Misc cleanups for style and consistency.

Change: 147297316

tensorflow/contrib/distributions/python/kernel_tests/dirichlet_multinomial_test.py

@@ -222,9 +222,10 @@ def testCovarianceFromSampling(self):
       dist = ds.DirichletMultinomial(n, alpha)
       x = dist.sample(int(250e3), seed=1)
       sample_mean = math_ops.reduce_mean(x, 0)
-      x_centered = x - sample_mean[None, ...]
+      x_centered = x - sample_mean[array_ops.newaxis, ...]
       sample_cov = math_ops.reduce_mean(math_ops.matmul(
-          x_centered[..., None], x_centered[..., None, :]), 0)
+          x_centered[..., array_ops.newaxis],
+          x_centered[..., array_ops.newaxis, :]), 0)
       sample_var = array_ops.matrix_diag_part(sample_cov)
       sample_stddev = math_ops.sqrt(sample_var)
       [
@@ -317,7 +318,7 @@ def testCovarianceNAlphaBroadcast(self):
       dist = ds.DirichletMultinomial(ns, alpha)
       covariance = dist.covariance()
       expected_covariance = shared_matrix * (
-          ns * (ns + alpha_0) / (1 + alpha_0))[..., None]
+          ns * (ns + alpha_0) / (1 + alpha_0))[..., array_ops.newaxis]
 
       self.assertEqual([4, 3, 3], covariance.get_shape())
       self.assertAllClose(expected_covariance, covariance.eval())

tensorflow/contrib/distributions/python/kernel_tests/multinomial_test.py

@@ -239,17 +239,18 @@ def testCovarianceFromSampling(self):
     # via broadcast between alpha, n.
     theta = np.array([[1., 2, 3],
                       [2.5, 4, 0.01]], dtype=np.float32)
-    theta /= np.sum(theta, 1)[..., None]
+    theta /= np.sum(theta, 1)[..., array_ops.newaxis]
     # Ideally we'd be able to test broadcasting but, the multinomial sampler
     # doesn't support different total counts.
     n = np.float32(5)
     with self.test_session() as sess:
       dist = ds.Multinomial(n, theta)  # batch_shape=[2], event_shape=[3]
       x = dist.sample(int(250e3), seed=1)
       sample_mean = math_ops.reduce_mean(x, 0)
-      x_centered = x - sample_mean[None, ...]
+      x_centered = x - sample_mean[array_ops.newaxis, ...]
       sample_cov = math_ops.reduce_mean(math_ops.matmul(
-          x_centered[..., None], x_centered[..., None, :]), 0)
+          x_centered[..., array_ops.newaxis],
+          x_centered[..., array_ops.newaxis, :]), 0)
       sample_var = array_ops.matrix_diag_part(sample_cov)
       sample_stddev = math_ops.sqrt(sample_var)
       [

tensorflow/contrib/distributions/python/kernel_tests/vector_student_t_test.py

@@ -176,7 +176,7 @@ def testProbScalarBaseDistributionNonScalarTransform(self):
     x = 2. * self._rng.rand(4, 3, 3).astype(np.float32) - 1.
 
     expected_mst = _FakeVectorStudentT(
-        df=np.tile(df, len(scale_diag)),
+        df=np.tile(df, reps=len(scale_diag)),
         loc=loc,
         scale_tril=scale_tril)
 
@@ -207,7 +207,7 @@ def testProbScalarBaseDistributionNonScalarTransformDynamic(self):
     x = 2. * self._rng.rand(4, 3, 3).astype(np.float32) - 1.
 
     expected_mst = _FakeVectorStudentT(
-        df=np.tile(df, len(scale_diag)),
+        df=np.tile(df, reps=len(scale_diag)),
         loc=loc,
         scale_tril=scale_tril)
 
@@ -236,8 +236,9 @@ def testProbNonScalarBaseDistributionScalarTransform(self):
 
     expected_mst = _FakeVectorStudentT(
         df=df,
-        loc=np.tile(loc[None, :], [len(df), 1]),
-        scale_tril=np.tile(scale_tril[None, :, :], [len(df), 1, 1]))
+        loc=np.tile(loc[array_ops.newaxis, :], reps=[len(df), 1]),
+        scale_tril=np.tile(scale_tril[array_ops.newaxis, :, :],
+                           reps=[len(df), 1, 1]))
 
     with self.test_session():
       actual_mst = _VectorStudentT(df=df, loc=loc, scale_diag=scale_diag,
@@ -261,8 +262,9 @@ def testProbNonScalarBaseDistributionScalarTransformDynamic(self):
 
     expected_mst = _FakeVectorStudentT(
         df=df,
-        loc=np.tile(loc[None, :], [len(df), 1]),
-        scale_tril=np.tile(scale_tril[None, :, :], [len(df), 1, 1]))
+        loc=np.tile(loc[array_ops.newaxis, :], reps=[len(df), 1]),
+        scale_tril=np.tile(scale_tril[array_ops.newaxis, :, :],
+                           reps=[len(df), 1, 1]))
 
     with self.test_session():
       df_pl = array_ops.placeholder(dtypes.float32, name="df")

tensorflow/contrib/distributions/python/ops/bernoulli.py

@@ -57,15 +57,15 @@ def __init__(self,
         Bernoulli distribution. Only one of `logits` or `probs` should be passed
         in.
       dtype: The type of the event samples. Default: `int32`.
-      validate_args: Python `Boolean`, default `False`. When `True` distribution
+      validate_args: Python `bool`, default `False`. When `True` distribution
         parameters are checked for validity despite possibly degrading runtime
         performance. When `False` invalid inputs may silently render incorrect
         outputs.
-      allow_nan_stats: Python `Boolean`, default `True`. When `True`,
+      allow_nan_stats: Python `bool`, default `True`. When `True`,
         statistics (e.g., mean, mode, variance) use the value "`NaN`" to
-        indicate the result is undefined.  When `False`, an exception is raised
+        indicate the result is undefined. When `False`, an exception is raised
         if one or more of the statistic's batch members are undefined.
-      name: `String` name prefixed to Ops created by this class.
+      name: Python `str` name prefixed to Ops created by this class.
 
     Raises:
       ValueError: If p and logits are passed, or if neither are passed.
@@ -114,7 +114,7 @@ def _event_shape(self):
     return tensor_shape.scalar()
 
   def _sample_n(self, n, seed=None):
-    new_shape = array_ops.concat(([n], self.batch_shape_tensor()), 0)
+    new_shape = array_ops.concat([[n], self.batch_shape_tensor()], 0)
     uniform = random_ops.random_uniform(
         new_shape, seed=seed, dtype=self.probs.dtype)
     sample = math_ops.less(uniform, self.probs)

tensorflow/contrib/distributions/python/ops/beta.py

@@ -139,15 +139,15 @@ def __init__(self,
       concentration0: Positive floating-point `Tensor` indicating mean
         number of failures; aka "beta". Otherwise has same semantics as
         `concentration1`.
-      validate_args: Python `Boolean`, default `False`. When `True` distribution
+      validate_args: Python `bool`, default `False`. When `True` distribution
         parameters are checked for validity despite possibly degrading runtime
         performance. When `False` invalid inputs may silently render incorrect
         outputs.
-      allow_nan_stats: Python `Boolean`, default `True`. When `True`, statistics
+      allow_nan_stats: Python `bool`, default `True`. When `True`, statistics
         (e.g., mean, mode, variance) use the value "`NaN`" to indicate the
-        result is undefined.  When `False`, an exception is raised if one or
+        result is undefined. When `False`, an exception is raised if one or
         more of the statistic's batch members are undefined.
-      name: `String` name prefixed to Ops created by this class.
+      name: Python `str` name prefixed to Ops created by this class.
     """
     parameters = locals()
     with ops.name_scope(name, values=[concentration1,
@@ -267,7 +267,7 @@ def _variance(self):
   @distribution_util.AppendDocstring(
       """Note: The mode is undefined when `concentration1 <= 1` or
       `concentration0 <= 1`. If `self.allow_nan_stats` is `True`, `NaN`
-      is used for undefined modes.  If `self.allow_nan_stats` is `False` an
+      is used for undefined modes. If `self.allow_nan_stats` is `False` an
       exception is raised when one or more modes are undefined.""")
   def _mode(self):
     mode = (self.concentration1 - 1.) / (self.total_concentration - 2.)