CODEWITHSUNDEEP
pythonnumpynumbajit
Aud
Aud

Reputation: 15

Error using jit with numpy linspace function

I am trying to use jit to compile my code. At the moment, the only library I'm using besides Numba is Numpy. When specifying nopython = True, it throws a bunch of errors with linspace arrays. I have recreated the error with a very simple function:

@jit(nopython=True)
def func(Nt):
    time = np.linspace(np.int_(0),np.int_(Nt-1),np.int_(Nt),dtype=np.int_)
    return time

Nt = 10
func(Nt)

When running, the following error message (see attached image) is shown.

Numba error

I have tried many different permutations messing with the types of the arguments with no success. All I am trying to do is make a linspace array of integers from 0 to Nt - 1. Any suggestions?

Upvotes: 1

Views: 86

Answers (1)

Andrej Kesely
Andrej Kesely

Reputation: 195573

Remove the dtype= parameter:

import numba


@numba.jit(nopython=True)
def func(Nt):
    time = np.linspace(np.int_(0), np.int_(Nt - 1), np.int_(Nt))
    return time


Nt = 10
print(func(Nt))

Prints:

[0. 1. 2. 3. 4. 5. 6. 7. 8. 9.]

If you want integers, just recast it afterwards:

@numba.jit(nopython=True)
def func(Nt):
    time = np.linspace(np.int_(0), np.int_(Nt - 1), np.int_(Nt)).astype("int_")
    return time

Upvotes: 1

Related Questions