np.reshape 1d array to 2d array - 选择正确的尺寸

Question

I'm trying to reshape a 1d array to a 2d array with numpy's reshape:

import numpy as np
inputArray =np.random.randint(low=0, high=4, size=160000)
inputArray_ = inputArray.reshape(-1,4000, 4000,4)

Which returns a value error:

ValueError: cannot reshape array of size 160000 into shape (400,400,4)

Answer 1

Use

inputArray_  = np.reshape(inputArray, (-1, 2))

Or

inputArray_ = np.reshape(inputArray, (len(inputArray)/2,2))

Answer 2

since 400*400*4 = 640,000 is bigger than 160000 you cannot reshape.

You don't have enough values to fill the new shape.

640,000-160,000 = 480,000. you lack 480,000 values.

divide your shape of 160000 by the other dimensions-multiplicated, if a int is the result, it works.

eg

inputArray_ = inputArray.reshape(-1,40, 40, 10)

this will result in a shape of [10,40,40,10]

since 160000 / (40*40*10) = 10 ---> 10 is the dim that the "-1" takes