From edb14eb1154edb97c8bbb1035b18521fa7cf02ea Mon Sep 17 00:00:00 2001
From: "Thomas S. Binns" <t.s.binns@outlook.com>
Date: Tue, 17 Sep 2024 18:14:11 +0200
Subject: [PATCH] [MAINT] Update NumPy random generation (#11)

---
 src/pyparrm/parrm.py |  8 ++++----
 tests/test_parrm.py  | 29 +++++++++++++++++------------
 2 files changed, 21 insertions(+), 16 deletions(-)

diff --git a/src/pyparrm/parrm.py b/src/pyparrm/parrm.py
index ff80e6f..fdeba21 100644
--- a/src/pyparrm/parrm.py
+++ b/src/pyparrm/parrm.py
@@ -282,7 +282,7 @@ def _standardise_data(self) -> None:
 
     def _optimise_period_estimate(self) -> None:
         """Optimise artefact period estimate."""
-        random_state = np.random.RandomState(self._random_seed)
+        random_state = np.random.default_rng(self._random_seed)
 
         estimated_period = self._assumed_periods
 
@@ -329,7 +329,7 @@ def _get_centre_indices(
         self,
         use_n_samples: int,
         ignore_portion: float,
-        random_state: np.random.RandomState,
+        random_state: np.random.Generator,
     ) -> np.ndarray:
         """Get indices for samples in the centre of the data segment.
 
@@ -341,7 +341,7 @@ def _get_centre_indices(
         ignore_portion : float
             Portion of the data segment to ignore when getting the indices.
 
-        random_state : numpy.random.RandomState
+        random_state : numpy.random.Generator
             Random state object to use to generate numbers if the available number of
             samples is less than that requested.
 
@@ -367,7 +367,7 @@ def _get_centre_indices(
         )
         return (
             np.unique(
-                random_state.randint(
+                random_state.integers(
                     0, end_idx - start_idx, np.min((use_n_samples, end_idx - start_idx))
                 )
             )
diff --git a/tests/test_parrm.py b/tests/test_parrm.py
index b97f7ab..473d84f 100644
--- a/tests/test_parrm.py
+++ b/tests/test_parrm.py
@@ -13,8 +13,6 @@
 from pyparrm.data import DATASETS
 from pyparrm._utils._power import compute_psd
 
-
-random = np.random.RandomState(44)
 sampling_freq = 20  # Hz
 artefact_freq = 10  # Hz
 
@@ -25,7 +23,8 @@
 @pytest.mark.parametrize("n_jobs", [1, -1])
 def test_parrm(n_chans: int, n_samples: int, verbose: bool, n_jobs: int):
     """Test that PARRM can run."""
-    data = random.rand(n_chans, n_samples)
+    random = np.random.default_rng(44)
+    data = random.standard_normal((n_chans, n_samples))
 
     parrm = PARRM(
         data=data,
@@ -43,7 +42,7 @@ def test_parrm(n_chans: int, n_samples: int, verbose: bool, n_jobs: int):
     assert filtered_data.shape == data.shape
     assert isinstance(filtered_data, np.ndarray)
 
-    other_data = random.rand(1, 50)
+    other_data = random.standard_normal((1, 50))
     other_filtered_data = parrm.filter_data(other_data)
     assert other_filtered_data.shape == other_data.shape
 
@@ -62,7 +61,8 @@ def test_parrm_attrs():
     The returned attributes should simply be a copy of their private
     counterparts.
     """
-    data = random.rand(1, 100)
+    random = np.random.default_rng(44)
+    data = random.standard_normal((1, 100))
 
     parrm = PARRM(
         data=data,
@@ -98,7 +98,8 @@ def test_parrm_attrs():
 
 def test_parrm_wrong_type_inputs():
     """Test that inputs of wrong types to PARRM are caught."""
-    data = random.rand(1, 100)
+    random = np.random.default_rng(44)
+    data = random.standard_normal((1, 100))
 
     # init object
     with pytest.raises(TypeError, match="`data` must be a NumPy array."):
@@ -185,12 +186,13 @@ def test_parrm_wrong_type_inputs():
 
 def test_parrm_wrong_value_inputs():
     """Test that inputs of wrong values to PARRM are caught."""
-    data = random.rand(1, 100)
+    random = np.random.default_rng(44)
+    data = random.standard_normal((1, 100))
 
     # init object
     with pytest.raises(ValueError, match="`data` must be a 2D array."):
         PARRM(
-            data=random.rand(1, 1, 1),
+            data=random.standard_normal((1, 1, 1)),
             sampling_freq=sampling_freq,
             artefact_freq=artefact_freq,
         )
@@ -281,13 +283,14 @@ def test_parrm_wrong_value_inputs():
 
     # filter_data
     with pytest.raises(ValueError, match="`data` must be a 2D array."):
-        parrm.filter_data(data=random.rand(100))
+        parrm.filter_data(data=random.standard_normal((100,)))
 
 
 def test_parrm_premature_method_attribute_calls():
     """Test that errors raised for PARRM methods/attrs. called prematurely."""
+    random = np.random.default_rng(44)
     parrm = PARRM(
-        data=random.rand(1, 100),
+        data=random.standard_normal((1, 100)),
         sampling_freq=sampling_freq,
         artefact_freq=artefact_freq,
         verbose=False,
@@ -316,8 +319,9 @@ def test_parrm_premature_method_attribute_calls():
 
 def test_parrm_missing_filter_inputs():
     """Test that PARRM can compute values for missing filter inputs."""
+    random = np.random.default_rng(44)
     parrm = PARRM(
-        data=random.rand(1, 100),
+        data=random.standard_normal((1, 100)),
         sampling_freq=sampling_freq,
         artefact_freq=artefact_freq,
         verbose=False,
@@ -333,7 +337,8 @@ def test_parrm_missing_filter_inputs():
 @pytest.mark.parametrize("n_jobs", [1, 2])
 def test_compute_psd(n_chans: int, n_jobs: int):
     """Test that PSD computation runs."""
-    data = random.rand(n_chans, 100)
+    random = np.random.default_rng(44)
+    data = random.standard_normal((n_chans, 100))
 
     n_freqs = 5
     freqs, psd = compute_psd(