Plot params beforehand