star

venv/lib/python3.9/site-packages/sklearn/utils/tests/test_arpack.py

@@ -0,0 +1,16 @@
+import pytest
+from numpy.testing import assert_allclose
+
+from sklearn.utils import check_random_state
+from sklearn.utils._arpack import _init_arpack_v0
+
+
+@pytest.mark.parametrize("seed", range(100))
+def test_init_arpack_v0(seed):
+    # check that the initialization a sampling from an uniform distribution
+    # where we can fix the random state
+    size = 1000
+    v0 = _init_arpack_v0(size, seed)
+
+    rng = check_random_state(seed)
+    assert_allclose(v0, rng.uniform(-1, 1, size=size))