treewide: Prefer ravel() instead of flatten()

Unless flatten()'s copying semantics are required. Signed-off-by: Jan Vesely <[email protected]>
jvesely · Jan 10, 2025 · b772fc3 · b772fc3
1 parent 4a5472c
commit b772fc3
Show file tree

Hide file tree

Showing 9 changed files with 20 additions and 18 deletions.
diff --git a/psyneulink/core/components/functions/nonstateful/optimizationfunctions.py b/psyneulink/core/components/functions/nonstateful/optimizationfunctions.py
@@ -2132,8 +2132,7 @@ def _function(self,
 
                 # Find the optimal value(s)
                 optimal_value_count = 1
-                value_sample_pairs = zip(all_values.flatten(),
-                                         [all_samples[:,i] for i in range(all_samples.shape[1])])
+                value_sample_pairs = zip(all_values.ravel(), [all_samples[:,i] for i in range(all_samples.shape[1])])
                 optimal_value, optimal_sample = next(value_sample_pairs)
 
                 # The algorithm below implements "Reservoir sampling"[0]. This

diff --git a/psyneulink/core/components/functions/stateful/integratorfunctions.py b/psyneulink/core/components/functions/stateful/integratorfunctions.py
@@ -3160,7 +3160,7 @@ def _function(self,
         previous_value = self.parameters.previous_value._get(context)
 
         try:
-            variable = variable.flatten()
+            variable = variable.ravel()
             drift = variable if len(variable) == dimension - 1 else np.full(dimension - 1, variable)
 
         except ValueError:

diff --git a/psyneulink/core/compositions/parameterestimationcomposition.py b/psyneulink/core/compositions/parameterestimationcomposition.py
@@ -609,7 +609,7 @@ def __init__(
 
                 # If the include mask is 2D, make it 1D
                 if likelihood_include_mask.ndim == 2:
-                    likelihood_include_mask = likelihood_include_mask.flatten()
+                    likelihood_include_mask = likelihood_include_mask.ravel()
 
                 self.likelihood_include_mask = likelihood_include_mask
 

diff --git a/psyneulink/core/globals/parameters.py b/psyneulink/core/globals/parameters.py
@@ -1744,7 +1744,7 @@ def _deliver_value(self, value, context=None):
                         context=execution_id,
                         value=ndArray(
                             shape=list(value.shape),
-                            data=list(value.flatten())
+                            data=list(value.ravel())
                         )
                     )
                 )

diff --git a/psyneulink/library/components/mechanisms/modulatory/learning/EMstoragemechanism.py b/psyneulink/library/components/mechanisms/modulatory/learning/EMstoragemechanism.py
@@ -676,7 +676,7 @@ def _validate_params(self, request_set, target_set=None, context=None):
         for i, learning_signal in enumerate(learning_signals[:num_match_fields]):
             learning_signal_shape = learning_signal.parameters.matrix._get(context).shape
             if concatenate_queries:
-                memory_matrix_field_shape = np.array([np.concatenate(row, dtype=object).flatten()
+                memory_matrix_field_shape = np.array([np.concatenate(row, dtype=object).ravel()
                                                       for row in memory_matrix[:,0:num_keys]]).T.shape
             else:
                 memory_matrix_field_shape = np.array(memory_matrix[:,key_indices[i]].tolist()).T.shape

diff --git a/tests/composition/test_autodiffcomposition.py b/tests/composition/test_autodiffcomposition.py
@@ -1300,9 +1300,7 @@ def test_xor_nested_train_then_no_train(self, num_epochs, learning_rate,
 
         # -----------------------------------------------------------------
 
-        xor_autodiff = AutodiffComposition(
-            learning_rate=learning_rate,
-        )
+        xor_autodiff = AutodiffComposition(learning_rate=learning_rate)
 
         xor_autodiff.add_node(xor_in)
         xor_autodiff.add_node(xor_hid)
@@ -1319,15 +1317,20 @@ def test_xor_nested_train_then_no_train(self, num_epochs, learning_rate,
         parentComposition = pnl.Composition()
         parentComposition.add_node(xor_autodiff)
 
-        input = {xor_autodiff: input_dict}
         no_training_input = {xor_autodiff: no_training_input_dict}
 
         learning_context = Context()
-        xor_autodiff.learn(inputs=input_dict, execution_mode=autodiff_mode, epochs=num_epochs, context=learning_context, patience=patience, min_delta=min_delta)
-        result1 = np.array(xor_autodiff.learning_results).flatten()
-        np.testing.assert_allclose(result1, np.array(xor_targets).flatten(), atol=0.1)
-        result2 = parentComposition.run(inputs=no_training_input, execution_mode=autodiff_mode, context=learning_context)
+        xor_autodiff.learn(inputs=input_dict,
+                           execution_mode=autodiff_mode,
+                           epochs=num_epochs,
+                           context=learning_context,
+                           patience=patience,
+                           min_delta=min_delta)
+
+        result1 = np.array(xor_autodiff.learning_results).ravel()
+        np.testing.assert_allclose(result1, np.array(xor_targets).ravel(), atol=0.1)
 
+        result2 = parentComposition.run(inputs=no_training_input, execution_mode=autodiff_mode, context=learning_context)
         np.testing.assert_allclose(result2, [[0]], atol=0.1)
 
     @pytest.mark.parametrize(

diff --git a/tests/composition/test_control.py b/tests/composition/test_control.py
@@ -2470,7 +2470,7 @@ def test_modulation_simple(self, cost, expected, exp_values, comp_mode):
         ret = comp.run(inputs={mech: [2]}, num_trials=1, execution_mode=comp_mode)
         np.testing.assert_allclose(ret, expected)
         if comp_mode == pnl.ExecutionMode.Python:
-            np.testing.assert_allclose(comp.controller.function.saved_values.flatten(), exp_values)
+            np.testing.assert_allclose(comp.controller.function.saved_values.ravel(), exp_values)
 
     @pytest.mark.benchmark
     @pytest.mark.control
@@ -3360,7 +3360,7 @@ def comp_run(inputs, execution_mode):
 
         np.testing.assert_array_equal(results, result)
         if mode == pnl.ExecutionMode.Python:
-            np.testing.assert_array_equal(saved_values.flatten(), [0.75, 1.5, 2.25])
+            np.testing.assert_array_equal(saved_values.ravel(), [0.75, 1.5, 2.25])
 
     def test_model_based_ocm_with_buffer(self):
 

diff --git a/tests/composition/test_emcomposition.py b/tests/composition/test_emcomposition.py
@@ -146,7 +146,7 @@ def test_structure(self,
             params.update({'softmax_gain': softmax_gain})
 
         em = EMComposition(**params)
-        assert np.hstack(np.array(em.memory, dtype=object).flatten()).size < 30
+        assert np.hstack(np.array(em.memory, dtype=object).ravel()).size < 30
 
         # Validate basic structure
         assert len(em.memory) == memory_capacity

diff --git a/tests/models/test_greedy_agent.py b/tests/models/test_greedy_agent.py
@@ -230,7 +230,7 @@ def action_fn(variable):
             # FIXME: The results are 'close' for both Philox and MT,
             #        because they're dominated by costs
             # FIX: Requires 1e-5 tolerance
-            np.testing.assert_allclose(np.asarray(ocm.function.saved_values).flatten(),
+            np.testing.assert_allclose(np.asarray(ocm.function.saved_values).ravel(),
                                        [-2.66258741, -22027.9970321, -22028.17515945, -44053.59867802,
                                         -22028.06045185, -44053.4048842, -44053.40736234, -66078.90687915],
                                        rtol=1e-5, atol=1e-5)