Fixed #1112 Migrate to New NumPy Random Number Generator (RNG) API by seanlaw · Pull Request #1131 · stumpy-dev/stumpy

seanlaw · 2026-02-14T05:30:35Z

Pull Request Checklist

Below is a simple checklist but please do not hesitate to ask for assistance!

To Do List

Search for np.random
Replace np.random.rand(x) with pytest.RNG.random(x)
Replace np.random.rand(x*y).reshape(x,y) with pytest.RNG.random(size=(x, y))
Replace np.random.uniform(x, y, [z]) with pytest.RNG.uniform(x, y, size=z)
Replace np.random.permutation([x, y, z]) with pytest.RNG.permutation([x, y, z])
Replace np.random.randint(x, y, z) with pytest.RNG.integers(x, y, z)
Replace np.random.choice([x, y], l, replace=True) with pytest.RNG.choice([x, y], l, replace=True)
Replace np.random.normal(size=x) with pytest.RNG.normal(size=x)

gitnotebooks · 2026-02-14T05:30:39Z

Review these changes at https://app.gitnotebooks.com/stumpy-dev/stumpy/pull/1131

seanlaw · 2026-02-14T05:31:49Z

@NimaSarajpoor I've tested the migration on test_core.py as an initial trial run. Would you mind taking a look and providing any feedback before I do the same thing on the other tests?

NimaSarajpoor

@seanlaw
I've left just one comment for your consideration.

test.sh

seanlaw · 2026-02-15T13:33:39Z

@NimaSarajpoor If it's okay with you, I'm going to add ~5 files at a time for you to incrementally review (there are 46 test files in total and I'll need to work on tutorials and other docs too).

seanlaw · 2026-02-16T02:32:56Z

Things are starting to get tricky. Previously, np.random.seed(seed) would set the seed for the global random number generator. However, with the new approach, we'll need to synchronize things across to naive.py (i.e., pass the RNG state). Extra care must be taken. Otherwise, things will fall out of sync and cause assertions to fail!

conftest.py

tests/test_core.py

seanlaw · 2026-02-26T04:57:21Z

@NimaSarajpoor I added three new files for you to review

NimaSarajpoor · 2026-02-27T02:34:58Z

tests/test_aamp_motifs.py

-    np.random.seed(1234)
+    pytest.fix_rng_state()

    # The time series is random noise with two motifs for m=10:


Suggested change

# The time series is random noise with two motifs for m=10:

# The time series is random noise with two motifs for m=20:

According to Line 80: m = 20

NimaSarajpoor · 2026-02-27T02:45:07Z

tests/test_aamp_motifs.py

    # naive.distance(naive.z_norm(T[10:30]), naive.z_norm(T[110:130])) = 0.47
    # naive.distance(naive.z_norm(T[10:30]), naive.z_norm(T[210:230])) = 0.24
    # naive.distance(naive.z_norm(T[110:130]), naive.z_norm(T[210:230])) = 0.72
    # Hence T[10:30] is the motif representative for this motif


I think these comments should be revised. They show the calculation of z-normalized distances. However, aamp_motifs is for non-normalized case.

NimaSarajpoor · 2026-02-27T03:08:34Z

tests/test_aamp_mmotifs.py

-    )
-    motif_indices_ref = np.array([[19, 77, 63, 52, 71]])
-    motif_subspaces_ref = [np.array([2])]
+    motif_distances_ref = np.array([[0.0, 0.10660435, 0.27927693]])


I cannot figure out how these numbers are obtained. Since tests are passing, I guess all is good here. I guess these numbers are revised to match the actual output. So... an off-topic question: are we cheating here by using ref values that are based on the calculation of outputs obtained from performant function ?

NimaSarajpoor · 2026-02-27T03:24:36Z

tests/test_aamp_motifs.py

    # * (almost) identical step functions at indices 10, 110 and 210
    # * identical linear slopes at indices 70 and 170
-    T = np.random.normal(size=300)
+    T = pytest.RNG.normal(size=300)


To be consistent with other test functions, I was wondering if we should go with random rather than normal distribution.

NimaSarajpoor · 2026-02-27T03:41:40Z

tests/test_aamp_ostinato.py

            npt.assert_almost_equal(ref_subseq_idx, comp_subseq_idx)


+@pytest.mark.parametrize("seed", [41, 88, 290, 292, 310, 328, 538, 556, 563, 570])


I think you missed removing this one.

NimaSarajpoor · 2026-02-27T03:42:09Z

tests/test_aamp_ostinato.py



+@pytest.mark.parametrize("seed", [41, 88, 290, 292, 310, 328, 538, 556, 563, 570])
+def test_deterministic_ostinatoed(seed, dask_cluster):


Suggested change

def test_deterministic_ostinatoed(seed, dask_cluster):

def test_deterministic_ostinatoed(dask_cluster):

NimaSarajpoor · 2026-02-27T04:01:26Z

tests/test_aamp_motifs.py


    T[70:90] = np.arange(m) * 0.1
    T[170:190] = np.arange(m) * 0.1
    # naive.distance(naive.z_norm(T[70:90]), naive.z_norm(T[170:190])) = 0.0


Suggested change

# naive.distance(T[70:90], T[170:190]) = 0.0

NimaSarajpoor · 2026-02-27T04:04:21Z

tests/test_aamp_ostinato.py

-def test_random_ostinatoed(seed, dask_cluster):
-    with Client(dask_cluster) as dask_client:
+def test_deterministic_ostinato():
+    pytest.fix_rng_state()


Do we need to fix state here? I can see we do not fix state for test_random_ostinato, and that should be OK.

NimaSarajpoor · 2026-02-27T04:05:29Z

tests/test_aamp_ostinato.py


+@pytest.mark.parametrize("seed", [41, 88, 290, 292, 310, 328, 538, 556, 563, 570])
+def test_deterministic_ostinatoed(seed, dask_cluster):
+    pytest.fix_rng_state()


Do we need to fix state here? I can see we do not fix state for test_random_ostinato.

Initial migration conversion

8fe16ec

seanlaw requested a review from NimaSarajpoor February 14, 2026 05:30

Fixed typo

da87749

NimaSarajpoor reviewed Feb 14, 2026

View reviewed changes

test.sh Outdated Show resolved Hide resolved

Removed --exclude flag

9efa0f5

Added ability to set state like np.random.seed(seed)

ade0375

Removed conftest.py from coverage testing

87f06a6

NimaSarajpoor reviewed Feb 16, 2026

View reviewed changes

conftest.py Outdated Show resolved Hide resolved

tests/test_core.py Show resolved Hide resolved

seanlaw added 2 commits February 16, 2026 17:18

Minor change

acf1394

Updated three more tests

2d799e8

NimaSarajpoor reviewed Feb 27, 2026

View reviewed changes

	# The time series is random noise with two motifs for m=10:
	# The time series is random noise with two motifs for m=20:

		npt.assert_almost_equal(ref_subseq_idx, comp_subseq_idx)


		@pytest.mark.parametrize("seed", [41, 88, 290, 292, 310, 328, 538, 556, 563, 570])



		@pytest.mark.parametrize("seed", [41, 88, 290, 292, 310, 328, 538, 556, 563, 570])
		def test_deterministic_ostinatoed(seed, dask_cluster):

	def test_deterministic_ostinatoed(seed, dask_cluster):
	def test_deterministic_ostinatoed(dask_cluster):

Conversation

seanlaw commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Checklist

To Do List

Uh oh!

gitnotebooks bot commented Feb 14, 2026

Uh oh!

seanlaw commented Feb 14, 2026

Uh oh!

NimaSarajpoor left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

seanlaw commented Feb 15, 2026

Uh oh!

seanlaw commented Feb 16, 2026

Uh oh!

Uh oh!

Uh oh!

seanlaw commented Feb 26, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

seanlaw commented Feb 14, 2026 •

edited

Loading