Describe the issue:

A TypeError is raised on line 1253 when both a particular axis is requested and varargs is passed with length > 1. The implementation of gradient seems not to consider the case where a user might pass in spacing for each dimension AND want only a specific dimension of the gradient out. It is easy to reproduce this error by simply combining two of the examples given in the gradient documentation.

Reproduce the code example:

import numpy as np
dx = 2.
y = [1., 1.5, 3.5]
np.gradient(np.array([[1, 2, 6], [3, 4, 5]]), dx, y, axis=0)

Error message:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[6], line 1
----> 1 np.gradient(np.array([[1, 2, 6], [3, 4, 5]]), dx, y, axis=0)

File ~\miniforge3\Lib\site-packages\numpy\lib\_function_base_impl.py:1253, in gradient(f, axis, edge_order, *varargs)
   1251         dx[i] = diffx
   1252 else:
-> 1253     raise TypeError("invalid number of arguments")
   1255 if edge_order > 2:
   1256     raise ValueError("'edge_order' greater than 2 not supported")

TypeError: invalid number of arguments

Python and NumPy Versions:

2.3.0
3.12.10 | packaged by conda-forge | (main, Apr 10 2025, 22:08:16) [MSC v.1943 64 bit (AMD64)]

Runtime Environment:

No response

Context for the issue:

I need to calculate vorticity of a fluid velocity field in as efficient a manner as possible for a mathematics research code that I am developing at a large state institution. This means that I only need certain components of the gradient, and fluid velocity field comes from a rectilinear grid. So I need to be able to pass both arguments into numpy.gradient. Beyond that, calculating the gradient of a vector field is a typical operations for numerical work; the API for such a function should work as advertised. Also, this does not seem like something that would be too time consuming to fix.