Numba nopython 模式不能接受二维布尔索引

Question

我正在尝试使用 numba 加速代码（目前我正在使用 numba 0.45.1）并遇到布尔索引问题。代码如下：

from numba import njit
import numpy as np

n_max = 1000

n_arr = np.hstack((np.arange(1,3),
                   np.arange(3,n_max, 3)
                   ))

@njit
def func(arr):
    idx =  np.arange(arr[-1]).reshape((-1,1)) < arr -2
    result = np.zeros(idx.shape)
    result[idx] = 10.1
    return result

new_arr = func(n_arr)

一旦我运行代码，我就会收到以下消息

TypingError: Invalid use of Function(<built-in function setitem>) with argument(s) of type(s): (array(float64, 2d, C), array(bool, 2d, C), float64)
 * parameterized
In definition 0:
    All templates rejected with literals.
In definition 1:
    All templates rejected without literals.
In definition 2:
    All templates rejected with literals.
In definition 3:
    All templates rejected without literals.
In definition 4:
    All templates rejected with literals.
In definition 5:
    All templates rejected without literals.
In definition 6:
    All templates rejected with literals.
In definition 7:
    All templates rejected without literals.
In definition 8:
    TypeError: unsupported array index type array(bool, 2d, C) in [array(bool, 2d, C)]
    raised from C:\Users\User\Anaconda3\lib\site-packages\numba\typing\arraydecl.py:71
In definition 9:
    TypeError: unsupported array index type array(bool, 2d, C) in [array(bool, 2d, C)]
    raised from C:\Users\User\Anaconda3\lib\site-packages\numba\typing\arraydecl.py:71
This error is usually caused by passing an argument of a type that is unsupported by the named function.
[1] During: typing of setitem at C:/Users/User/Desktop/all python file/5.5.5/numba index broadcasting2.py (29)

注意最后一行的(29)对应的是第29行，也就是result[idx] = 10.1，这一行我尝试给索引为idx的result赋值，一个2 -D 布尔索引。

我想解释一下，必须在 @njit 中包含该语句 result[idx] = 10.1。尽管我想在 @njit 中排除这条语句，但我做不到，因为这一行恰好位于我正在处理的代码的中间。

如果我坚持要在 @njit 中包含赋值语句 result[idx] = 10.1，究竟需要更改什么才能使其生效？如果可能的话，我希望看到一些代码示例，其中涉及 @njit 中的二维布尔索引，可以是运行.

谢谢

Answer 1

Numba 当前不支持使用二维数组进行花式索引。参见：

https://numba.pydata.org/numba-doc/dev/reference/numpysupported.html#array-access

但是，您可以通过明确地使用 for 循环重写您的函数而不是依赖广播来获得等效的行为：

from numba import njit
import numpy as np

n_max = 1000

n_arr = np.hstack((np.arange(1,3),
                   np.arange(3,n_max, 3)
                   ))

def func(arr):
    idx =  np.arange(arr[-1]).reshape((-1,1)) < arr -2
    result = np.zeros(idx.shape)
    result[idx] = 10.1
    return result

@njit
def func2(arr):
    M = arr[-1]
    N = arr.shape[0]
    result = np.zeros((M, N))
    for i in range(M):
        for j in range(N):
            if i < arr[j] - 2:
                result[i, j] = 10.1

    return result

new_arr = func(n_arr)
new_arr2 = func2(n_arr)
print(np.allclose(new_arr, new_arr2))  # True

在我的机器上，使用您提供的示例输入，func2 比 func 快 3.5 倍。

Numba nopython 模式不能接受二维布尔索引

Numba nopython mode cannot accept 2-D boolean indexing

python

indexing

numpy

python-3.x

numba