Python 值错误：要解压缩的值太多（应为3个）

我试图实现值迭代算法。 我有一个网格

grid = [[0, 0, 0, +1],
    [0, "W", 0, -1],
    [0, 0, 0, 0]]

行动清单

actlist = {UP:1, DOWN:2, LEFT:3, RIGHT:4}

和一个奖励函数

reward = [[0, 0, 0, 0],
      [0, 0, 0, 0],
      [0, 0, 0,  0]] 

我编写了一个函数T，它返回3个元组的元组

def T(i,j,actions):
if(i == 0 and j == 0):
    if(actions == UP):
        return (i,i,0.8),(i,i,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i+1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j,0.8),(i,j,0.1),(i+1,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i,i,0.1),(i+1,j,0.1)
elif (i == 0 and j == 1):
    if(actions == UP):
        return (i,i,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i,j,0.1),(i,j,0.1)
elif(i == 0 and j == 2):
    if(actions == UP):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return(i+1,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i,j,0.1),(i+1,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i,j,0.1),(i+1,j,0.1)
elif(i == 0 and j == 3):
    if(actions == UP):
        return (-1,-1,0.8),(-1,-1,0.1),(-1,-1,0.1)
    elif(actions == DOWN):
        return (-1,-1,0.8),(-1,-1,0.1),(-1,-1,0.1)
    elif(actions == LEFT):
        return (-1,-1,0.8),(-1,-1,0.1),(-1,-1,0.1)
    elif(actions == RIGHT):
        return (-1,-1,0.8),(-1,-1,0.1),(-1,-1,0.1)
# 2nd row
elif (i == 1 and j == 0):
    if(actions == UP):
        return (i-1,j,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == DOWN):
        return (i+1,j,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == LEFT):
        return (i,j,0.8),(i-1,j,0.1),(i+1,j,0.1)
    elif(actions == RIGHT):
        return (i,j,0.8),(i-1,j,0.1),(i+1,j,0.1)
elif(i == 1 and j ==1):
    if(actions == UP):
        return (i,j,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == LEFT):
        return (i,j,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j,0.8),(i,j,0.1),(i,j,0.1)
elif (i == 1 and j == 2):
    if(actions == UP):
        return (i-1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i+1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j,0.8),(i-1,j,0.1),(i+1,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i-1,j,0.1),(i+1,j,0.1)
elif(i == 1 and j == 3):
    if(actions == UP):
        return (-2,-2,0.8),(-2,-2,0.1),(-2,-2,0.1)
    elif(actions == DOWN):
        return (-2,-2,0.8),(-2,-2,0.1),(-2,-2,0.1)
    elif(actions == LEFT):
        return (-2,-2,0.8),(-2,-2,0.1),(-2,-2,0.1)
    elif(actions == RIGHT):
        return (-2,-2,0.8),(-2,-2,0.1),(-2,-2,0.1)      
# 3rd row
elif(i == 2 and j == 0):
    if(actions == UP):
        return (i-1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j,0.1),(i,j+1,1,0.1)
    elif(actions == LEFT):
        return (i,j,0.8),(i-1,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i-1,j,0.1),(i,j,0.1)
elif (i == 2 and j == 1):
    if(actions == UP):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i,j,0.1),(i,j,0.1)
elif(i == 2 and j == 2):
    if(actions == UP):
        return (i-1,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i-1,j,0.1),(i,j,1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i-1,j,0.1),(i,j,0.1)
elif(i == 2 and j == 3):
    if(actions == UP):
        return (i-1,j,0.8),(i,j-1,0.1),(i,j,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i-1,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j,0.8),(i-1,j,0.1),(i,j,0.1) 

在值迭代函数中调用此函数：

def value_iteration():
U1 = [[0, 0, 0, 0],
      [0, 0, 0, 0],
      [0, 0, 0, 0]] 
while True:
    U=U1.copy()
    delta = 0
    for i in range(len(grid)):
        for j in range(len(grid[i])):
            U1[i][j] = max(sum(p*(R(k,l)+gamma*U[k][l]) for (k,l,p) in T(i,j,a)) for a in actlist)
            print(i,j,U1[i][j])
            delta = max(delta, abs(U1[i][j] - U[i][j]))
    if delta <= epsilon*(1 - gamma)/gamma:
        return U

但随后代码停止并出现错误

ValueError: too many values to unpack (expected 3)

---

看看****中的元组，也许这就是原因

# 3rd row
elif(i == 2 and j == 0):
    if(actions == UP):
        return (i-1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j,0.1),**(i,j+1,1,0.1)**
    elif(actions == LEFT):
        return (i,j,0.8),(i-1,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i-1,j,0.1),(i,j,0.1)
elif (i == 2 and j == 1):
    if(actions == UP):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
       return (i,j+1,0.8),(i,j,0.1),(i,j,0.1)

---

看看****中的元组，也许这就是原因

# 3rd row
elif(i == 2 and j == 0):
    if(actions == UP):
        return (i-1,j,0.8),(i,j,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j,0.1),**(i,j+1,1,0.1)**
    elif(actions == LEFT):
        return (i,j,0.8),(i-1,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
        return (i,j+1,0.8),(i-1,j,0.1),(i,j,0.1)
elif (i == 2 and j == 1):
    if(actions == UP):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == DOWN):
        return (i,j,0.8),(i,j-1,0.1),(i,j+1,0.1)
    elif(actions == LEFT):
        return (i,j-1,0.8),(i,j,0.1),(i,j,0.1)
    elif(actions == RIGHT):
       return (i,j+1,0.8),(i,j,0.1),(i,j,0.1)

---

正如@EdwardMinnix所提到的，您应该使用map，并且永远不要使用这样的if/else构造。或者，如果有任何与这些值相关的模式，请查看

这将对您有所帮助：

...

VALUE_A = 0.8
VALUE_B = 0.1


def new_T(i, j, actions):
    result_map = {(0, 0, 1): ((i, i, VALUE_A), (i, i, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 0, 2): ((i + 1, j, VALUE_A), (i, j, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 0, 3): ((i, j, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 0, 4): ((i, j + 1, VALUE_A), (i, i, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 1, 1): ((i, i, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 1, 2): ((i, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 1, 3): ((i, j - 1, VALUE_A), (i, j, VALUE_B), (i, j, VALUE_B)),
              (0, 1, 4): ((i, j + 1, VALUE_A), (i, j, VALUE_B), (i, j, VALUE_B)),
              (0, 2, 1): ((i, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 2, 2): ((i + 1, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 2, 3): ((i, j - 1, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 2, 4): ((i, j + 1, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B))}

    return result_map.get((i, j, actions))


for i, j, action in itertools.product(range(4), range(4), range(1, 5)):
    print('%s %s %s' % (i, j, action))
    T_value = T(i, j, action)
    new_T_value = new_T(i, j, action)
    if T_value != new_T_value:
        raise AssertionError('Error! \nT:     %s \nNew T: %s' % (T_value, new_T_value))

---

正如@EdwardMinnix所提到的，您应该使用map，并且永远不要使用这样的if/else构造。或者，如果有任何与这些值相关的模式，请查看

这将对您有所帮助：

...

VALUE_A = 0.8
VALUE_B = 0.1


def new_T(i, j, actions):
    result_map = {(0, 0, 1): ((i, i, VALUE_A), (i, i, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 0, 2): ((i + 1, j, VALUE_A), (i, j, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 0, 3): ((i, j, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 0, 4): ((i, j + 1, VALUE_A), (i, i, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 1, 1): ((i, i, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 1, 2): ((i, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 1, 3): ((i, j - 1, VALUE_A), (i, j, VALUE_B), (i, j, VALUE_B)),
              (0, 1, 4): ((i, j + 1, VALUE_A), (i, j, VALUE_B), (i, j, VALUE_B)),
              (0, 2, 1): ((i, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 2, 2): ((i + 1, j, VALUE_A), (i, j - 1, VALUE_B), (i, j + 1, VALUE_B)),
              (0, 2, 3): ((i, j - 1, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B)),
              (0, 2, 4): ((i, j + 1, VALUE_A), (i, j, VALUE_B), (i + 1, j, VALUE_B))}

    return result_map.get((i, j, actions))


for i, j, action in itertools.product(range(4), range(4), range(1, 5)):
    print('%s %s %s' % (i, j, action))
    T_value = T(i, j, action)
    new_T_value = new_T(i, j, action)
    if T_value != new_T_value:
        raise AssertionError('Error! \nT:     %s \nNew T: %s' % (T_value, new_T_value))

---

你能提供完整的错误信息以便我们知道错误被触发的确切位置吗？

U1[i][j]=max（sum（p*（R（k，l）+gamma*U[k][l]）for（k，l，p）in T（i，j，a））for a in actlist）

错误是在这一行触发的。附带问题：如果你所有的函数都像一个表一样，为什么不使用字典呢？@Edwardminix，我不知道该怎么做。@ShifatEArman您只需制作一个字典，其中键是三元素元组，其值是

（I，j，action）

，其值是您将从函数返回的结果（虽然它们需要硬编码。如果你不熟悉字典，我建议你参考一下。你能提供完整的错误信息吗？这样我们就知道触发错误的确切位置了吗？

U1[I][j]=max（对于actlist中的a，T（I，j，a）中的（k，l，p）的和（p*（R（k，l）+gamma*U[k][l]））

这一行触发了错误。附带问题：如果你的函数只是像一个表一样工作，为什么不使用字典呢？@EdwardMinnix，我不知道怎么做。@ShifatEArman你只需要制作一个字典，其中键是三元素元组，值是

（I，j，action）

，其值是您将从函数返回的结果（尽管它们需要硬编码。如果您不熟悉字典，我建议您引用愚蠢的错误。非常感谢。：）当然。另外，我正在更新

U=U1.copy（）

在两个for循环都完成后。但是它在我的for循环中自动更新。我不知道原因。不确定这能解决您的问题，但每次使用该功能时，您都会“重置”U和U1，并返回

U

，这是一个仅为零的错误。非常感谢。）当然。还有一件事，在两个for循环完成后，我正在更新

U=U1.copy（）

。但是它在我的for循环中自动更新。我不知道原因。我不确定这是否解决了您的问题，但每次使用函数时，您都会“重置”U和

U1

，并返回

U

，仅为零