如何为不同红色射手准确率运行代码？异常结果排查求助

正常运行的代码及输出 #

当单独调用函数时，代码运行正常：

from IPython.core import history
from numpy.random import default_rng
import numpy as np

rng = default_rng(42)

def trial(red,blue , red_accurcy = 1, blue_accurcy = 1, history = False ,debug = False):
    if history:
        red_history = [red]
        blue_history = [blue]
        
    if debug:
        print(f"Start:\t red = {red:3d} blue = {blue:3d}")
    while red and blue:
        p = np.array([red,blue] , dtype = float)
        p /= p.sum()
        arrow = rng.choice(['red' , 'blue'] , p=p)
        if arrow == 'red':
            blue -= 1 if rng.uniform() < red_accurcy else 0
        else:
            red -= 1 if rng.uniform() < blue_accurcy else 0
        if debug:
            print(f"\t red = {red:3d} blue = {blue:3d} arrow = {arrow}")
        
        if history:
            red_history.append(red)
            blue_history.append(blue)
        
    if debug:
        print(f"End:\t red = {red:3d} blue = {blue:3d}") 
    if history:    
        return red,blue, red_history, blue_history
    else:
        return red, blue ,red_accurcy ,blue_accurcy

trial(90,45,0.1,0.2) 

输出：

(60, 0, 0.1, 0.2)

异常场景的代码及输出 #

尝试遍历红色射手不同准确率时，结果异常：

red_ac = np.linspace(0,1,10)
battles = [trial(90,45,red_accurcy=red_ac,blue_accurcy= 1) for red_ac in range(10)]
battles

输出：

[(0, 45, 0, 1),
 (75, 0, 1, 1),
 (82, 0, 2, 1),
 (80, 0, 3, 1),
 (78, 0, 4, 1),
 (74, 0, 5, 1),
 (69, 0, 6, 1),
 (74, 0, 7, 1),
 (81, 0, 8, 1),
 (81, 0, 9, 1)]

错误原因分析 #

1. 循环变量误用：你生成了red_ac = np.linspace(0,1,10)来获取0到1之间的10个浮点准确率，但列表推导式中却用for red_ac in range(10)遍历0-9的整数，完全没用到之前生成的浮点数组。
2. 准确率参数越界：循环中传入red_accurcy的是0-9的整数，而准确率的合理范围是0到1。当red_accurcy > 1时，rng.uniform() < red_accurcy永远为True，红色射手每次射箭必中，导致蓝色射手几乎不可能获胜（仅当red_accurcy=0时红色全不中，蓝色才能赢）。

修正后的代码 #

red_ac = np.linspace(0,1,10)
# 改用遍历生成的浮点数组，避免变量名冲突
battles = [trial(90,45,red_accurcy=rac,blue_accurcy=1) for rac in red_ac]
battles

修正后，传入的red_accurcy会是0、0.111...、0.222...直到1的合理数值，模拟结果会符合预期。

内容的提问来源于stack exchange，提问作者Ahmad Morwat