如何用广播替换我的嵌套for循环?

huangapple go评论67阅读模式
英文:

How do I replace my nested for loop with broadcasting?

问题

可以使用广播(broadcasting)来替代嵌套的for循环,以实现不同方式的矩阵元素交互。希望避免嵌套for循环的话,可以使用NumPy库来进行广播操作。

以下是使用广播重写的代码:

import numpy as np

values = np.array([[33, 23, 0.6, 0.8, 1.5], [34, 25, 0.9, 0.7, 1.6]])
points = np.array([[0.2, 19.8], [0.6, 19.8]])

# 使用广播计算差值
pa_x = values[:, 0] - points[:, 0][:, np.newaxis]
pa_y = values[:, 1] - points[:, 1][:, np.newaxis]

这个代码片段使用NumPy数组来执行广播操作,将valuespoints两个不同大小的矩阵的元素进行不同方式的交互,同时避免了嵌套的for循环。

英文:

I have two different matrices of unequal sizes and I want each element in one matrice to interact with the second matrice differently. Can I replace the nested for-loops with broadcasting? I want to ideally avoid a nested for loop.

values=[[33,23,0.6,0.8,1.5],[34,25,0.9,0.7,1.6]]
points=[[0.2,19.8],[0.6,19.8]]

for i in points:
   for j in values:
      pa_x = j[0]-i[0]
      pa_y = j[1]-i[1]

答案1

得分: 1

在这种情况下,您可以使用广播替代嵌套的for循环。广播是NumPy中的一个强大功能,允许您在不同形状的数组之间执行逐元素操作。

import numpy as np

values = np.array([[33, 23, 0.6, 0.8, 1.5], [34, 25, 0.9, 0.7, 1.6]])
points = np.array([[0.2, 19.8], [0.6, 19.8]])

# 使用广播从values中减去points
pa_x = values[:, 0] - points[:, 0][:, np.newaxis]
pa_y = values[:, 1] - points[:, 1][:, np.newaxis]

np.newaxis 用于为 points[:, 0] 和 points[:, 1] 添加新的轴,以便在从 values[:, 0] 和 values[:, 1] 减去时可以正确广播它们。

运行此代码后,pa_x 和 pa_y 将包含values中每个元素与points中每个元素之间的差异,而无需使用嵌套的for循环。

英文:

You can replace the nested for loop with broadcasting in this case. Broadcasting is a powerful feature in NumPy that allows you to perform element-wise operations between arrays of different shapes.

import numpy as np

values = np.array([[33, 23, 0.6, 0.8, 1.5], [34, 25, 0.9, 0.7, 1.6]])
points = np.array([[0.2, 19.8], [0.6, 19.8]])

# Broadcasting to subtract points from values
pa_x = values[:, 0] - points[:, 0][:, np.newaxis]
pa_y = values[:, 1] - points[:, 1][:, np.newaxis]

np.newaxis is used to add a new axis to points[:, 0] and points[:, 1], so that they can be broadcasted properly when subtracted from values[:, 0] and values[:, 1], respectively.

After running this code, pa_x and pa_y will contain the differences between each element in values and each element in points, without using a nested for loop.

huangapple
  • 本文由 发表于 2023年6月22日 17:11:33
  • 转载请务必保留本文链接:https://go.coder-hub.com/76530289.html
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定