时间戳

首页 > 解决方案 > 将网格网格作为参数传递，将数组作为另一个参数

python - 将网格网格作为参数传递，将数组作为另一个参数

问题描述

我似乎记得看到可以将 numpy 网格网格作为参数与数组一起传递给函数，并让函数在迭代网格网格的元素时保持数组的完整性。我梦到这个还是有人知道我指的是什么？我搜索了 SOF、google 和 numpy 文档，但没有成功。

例子：

def makeGridandArray(i=60, j=40, ratio=10):
    arr = np.arange(i*j).reshape(i, j)
    x,y = np.meshgrid(np.arange(0, i, ratio), np.arange(0, j, ratio), sparse=True)
    return crop(x, y, ratio, arr)

def crop(x, y, ratio,  Arr):
    return np.average(Arr[x:x+ratio, y:y+ratio])

上面的代码将抛出一个无效切片错误，因为 Arr 正在与网格网格一起迭代。

>>> x=makeGridandArray()
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/private/var/mobile/Containers/Shared/AppGroup/A181F23F-E674-4B44-882C-B03A93E0E84A/Pythonista3/Documents/so.py", line 752, in makeGridandArray
    return crop(x,y, ratio, arr)
  File "/private/var/mobile/Containers/Shared/AppGroup/A181F23F-E674-4B44-882C-B03A93E0E84A/Pythonista3/Documents/so.py", line 755, in crop
    return np.average(Arr[x:x+ratio, y:y+ratio])
IndexError: invalid slice

在上面的示例中，我希望返回一个 6x4 数组，其中包含 Arr 的每个 10x10 子集的平均值。我意识到这可以通过 np.average 和重塑来实现，但我正在寻找的是如何将 Arr 整体传递给函数的语法，同时利用将网格网格传递给函数。谢谢！

标签： pythonnumpy

解决方案

meshgrid is just a way of generating two arrays that can index and broadcast together. For example:

In [302]: x,y=np.meshgrid([1,2,3],[4,5,6],sparse=True)                                                 
In [303]: x                                                                                            
Out[303]: array([[1, 2, 3]])
In [304]: y                                                                                            
Out[304]: 
array([[4],
       [5],
       [6]])

One is (1,3) shape, the other (3,1); together the produce a (3,3) result

With addition (or other binary math):

In [305]: x+y                                                                                          
Out[305]: 
array([[5, 6, 7],
       [6, 7, 8],
       [7, 8, 9]])

As indices in a 2d array:

In [306]: arr = np.arange(1,101).reshape(10,10)                                                        
In [307]: arr[x,y]                                                                                     
Out[307]: 
array([[15, 25, 35],
       [16, 26, 36],
       [17, 27, 37]])

An array cannot be used as a slice endpoint:

In [310]: np.arange(10)[x: x+3]                                                                        
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-310-adcd5e98b614> in <module>
----> 1 np.arange(10)[x: x+3]

TypeError: only integer scalar arrays can be converted to a scalar index

Just because an array is generated by meshgrid, it doesn't mean it has special properties, or be used in ways that any other array can't.

推荐阅读