Numpy recarray.dot()函数| Python
原文:https://www . geesforgeks . org/numpy-recarray-dot-function-python/
在 numpy 中,数组可能具有包含字段的数据类型,类似于电子表格中的列。一个例子是[(a, int), (b, float)]
,数组中的每个条目都是一对(int,float)。通常,这些属性是使用字典查找来访问的,例如arr['a'] and arr['b']
。使用arr.a and arr.b
,记录数组允许字段作为数组的成员被访问。
numpy.recarray.dot()
函数返回两个记录数组的乘积。
语法:
numpy.recarray.dot(arr, out=None)
参数: arr : 秒录阵。 out:【n 数组,可选】存储结果的位置。 - >如果提供,它必须具有输入广播到的形状。 - >如果未提供或无,则返回新分配的阵列。
Return : 除非指定 out,否则将返回一个保存结果的新数组,在这种情况下,将返回该数组。
代码#1 :
```py
Python program explaining
numpy.recarray.dot() method
importing numpy as geek
import numpy as geek
creating 2 input array with 2 different field
in_arr1 = geek.array([[(5.0, 2), (3.0, -4), (6.0, 9)], [(9.0, 1), (5.0, 4), (-12.0, -7)]], dtype =[('a', float), ('b', int)])
in_arr2 = geek.array([[(2.0, 1), (4.0, -3)], [(8.0, 3), (6.0, 5)], [(6.0, -5), (-5.0, 4)]], dtype =[('a', float), ('b', int)])
print ("1st Input array : ", in_arr1) print ("2nd Input array : ", in_arr2)
convert it to a record array,
using arr.view(np.recarray)
rec_arr1 = in_arr1.view(geek.recarray) print("1st Record array of float: ", rec_arr1.a) print("1st Record array of int: ", rec_arr1.b) rec_arr2 = in_arr2.view(geek.recarray) print("2nd Record array of float: ", rec_arr2.a) print("2nd Record array of int: ", rec_arr2.b)
applying recarray.dot methods
between two float record array
out_arr1 = rec_arr1.a.dot( rec_arr2.a) print ("Output float array : ", out_arr1)
applying recarray.dot methods
between two int record array
out_arr1 = rec_arr1.b.dot( rec_arr2.b) print ("Output int array : ", out_arr1) ```
Output:
```py 1st Input array : [[( 5., 2) ( 3., -4) ( 6., 9)] [( 9., 1) ( 5., 4) (-12., -7)]] 2nd Input array : [[( 2., 1) ( 4., -3)] [( 8., 3) ( 6., 5)] [( 6., -5) (-5., 4)]] 1st Record array of float: [[ 5. 3. 6.] [ 9. 5. -12.]] 1st Record array of int: [[ 2 -4 9] [ 1 4 -7]] 2nd Record array of float: [[ 2. 4.] [ 8. 6.] [ 6. -5.]] 2nd Record array of int: [[ 1 -3] [ 3 5] [-5 4]] Output float array : [[ 70. 8.] [-14. 126.]] Output int array : [[-55 10] [ 48 -11]]
```