Cython says buffer types only allowed as function local variables even for ndarray.copy()

Question

I am new to Cython and encountered this code snippet:

import numpy as np
cimport numpy as np

testarray = np.arange(5)
cdef np.ndarray[np.int_t, ndim=1] testarray1 = testarray.copy()
cdef np.ndarray[np.float_t, ndim=1] testarray2 = testarray.astype(np.float)

During compilation, it said Buffer types only allowed as function local variables. However, I am using .copy() or .astype() which is returning not a memoryview, but a copy. Why is this still happening? How can I get around this?

Thanks!

testarray itself is a buffer. So why not put everything into a function and call it from Python? — Midnighter
– Midnighter, Commented May 23, 2014 at 21:06
@Midnighter Thanks! I was trying to define some global constants to save several lines of code... And I never thought that testarray is a buffer itself before. Thanks for the input - I'd go with more verbose code then I guess! :) — Yuxiang Wang
– Yuxiang Wang, Commented May 23, 2014 at 21:08

JoshAdel · Accepted Answer · 2014-05-24 00:48:33Z

17

When you define an array in cython using np.ndarray[Type, dim], that is accessing the python buffer interface, and those can't be set as module level variables. This is a separate issue from views vs copies of numpy array data.

Typically if I want to have an array as a module level variable (i.e not local to a method), I define a typed memoryview and then set it within a method using something like (untested):

import numpy as np
cimport numpy as np

cdef np.int_t[:] testarray1

def init_arrays(np.int_t[:] testarray):
    global testarray1
    testarray1 = testarray.copy()

answered May 24, 2014 at 0:48

JoshAdel

69.1k27 gold badges146 silver badges146 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Yuxiang Wang Over a year ago

Thank you so much Josh! It is very helpful. However, the memoryviewslice would not have ndarray methods such as ndarray.mean() or ndarray.max()... Any chance you know a workaround?

JoshAdel Over a year ago

As long as it doesn't incur too much overhead for what you're doing, you can use np.asarray to coerce the typed memoryview to a numpy array docs.cython.org/src/userguide/… within the method that you are using the memoryview in.

Yuxiang Wang Over a year ago

Excellent! That is awesome functionality that I didn't notice. Thank you so much Josh!

Kyle Barron Over a year ago

Why do you use testarray.copy() instead of just testarray1 = testarray?

Kyle Barron Over a year ago

From stackoverflow.com/a/24166170, "I am using .copy() to guarantee that you have contiguous data in memory. This would be unnecessary if positions was Fortran-contiguous.". So I'm guessing the copy() here is also to ensure contiguous memory

Collectives™ on Stack Overflow

Cython says buffer types only allowed as function local variables even for ndarray.copy()

1 Answer 1

5 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related