Merge pull request #196 from wbenoit26/seed-tests

Set random seed for unit tests

tests/conftest.py

@@ -1,10 +1,22 @@
+import random
+
 import numpy as np
 import pytest
 import torch
 from scipy.special import erfinv
 from torch.distributions import Uniform
 
 
+# If a fixture is doing anything random,
+# it should take this function as an argument
+@pytest.fixture(autouse=True)
+def seed_everything():
+    seed = 101589
+    torch.manual_seed(seed)
+    np.random.seed(seed)
+    random.seed(seed)
+
+
 @pytest.fixture
 def compare_against_numpy():
     """
@@ -81,9 +93,8 @@ def validate(whitened, highpass, lowpass, sample_rate, df):
     return validate
 
 
-# number of samples to draw from
-# the distributions for testing
-N_SAMPLES = 1000
+# A num_samples fixture should be defined for any
+# test that wants to use these fixtures
 
 
 @pytest.fixture(params=[256, 1024, 2048])
@@ -92,90 +103,90 @@ def sample_rate(request):
 
 
 @pytest.fixture()
-def chirp_mass(request):
+def chirp_mass(num_samples, seed_everything):
     dist = Uniform(5, 100)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def mass_ratio():
+def mass_ratio(num_samples, seed_everything):
     dist = Uniform(0.125, 0.99)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def a_1(request):
+def a_1(num_samples, seed_everything):
     dist = Uniform(0, 0.90)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def a_2(request):
+def a_2(num_samples, seed_everything):
     dist = Uniform(0, 0.90)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def tilt_1(request):
+def tilt_1(num_samples, seed_everything):
     dist = Uniform(0, torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def tilt_2(request):
+def tilt_2(num_samples, seed_everything):
     dist = Uniform(0, torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def phi_12(request):
+def phi_12(num_samples, seed_everything):
     dist = Uniform(0, 2 * torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def phi_jl(request):
+def phi_jl(num_samples, seed_everything):
     dist = Uniform(0, 2 * torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def distance(request):
+def distance(num_samples, seed_everything):
     dist = Uniform(100, 3000)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def distance_far(request):
+def distance_far(num_samples, seed_everything):
     dist = Uniform(500, 3000)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def distance_close(request):
+def distance_close(num_samples, seed_everything):
     dist = Uniform(100, 500)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def theta_jn(request):
+def theta_jn(num_samples, seed_everything):
     dist = Uniform(0, torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def phase(request):
+def phase(num_samples, seed_everything):
     dist = Uniform(0, 2 * torch.pi)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def chi1(request):
+def chi1(num_samples, seed_everything):
     dist = Uniform(-0.999, 0.999)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))
 
 
 @pytest.fixture()
-def chi2(request):
+def chi2(num_samples, seed_everything):
     dist = Uniform(-0.999, 0.999)
-    return dist.sample((N_SAMPLES,))
+    return dist.sample((num_samples,))

tests/transforms/test_iirfilter.py

@@ -5,7 +5,6 @@
 import torch
 from astropy import units as u
 from scipy.signal import filtfilt, iirfilter
-from torch.distributions import Uniform
 
 from ml4gw.constants import MSUN
 from ml4gw.transforms.iirfilter import IIRFilter
@@ -124,73 +123,9 @@ def test_filters_synthetic_signal(sample_rate, order):
         )
 
 
-N_SAMPLES = 1
-
-
-@pytest.fixture()
-def chirp_mass(request):
-    dist = Uniform(5, 100)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def mass_ratio():
-    dist = Uniform(0.125, 0.99)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def a_1(request):
-    dist = Uniform(0, 0.90)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def a_2(request):
-    dist = Uniform(0, 0.90)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def tilt_1(request):
-    dist = Uniform(0, torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def tilt_2(request):
-    dist = Uniform(0, torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def phi_12(request):
-    dist = Uniform(0, 2 * torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def phi_jl(request):
-    dist = Uniform(0, 2 * torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def distance(request):
-    dist = Uniform(100, 3000)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
-@pytest.fixture()
-def theta_jn(request):
-    dist = Uniform(0, torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
-
-
 @pytest.fixture()
-def phase(request):
-    dist = Uniform(0, 2 * torch.pi)
-    return dist.sample(torch.Size((N_SAMPLES,)))
+def num_samples():
+    return 1
 
 
 @pytest.fixture(params=[20, 40])