cannot determine Numba type of <class 'function'>

See original GitHub issue

I am trying to run a function in parallel mode. It is iterating over elements from a list and performing some operations. Since the elements and operations are independent of each other, the operations can be done in parallel, however when i go to compile it(i am using spyder btw) using @njit(parallel= True), I am getting an error saying that ‘cannot determine Numba type of <class ‘function’>’

The loop which is doing the iteration has functions like:

Addition
Division and Multiplication
np.cross(cross product)
np. array
Calling functions that are defined earlier(outside @njit)

The error dump is pasted below:

runfile('C:/Users/Arghyadeep/Desktop/p1.py', wdir='C:/Users/Arghyadeep/Desktop')
Traceback (most recent call last):

  File "<ipython-input-38-1f08c1876264>", line 1, in <module>
    runfile('C:/Users/Arghyadeep/Desktop/p1.py', wdir='C:/Users/Arghyadeep/Desktop')

  File "C:\Users\Arghyadeep\Anaconda3\lib\site-packages\spyder_kernels\customize\spydercustomize.py", line 827, in runfile
    execfile(filename, namespace)

  File "C:\Users\Arghyadeep\Anaconda3\lib\site-packages\spyder_kernels\customize\spydercustomize.py", line 110, in execfile
    exec(compile(f.read(), filename, 'exec'), namespace)

  File "C:/Users/Arghyadeep/Desktop/p1.py", line 73, in <module>
    parallelBoris(10)

  File "C:\Users\Arghyadeep\Anaconda3\lib\site-packages\numba\dispatcher.py", line 376, in _compile_for_args
    error_rewrite(e, 'typing')

  File "C:\Users\Arghyadeep\Anaconda3\lib\site-packages\numba\dispatcher.py", line 343, in error_rewrite
    reraise(type(e), e, None)

  File "C:\Users\Arghyadeep\Anaconda3\lib\site-packages\numba\six.py", line 658, in reraise
    raise value.with_traceback(tb)

TypingError: cannot determine Numba type of <class 'function'>

Issue Analytics

State:
Created 4 years ago
Comments:7 (5 by maintainers)

Top GitHub Comments

6reactions

stuartarchibaldcommented, Oct 1, 2019

Did you do something like:

In [1]: from numba import njit                                                                                                  

In [2]: def foo(x): 
   ...:     return x + 1 
   ...:                                                                                                                         

In [3]: @njit 
   ...: def call_foo(y): 
   ...:     return foo(y) 
   ...:                                                                                                                         

In [4]: call_foo(10)                

TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Untyped global name 'foo': cannot determine Numba type of <class 'function'>

File "<ipython-input-3-8c790192a815>", line 3:
def call_foo(y):
    return foo(y)
    ^

the root cause of which is that the functions called by @jit decorated functions also need to be @jit decorated. i.e.

In [1]: from numba import njit                                                                                                  

In [2]: @njit 
   ...: def foo(x): 
   ...:     return x + 1 
   ...:                                                                                                                         

In [3]: @njit 
   ...: def call_foo(y): 
   ...:     return foo(y) 
   ...:                                                                                                                         

In [4]: call_foo(10)                                                                                                            
Out[4]: 11

0reactions

stuartarchibaldcommented, Oct 4, 2019

Closing this issue as it seems to be resolved. If this is not the case please re-open with a comment about any item that appears to be unresolved. Many thanks.

Top Results From Across the Web

TypingError: cannot determine Numba type of <class ...

I have a simple function to rank poker hands ...

Cannot determine Numba type of <class 'type'> - Support

Cannot determine Numba type of <class 'type'>. This is my code and the input xq is a (N,3) ndarray. @njit(parallel=True) def fraction_field(xq): N...

First Steps with numba — numba 0.12.0 documentation

Starting with numba version 0.12, it is possible to use numba.jit without providing a type-signature for the function. This functionality was provided by...

TypingError: cannot determine Numba type of class ... - YouTube

Become part of the top 3% of the developers by applying to Toptal https://topt.al/25cXVn--Music by Eric Matyashttps://www.soundimage.

Types and signatures - Numba documentation

As an optimizing compiler, Numba needs to decide on the type of each variable to ... First-class function support is enabled for all...