Add basic optimization canonicalizations to "fast_compile" mode by default

Author: brandonwillard

aesara/tensor/basic_opt.py

@@ -507,7 +507,9 @@ def register(inner_lopt):
         return register
     else:
         name = kwargs.pop("name", None) or lopt.__name__
-        compile.optdb["canonicalize"].register(name, lopt, "fast_run", *tags, **kwargs)
+        compile.optdb["canonicalize"].register(
+            name, lopt, "fast_run", "fast_compile", *tags, **kwargs
+        )
         return lopt
 
 

tests/d3viz/test_formatting.py

@@ -48,7 +48,7 @@ def test_ofg(self):
         assert len(sub_graphs) == 2
         ofg1, ofg2 = sub_graphs
         if config.mode == "FAST_COMPILE":
-            assert len(ofg1.get_nodes()) == 9
+            assert len(ofg1.get_nodes()) == 8
         else:
             assert len(ofg1.get_nodes()) == 5
         assert len(ofg1.get_nodes()) == len(ofg2.get_nodes())

tests/sparse/test_basic.py

@@ -1563,18 +1563,8 @@ def f_b(x, y):
 
                 assert np.all(f_a(vx, vy) == f_b(vx, vy))
                 topo = f_a.maker.fgraph.toposort()
-                if aesara.config.mode != "FAST_COMPILE":
-                    nb = 0
-                else:
-                    nb = 1
-                assert (
-                    sum(
-                        [
-                            isinstance(node.op, (Dot, Usmm, UsmmCscDense))
-                            for node in topo
-                        ]
-                    )
-                    == nb
+                assert not any(
+                    isinstance(node.op, (Dot, Usmm, UsmmCscDense)) for node in topo
                 )
 
     def test_int32_dtype(self):
@@ -1822,13 +1812,8 @@ def f_b(z, a, x, y):
             )
             assert all(f_shape(a_data, x_data, y_data) == f_b_out.shape)
             topo = f_shape.maker.fgraph.toposort()
-            if aesara.config.mode != "FAST_COMPILE":
-                nb = 0
-            else:
-                nb = 1
-            assert (
-                sum([isinstance(node.op, (Dot, Usmm, UsmmCscDense)) for node in topo])
-                == nb
+            assert not any(
+                isinstance(node.op, (Dot, Usmm, UsmmCscDense)) for node in topo
             )
 
 

tests/tensor/test_basic.py

@@ -1028,11 +1028,6 @@ def setup_method(self):
         self.split_op_class = Split
         self.make_vector_op = MakeVector()
         self.floatX = config.floatX
-        self.hide_error = config.mode not in [
-            "DebugMode",
-            "DEBUG_MODE",
-            "FAST_COMPILE",
-        ]
         self.shared = shared
 
     def eval_outputs_and_check_join(self, outputs):
@@ -1712,17 +1707,6 @@ def get_mat(s1, s2):
             for node in topo:
                 assert not isinstance(node.op, type(self.join_op))
 
-        with config.change_flags(compute_test_value="off"):
-            # Test hide error
-            x1.set_value(get_mat(3, 4))
-            x2.set_value(get_mat(3, 4))
-            x3.set_value(get_mat(2, 5))
-            if not self.hide_error:
-                with pytest.raises(ValueError):
-                    f()
-            else:
-                f()
-
     def test_rebroadcast(self):
         # Regression test for a crash that used to happen when rebroadcasting.
         x = TensorType(self.floatX, [False, False, True])()

tests/tensor/test_sharedvar.py

@@ -439,14 +439,6 @@ def test_specify_shape(self):
             with pytest.raises(AssertionError):
                 specify_shape_fct()
 
-            # No assertion will be raised as the Op is removed from the graph
-            # when their is optimization
-            if aesara.config.mode not in ["FAST_COMPILE", "DebugMode", "DEBUG_MODE"]:
-                shape_constant_fct()
-            else:
-                with pytest.raises(AssertionError):
-                    shape_constant_fct()
-
         def test_specify_shape_partial(self):
             dtype = self.dtype
             if dtype is None:
@@ -502,13 +494,6 @@ def test_specify_shape_partial(self):
             with pytest.raises(AssertionError):
                 specify_shape_fct()
 
-            # No assertion will be raised as the Op is removed from the graph
-            if aesara.config.mode not in ["FAST_COMPILE", "DebugMode", "DEBUG_MODE"]:
-                shape_constant_fct()
-            else:
-                with pytest.raises(AssertionError):
-                    shape_constant_fct()
-
         def test_specify_shape_inplace(self):
             # test that specify_shape don't break inserting inplace op
 

tests/tensor/test_subtensor.py

@@ -1077,7 +1077,8 @@ def fct2(t):
             else:
                 ops = subtensor_ops
             if idx is idxs[0]:
-                f = self.function([], [gn.shape, n[idx_].shape], op=ops, N=0, N_fast=2)
+                # TODO FIXME: This is a very poorly specified test.
+                f = self.function([], [gn.shape, n[idx_].shape], op=ops, N=0, N_fast=0)
                 f()
 
     def test_wrong_exception_regression(self):
@@ -1129,7 +1130,7 @@ def test_shape_list(self):
             data = np.asarray(data, dtype=self.dtype)
             n = self.shared(data)
             t = n[idx]
-            f = self.function([], t.shape, op=subtensor_ops, N=0, N_fast=1)
+            f = self.function([], t.shape, op=subtensor_ops, N=0, N_fast=0)
             val = f()
             assert np.allclose(val, data[idx].shape)
 

tests/tensor/test_subtensor_opt.py

@@ -1174,18 +1174,14 @@ def test_const6(self):
         data = self.rng.uniform(size=(8, 8, 8)).astype(config.floatX)
         x = tensor3("x")
 
-        nops = 1
-        if config.mode == "FAST_COMPILE":
-            nops = 2
-
         # test 1)
         y = x[3:6, 2:6, 1:7][1]
         fun = function([x], y)
         val = fun(data)
         assert np.all(val == data[3:6, 2:6, 1:7][1])
         assert (
             len([n for n in fun.maker.fgraph.toposort() if isinstance(n.op, Subtensor)])
-            == nops
+            == 1
         )
 
         # test 2)
@@ -1195,7 +1191,7 @@ def test_const6(self):
         assert np.all(val == data[2, 3][1])
         assert (
             len([n for n in fun.maker.fgraph.toposort() if isinstance(n.op, Subtensor)])
-            == nops
+            == 1
         )
 
         # test 3)
@@ -1205,7 +1201,7 @@ def test_const6(self):
         assert np.all(val == data[3:6, 2, 1:7][1])
         assert (
             len([n for n in fun.maker.fgraph.toposort() if isinstance(n.op, Subtensor)])
-            == nops
+            == 1
         )
 
     def test_scalar6(self):
@@ -1590,10 +1586,7 @@ def test_incsubtensor_x_zeros(self):
 
         assert len(inc_nodes) == 1
         node_is_set_instead_of_inc = inc_nodes[0].op.set_instead_of_inc
-        mode = config.mode
-        assert (mode != "FAST_COMPILE" and node_is_set_instead_of_inc) or (
-            mode == "FAST_COMPILE" and not node_is_set_instead_of_inc
-        )
+        assert node_is_set_instead_of_inc
         test_X = np.random.random((4, 4)).astype(config.floatX)
         utt.assert_allclose(f(test_X), test_X)