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]]
Action 列表
actlist = {UP:1, DOWN:2, LEFT:3, RIGHT:4}
还有奖励函数
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)
该函数在值迭代函数中调用:
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],U[i][j])
delta = max(delta, abs(U1[i][j] - U[i][j]))
if delta <= epsilon*(1 - gamma)/gamma:
return U
我正在更新
U=U1.copy()
在 while 循环中。
问题是,输出如下所示:
0 0 0.0 0.0
0 1 0.0 0.0
0 2 0.0 0.0
0 3 1.0 1.0
1 0 0.0 0.0
1 2 0.0 0.0
1 3 -1.0 -1.0
2 0 0.0 0.0
2 1 0.0 0.0
2 2 0.7000000000000001 0.7000000000000001
2 3 0.9630000000000001 0.9630000000000001
但我没有在 for 循环内更新 U。 U 应该保持不变(这意味着全为零),而 U1 应该只发生变化。为什么在 for 循环中 U 会自动设置为 U1 的值?
最佳答案
U1(和 U)是列表的列表,实际上是对列表的引用的列表。
您正在(浅)复制外部列表,但副本的内容仍然是对相同内部列表的引用。
尝试:
import copy
U = copy.deepcopy(U1)
看看会发生什么。 deepcopy
将正确地递归复制列表。
关于python - 值在 for 循环内自动更新,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/51710667/