用 astype 重铸 NumPy 数组切片

Question

The code

a = 100. * np.random.randn(200)                                                                        
a = a.astype(int)  

recasts a as an array of integer numbers. Meanwhile, the code

a = 100. * np.random.randn(200)                                                                        
a[:] = a.astype(int) 

recasts a as an array of integer floats obtained by truncating the original values of a . Why doesn't the second case also give integers?

I'm currently using Python 3.8 and NumPy 1.23.

Answer 1

To clarify, a.astype(int) return a new array with dtype('int32') .

• Line a = a.astype(int) alters array a with the new array return by astype(int) .

• Meanwhile, a[:] = a.astype(int) only update the values of the a array, not the dtype.