Use testing.parametrize

chainer · Apr 17, 2017 · 45f1cfa · 45f1cfa
1 parent a460de3
commit 45f1cfa
Showing 1 changed file with 15 additions and 21 deletions.
diff --git a/tests/cupy_tests/linalg_tests/test_decomposition.py b/tests/cupy_tests/linalg_tests/test_decomposition.py
@@ -61,6 +61,9 @@ def test_mode(self):
         self.check_mode(numpy.random.randn(5, 4), mode=self.mode)
 
 
+@testing.parameterize(*testing.product({
+    'full_matrices': [True, False],
+}))
 @unittest.skipUnless(
     cuda.cusolver_enabled, 'Only cusolver in CUDA 8.0 is supported')
 @testing.gpu
@@ -69,39 +72,30 @@ class TestSVD(unittest.TestCase):
     _multiprocess_can_split_ = True
 
     @testing.for_float_dtypes(no_float16=True)
-    def check_usv(self, array, full_matrices, dtype):
+    def check_usv(self, array, dtype):
         a_cpu = numpy.asarray(array, dtype=dtype)
         a_gpu = cupy.asarray(array, dtype=dtype)
-        result_cpu = numpy.linalg.svd(a_cpu, full_matrices=full_matrices)
-        result_gpu = cupy.linalg.svd(a_gpu, full_matrices=full_matrices)
+        result_cpu = numpy.linalg.svd(a_cpu, full_matrices=self.full_matrices)
+        result_gpu = cupy.linalg.svd(a_gpu, full_matrices=self.full_matrices)
         for b_cpu, b_gpu in zip(result_cpu, result_gpu):
             # Use abs to support an inverse vector
             cupy.testing.assert_allclose(
                 numpy.abs(b_cpu), cupy.abs(b_gpu), atol=1e-4)
 
     @testing.for_float_dtypes(no_float16=True)
     @testing.numpy_cupy_allclose(atol=1e-4)
-    def check_singular(self, array, xp, dtype, full_matrices):
+    def check_singular(self, array, xp, dtype):
         a = xp.asarray(array, dtype=dtype)
         result = xp.linalg.svd(
-            a, full_matrices=full_matrices, compute_uv=False)
+            a, full_matrices=self.full_matrices, compute_uv=False)
         return result
 
-    def test_svd_full_matrices(self):
-        self.check_usv(numpy.random.randn(2, 3), full_matrices=True)
-        self.check_usv(numpy.random.randn(2, 2), full_matrices=True)
-        self.check_usv(numpy.random.randn(3, 2), full_matrices=True)
-
-    def test_svd_no_full_matrices(self):
-        self.check_usv(numpy.random.randn(2, 3), full_matrices=False)
-        self.check_usv(numpy.random.randn(2, 2), full_matrices=False)
-        self.check_usv(numpy.random.randn(3, 2), full_matrices=False)
+    def test_svd(self):
+        self.check_usv(numpy.random.randn(2, 3))
+        self.check_usv(numpy.random.randn(2, 2))
+        self.check_usv(numpy.random.randn(3, 2))
 
     def test_svd_no_uv(self):
-        self.check_singular(numpy.random.randn(2, 3), full_matrices=True)
-        self.check_singular(numpy.random.randn(2, 2), full_matrices=True)
-        self.check_singular(numpy.random.randn(3, 2), full_matrices=True)
-
-        self.check_singular(numpy.random.randn(2, 3), full_matrices=False)
-        self.check_singular(numpy.random.randn(2, 2), full_matrices=False)
-        self.check_singular(numpy.random.randn(3, 2), full_matrices=False)
+        self.check_singular(numpy.random.randn(2, 3))
+        self.check_singular(numpy.random.randn(2, 2))
+        self.check_singular(numpy.random.randn(3, 2))