implement dpnp.piecewise

[vtavana]

In this PR, dpnp.piecewise is implemented.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2550/index.html

[github-actions]

Array API standard conformance tests for dpnp=0.19.0dev3=py313h509198e_25 ran successfully.
Passed: 1227
Failed: 0
Skipped: 9

[coveralls]

coverage: 72.251% (+0.06%) from 72.19%
when pulling 642f9f8 on impl-piecewise
into d975818 on master.

[antonwolfy]

There is no check that condlist is exactly a list. Do we need that or it is assuming to accept any sequence?

[vtavana]

In practice, condlist can be a list, tuple or ndarray in both NumPy and CuPy. And dpnp follows the same practice.

import numpy
a = numpy.linspace(-2.5, 2.5, 6)
numpy.piecewise(a, [a < 0, a >= 0], [-1, 1])  # condlist is list
# array([-1., -1., -1.,  1.,  1.,  1.])

numpy.piecewise(a, (a < 0, a >= 0), [-1, 1])  # condlist is tuple
# array([-1., -1., -1.,  1.,  1.,  1.])

numpy.piecewise(a, numpy.array([a < 0, a >= 0]), [-1, 1])  # condlist is array
# array([-1., -1., -1.,  1.,  1.,  1.])

import cupy
x = cupy.array(a)

cupy.piecewise(x, [x < 0, x >= 0], [-1, 1])  # condlist is list
# array([-1., -1., -1.,  1.,  1.,  1.])

cupy.piecewise(x, (x < 0, x >= 0), [-1, 1])  # condlist is tuple
# array([-1., -1., -1.,  1.,  1.,  1.])

cupy.piecewise(x, cupy.array([x < 0, x >= 0]), [-1, 1])  # condlist is array
# array([-1., -1., -1.,  1.,  1.,  1.])

import dpnp
ia = dpnp.array(a)
dpnp.piecewise(ia, [ia < 0, ia >= 0], [-1, 1])  # condlist is list
# array([-1., -1., -1.,  1.,  1.,  1.])

dpnp.piecewise(ia, (ia < 0, ia >= 0), [-1, 1])  # condlist is tuple
# array([-1., -1., -1.,  1.,  1.,  1.])

dpnp.piecewise(ia, dpnp.array([ia < 0, ia >= 0]), [-1, 1])  # condlist is array
# array([-1., -1., -1.,  1.,  1.,  1.])

[antonwolfy]

No, I mean condlist might not be a subscriptable object and so condlist[0] would raise TypeError

[vlad-perevezentsev]

No more comments from my side
LGTM
Thank you @vtavana

[antonwolfy]

need to add exception text

[antonwolfy]

Should we check for usm_ndarray here?