ENH: Use NumPy random generators

tract_querier/tensor/tests/test_scalar_measures.py

@@ -4,8 +4,8 @@
 from numpy.testing import assert_array_almost_equal
 
 
-def test_fractional_anisotropy(N=10, random=numpy.random.RandomState(0)):
-    tensors = random.randn(N, 3, 3)
+def test_fractional_anisotropy(N=10, random=numpy.random.default_rng(0)):
+    tensors = random.standard_normal((N, 3, 3))
     fa = numpy.empty(N)
     for i, t in enumerate(tensors):
         tt = numpy.dot(t, t.T)
@@ -17,8 +17,8 @@ def test_fractional_anisotropy(N=10, random=numpy.random.RandomState(0)):
     assert_array_almost_equal(fa, scalar_measures.fractional_anisotropy(tensors))
 
 
-def test_volume_fraction(N=10, random=numpy.random.RandomState(0)):
-    tensors = random.randn(N, 3, 3)
+def test_volume_fraction(N=10, random=numpy.random.default_rng(0)):
+    tensors = random.standard_normal((N, 3, 3))
     vf = numpy.empty(N)
     for i, t in enumerate(tensors):
         tt = numpy.dot(t, t.T)
@@ -30,8 +30,8 @@ def test_volume_fraction(N=10, random=numpy.random.RandomState(0)):
     assert_array_almost_equal(vf, scalar_measures.volume_fraction(tensors))
 
 
-def test_tensor_determinant(N=10, random=numpy.random.RandomState(0)):
-    tensors = random.randn(N, 3, 3)
+def test_tensor_determinant(N=10, random=numpy.random.default_rng(0)):
+    tensors = random.standard_normal((N, 3, 3))
     dt = numpy.empty(N)
     for i, t in enumerate(tensors):
         tt = numpy.dot(t, t.T)
@@ -41,8 +41,8 @@ def test_tensor_determinant(N=10, random=numpy.random.RandomState(0)):
     assert_array_almost_equal(dt, scalar_measures.tensor_det(tensors))
 
 
-def test_tensor_traces(N=10, random=numpy.random.RandomState(0)):
-    tensors = random.randn(N, 3, 3)
+def test_tensor_traces(N=10, random=numpy.random.default_rng(0)):
+    tensors = random.standard_normal((N, 3, 3))
     res = numpy.empty(N)
     for i, t in enumerate(tensors):
         tt = numpy.dot(t, t.T)
@@ -52,9 +52,9 @@ def test_tensor_traces(N=10, random=numpy.random.RandomState(0)):
     assert_array_almost_equal(res, scalar_measures.tensor_trace(tensors))
 
 
-def test_tensor_contraction(N=10, random=numpy.random.RandomState(0)):
-    tensors1 = random.randn(N, 3, 3)
-    tensors2 = random.randn(N, 3, 3)
+def test_tensor_contraction(N=10, random=numpy.random.default_rng(0)):
+    tensors1 = random.standard_normal((N, 3, 3))
+    tensors2 = random.standard_normal((N, 3, 3))
 
     res = numpy.empty(N)
     for i in range(N):

tract_querier/tests/test_query_eval.py

@@ -1,12 +1,14 @@
 from .. import query_processor
 
-from numpy import random
 import ast
+import numpy as np
+
 
 # Ten tracts traversing random labels
 another_set = True
 while (another_set):
-    tracts_labels = dict([(i, set(random.randint(100, size=2))) for i in range(100)])
+    rng = np.random.default_rng()
+    tracts_labels = dict([(i, set(rng.integers(100, size=2))) for i in range(100)])
     labels_tracts = query_processor.labels_for_tracts(tracts_labels)
     another_set = 0 not in labels_tracts.keys() or 1 not in labels_tracts.keys()
 

tract_querier/tract_math/tract_obb.py

@@ -1007,13 +1007,14 @@ def draw_box_2d(obbs, **args):
 
 
 def draw_box_3d(obbs, tube_radius=1, color=None, **kwargs):
+        import numpy as np
         from mayavi.mlab import plot3d
-        from numpy.random import rand
+        rng = np.random.default_rng(1234)
         if isinstance(obbs, Box2D):
             obbs = [obbs]
         for obb in obbs:
             if color is None:
-                color_ = tuple(rand(3))
+                color_ = tuple(rng.random(3))
             else:
                 color_ = color
             box = obb.box

tract_querier/tractography/tests/test_tractography.py

@@ -16,8 +16,8 @@
 import copy
 from itertools import chain
 
+import numpy as np
 from numpy import all, eye, ones, allclose
-from numpy.random import randint, randn
 from numpy.testing import assert_array_equal
 
 dimensions = None
@@ -72,14 +72,15 @@ def setup_module(*args, **kwargs):
     else:
         test_active_data = False
 
-    dimensions = [(randint(5, max_tract_length), 3) for _ in range(n_tracts)]
-    tracts = [randn(*d) for d in dimensions]
+    rng = np.random.default_rng(1234)
+    dimensions = [(rng.integers(5, max_tract_length), 3) for _ in range(n_tracts)]
+    tracts = [rng.standard_normal(d) for d in dimensions]
     tracts_data = {
         'a%d' % i: [
-            randn(d[0], k)
+            rng.standard_normal((d[0], k))
             for d in dimensions
         ]
-        for i, k in zip(range(4), randint(1, 3, 9))
+        for i, k in zip(range(4), rng.integers(1, 3, 9))
     }
 
     if test_active_data: