numpy.clip accept numpy.array for a_max

---

Hi Numba team, Thanks for your contributions. Please could you add support of numpy.clip() so that it can accept a numpy.array() as the a_max and a_min?

The example code below works when running normally, but does not work when using @njit.

mask = a > b; out[mask] = (np.clip(c, 0, d)[mask]) where mask is a boolean of length 1000 c.shape = (1000, 10, 10) d.shape = (1000, 10, 10) (i.e. not just a number)

I get this error when running with @jit(nopython=True) "numba.core.errors.TypingError: Failed in nopython mode pipeline (step: nopython frontend) No implementation of function Function(<function clip at 0x000001F8477D5C10>) found for signature:

clip(array(float64, 3d, C), Literalint, array(int32, 3d, C))

There are 2 candidate implementations:

• Of which 2 did not match due to: Overload in function ‘np_clip’: File: numba\np\arrayobj.py: Line 1805. With argument(s): ‘(array(float64, 3d, C), int64, array(int32, 3d, C))’: Rejected as the implementation raised a specific error: TypingError: The argument “a_max” must be a number"

Feature request

[czgdp1807]

I think I can work on this. Please let me know if that’s okay.

Edit 1 - I figured that we first need to implement, np.broadcast because a can be a scalar as well in which case we need an easy to implement broadcasting all the inputs to the same shape before starting to iterate on the inputs.

Edit 2 - I checked, @guilhermeleobas has recently implemented np.broadcast_arrays, so it should be possible to implement this feature. So, may be I can proceed with enhancing numpy.clip? Please let me know.

[Twomem]

Great, thanks @czgdp1807 !