address reviews

.pre-commit-config.yaml

@@ -95,15 +95,14 @@ repos:
         entry: 'pytest\.xfail'
         language: pygrep
         types: [python]
-      - id: use-only-default-rng
-        name: use-only-default-rng
-        description: 'Enforce that `default_rng` is used instead of `np.random.seed` and it must be seeded.'
+      - id: no-unseeded-default-rng
+        name: no-unseeded-default-rng
+        description: 'Enforce that no non-seeded default_rng is used and default_rng is used instead of np.random.seed'
         entry: |
-            (?x)
-            # no unseeded default_rng
-            |default_rng\(\)
-            # no np.random.seed
-            |np.random.seed\(
+          # Check for usage of default_rng without seeding
+          default_rng\(\)|
+          # Check for usage of np.random.seed
+          np.random.seed\(
         language: pygrep
         types: [python]
       - id: cmake-format

python/cudf/cudf/tests/test_categorical.py

@@ -252,7 +252,7 @@ def test_cat_series_binop_error():
 @pytest.mark.parametrize("num_elements", [10, 100, 1000])
 def test_categorical_unique(num_elements):
     # create categorical series
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     pd_cat = pd.Categorical(
         pd.Series(
             rng.choice(

python/cudf/cudf/tests/test_dataframe.py

@@ -3903,7 +3903,7 @@ def test_select_dtype_datetime_with_frequency():
 
 
 def test_dataframe_describe_exclude():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     data_length = 10000
 
     df = cudf.DataFrame()
@@ -3919,7 +3919,7 @@ def test_dataframe_describe_exclude():
 
 
 def test_dataframe_describe_include():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     data_length = 10000
 
     df = cudf.DataFrame()
@@ -3934,7 +3934,7 @@ def test_dataframe_describe_include():
 
 
 def test_dataframe_describe_default():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     data_length = 10000
 
     df = cudf.DataFrame()
@@ -3948,7 +3948,7 @@ def test_dataframe_describe_default():
 
 
 def test_series_describe_include_all():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     data_length = 10000
 
     df = cudf.DataFrame()
@@ -3971,7 +3971,7 @@ def test_series_describe_include_all():
 
 
 def test_dataframe_describe_percentiles():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=12)
     data_length = 10000
     sample_percentiles = [0.0, 0.1, 0.33, 0.84, 0.4, 0.99]
 

python/cudf/cudf/tests/test_stats.py

@@ -19,8 +19,6 @@
 
 interpolation_methods = ["linear", "lower", "higher", "midpoint", "nearest"]
 
-rng = np.random.default_rng(seed=0)
-
 
 @pytest.mark.parametrize("method", methods)
 @pytest.mark.parametrize("dtype", params_dtypes)
@@ -86,6 +84,7 @@ def test_series_std(ddof):
 
 
 def test_series_unique():
+    rng = np.random.default_rng(seed=0)
     for size in [10**x for x in range(5)]:
         arr = rng.integers(low=-1, high=10, size=size)
         mask = arr != -1
@@ -131,6 +130,7 @@ def test_series_nunique(nan_as_null, dropna):
 
 
 def test_series_scale():
+    rng = np.random.default_rng(seed=0)
     arr = pd.Series(rng.integers(low=-10, high=10, size=100))
     sr = cudf.Series(arr)
 
@@ -231,8 +231,8 @@ def test_misc_quantiles(data, q):
 @pytest.mark.parametrize(
     "data",
     [
-        {"data": rng.normal(-100, 100, 1000)},
-        {"data": rng.integers(-50, 50, 1000)},
+        {"data": np.random.default_rng(seed=0).normal(-100, 100, 1000)},
+        {"data": np.random.default_rng(seed=0).integers(-50, 50, 1000)},
         {"data": (np.zeros(100))},
         {"data": np.repeat(np.nan, 100)},
         {"data": np.array([1.123, 2.343, np.nan, 0.0])},
@@ -282,8 +282,8 @@ def test_kurt_skew_error(op):
 @pytest.mark.parametrize(
     "data",
     [
-        cudf.Series(rng.normal(-100, 100, 1000)),
-        cudf.Series(rng.integers(-50, 50, 1000)),
+        cudf.Series(np.random.default_rng(seed=0).normal(-100, 100, 1000)),
+        cudf.Series(np.random.default_rng(seed=0).integers(-50, 50, 1000)),
         cudf.Series(np.zeros(100)),
         cudf.Series(np.repeat(np.nan, 100)),
         cudf.Series(np.array([1.123, 2.343, np.nan, 0.0])),
@@ -346,8 +346,8 @@ def test_series_median(dtype, num_na):
 @pytest.mark.parametrize(
     "data",
     [
-        rng.normal(-100, 100, 1000),
-        rng.integers(-50, 50, 1000),
+        np.random.default_rng(seed=0).normal(-100, 100, 1000),
+        np.random.default_rng(seed=0).integers(-50, 50, 1000),
         np.zeros(100),
         np.array([1.123, 2.343, np.nan, 0.0]),
         np.array([-2, 3.75, 6, None, None, None, -8.5, None, 4.2]),
@@ -381,8 +381,8 @@ def test_series_pct_change(data, periods, fill_method):
 @pytest.mark.parametrize(
     "data1",
     [
-        rng.normal(-100, 100, 1000),
-        rng.integers(-50, 50, 1000),
+        np.random.default_rng(seed=0).normal(-100, 100, 1000),
+        np.random.default_rng(seed=0).integers(-50, 50, 1000),
         np.zeros(100),
         np.repeat(np.nan, 100),
         np.array([1.123, 2.343, np.nan, 0.0]),
@@ -395,8 +395,8 @@ def test_series_pct_change(data, periods, fill_method):
 @pytest.mark.parametrize(
     "data2",
     [
-        rng.normal(-100, 100, 1000),
-        rng.integers(-50, 50, 1000),
+        np.random.default_rng(seed=0).normal(-100, 100, 1000),
+        np.random.default_rng(seed=0).integers(-50, 50, 1000),
         np.zeros(100),
         np.repeat(np.nan, 100),
         np.array([1.123, 2.343, np.nan, 0.0]),
@@ -425,8 +425,8 @@ def test_cov1d(data1, data2):
 @pytest.mark.parametrize(
     "data1",
     [
-        rng.normal(-100, 100, 1000),
-        rng.integers(-50, 50, 1000),
+        np.random.default_rng(seed=0).normal(-100, 100, 1000),
+        np.random.default_rng(seed=0).integers(-50, 50, 1000),
         np.zeros(100),
         np.repeat(np.nan, 100),
         np.array([1.123, 2.343, np.nan, 0.0]),
@@ -439,8 +439,8 @@ def test_cov1d(data1, data2):
 @pytest.mark.parametrize(
     "data2",
     [
-        rng.normal(-100, 100, 1000),
-        rng.integers(-50, 50, 1000),
+        np.random.default_rng(seed=0).normal(-100, 100, 1000),
+        np.random.default_rng(seed=0).integers(-50, 50, 1000),
         np.zeros(100),
         np.repeat(np.nan, 100),
         np.array([1.123, 2.343, np.nan, 0.0]),

python/cudf/cudf/tests/test_string.py

@@ -133,9 +133,14 @@ def test_string_get_item(ps_gs, item):
         np.array([False] * 5),
         cupy.asarray(np.array([True] * 5)),
         cupy.asarray(np.array([False] * 5)),
-        rng.integers(0, 2, 5).astype("bool").tolist(),
-        rng.integers(0, 2, 5).astype("bool"),
-        cupy.asarray(rng.integers(0, 2, 5).astype("bool")),
+        np.random.default_rng(seed=0)
+        .integers(0, 2, 5)
+        .astype("bool")
+        .tolist(),
+        np.random.default_rng(seed=0).integers(0, 2, 5).astype("bool"),
+        cupy.asarray(
+            np.random.default_rng(seed=0).integers(0, 2, 5).astype("bool")
+        ),
     ],
 )
 def test_string_bool_mask(ps_gs, item):
@@ -1079,6 +1084,7 @@ def test_string_set_scalar(scalar):
 
 
 def test_string_index():
+    rng = np.random.default_rng(seed=0)
     pdf = pd.DataFrame(rng.random(size=(5, 5)))
     gdf = cudf.DataFrame.from_pandas(pdf)
     stringIndex = ["a", "b", "c", "d", "e"]

python/cudf/cudf/utils/hash_vocab_utils.py

@@ -108,7 +108,7 @@ def _find_hash_for_internal(hash_bin, rng):
             return bins, a, b
 
 
-def _perfect_hash(integers, rng, max_constant):
+def _perfect_hash(integers, max_constant, rng):
     num_top_level_bins = len(integers) // 4
 
     init_bins, init_a, init_b = _pick_initial_a_b(
@@ -264,7 +264,7 @@ def hash_vocab(
         hash_table,
         inner_table_coeffs,
         offsets_into_ht,
-    ) = _perfect_hash(keys, rng, 10)
+    ) = _perfect_hash(keys, 10, rng)
 
     _pack_keys_and_values(hash_table, hashed_vocab)
     _store_func(

python/cudf/cudf_pandas_tests/test_cudf_pandas.py

@@ -1142,7 +1142,7 @@ def test_private_method_result_wrapped():
 
 
 def test_numpy_var():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=42)
     data = rng.random(1000)
     psr = pd.Series(data)
     sr = xpd.Series(data)

python/cudf/cudf_pandas_tests/test_profiler.py

@@ -23,7 +23,7 @@
     reason="function names change across versions of pandas, so making sure it only runs on latest version of pandas",
 )
 def test_profiler():
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=42)
     with Profiler() as profiler:
         df = pd.DataFrame(
             {

python/cudf/cudf_pandas_tests/third_party_integration_tests/tests/test_stumpy_distributed.py

@@ -31,14 +31,14 @@ def dask_client():
 
 
 def test_1d_distributed(dask_client):
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=42)
     ts = pd.Series(rng.random(100))
     m = 10
     return stumpy.stumped(dask_client, ts, m)
 
 
 def test_multidimensional_distributed_timeseries(dask_client):
-    rng = np.random.default_rng(seed=0)
+    rng = np.random.default_rng(seed=42)
     # Each row represents data from a different dimension while each column represents
     # data from the same dimension
     your_time_series = rng.random(3, 1000)