numpy.empty references previously created empty array

[montefer]

Reproducing code example:

import numpy as np

for x in range(5):
    print(np.empty([2,3])+x)

Expected Output

[[0.00000000e+000 4.67296746e-307 1.69121096e-306]
 [1.60219035e-306 8.90104239e-307 3.27489435e-032]]
[[1. 1. 1.]
 [1. 1. 1.]]
[[2. 2. 2.]
 [2. 2. 2.]]
[[3. 3. 3.]
 [3. 3. 3.]]
[[4. 4. 4.]
 [4. 4. 4.]]

So each time an empty array is initialized and then printed out, I think we should expect to see whatever the present value of x is to be added to each 'empty' index in the array? However, what actually occurs is this:

Actual Output

[[0.00000000e+000 4.67296746e-307 1.69121096e-306]
 [1.60219035e-306 8.90104239e-307 3.27489435e-032]]
[[1. 1. 1.]
 [1. 1. 1.]]
[[3. 3. 3.]
 [3. 3. 3.]]
[[6. 6. 6.]
 [6. 6. 6.]]
[[10. 10. 10.]
 [10. 10. 10.]]

It appears that numpy continues to reference the last created 'empty' array even though we're calling the function five different times, so the values add cumulatively.

I think we should expect it to act similarly to np.zeros in that the values aren't added cumulatively, see below for an example.

Output with np.zeros

>>>import numpy as np

>>>for x in range(5):
...    print(np.zeros([2,3])+x)
…
[[0. 0. 0.]
 [0. 0. 0.]]
[[1. 1. 1.]
 [1. 1. 1.]]
[[2. 2. 2.]
 [2. 2. 2.]]
[[3. 3. 3.]
 [3. 3. 3.]]
[[4. 4. 4.]
 [4. 4. 4.]]

Is this a bug? Or is this just the way that np.empty is supposed to act?

Numpy/Python version information:

1.15.1 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:06:47) [MSC v.1914 32 bit (Intel)]

[pv]

The behavior is correct, that's how the allocator works.

[eric-wieser]

empty does not mean zeros