Merge pull request #270 from jakobj/maint/remove-graph-call

Remove support for calling graph directly

cgp/cartesian_graph.py

@@ -202,20 +202,6 @@ def determine_active_regions(self) -> List[int]:
 
         return active_regions
 
-    def __call__(self, x: List[float]) -> List[float]:
-        # store values of x in input nodes
-        for i, xi in enumerate(x):
-            assert isinstance(self._nodes[i], InputNode)
-            self._nodes[i]._output = xi
-
-        # evaluate active nodes in order
-        active_nodes_by_hidden_column_idx = self._determine_active_nodes()
-        for hidden_column_idx in sorted(active_nodes_by_hidden_column_idx):
-            for node in active_nodes_by_hidden_column_idx[hidden_column_idx]:
-                node(x, self)
-
-        return [node._output for node in self.output_nodes]
-
     def __getitem__(self, key: int) -> Node:
         return self._nodes[key]
 

cgp/node.py

@@ -57,9 +57,6 @@ def __init_subclass__(cls: Type["Node"]) -> None:
         super().__init_subclass__()
         register(cls)
 
-    def __call__(self, x: List[float], graph: "CartesianGraph") -> None:
-        raise NotImplementedError
-
     @property
     def arity(self) -> int:
         return self._arity
@@ -210,19 +207,6 @@ def initial_value(cls, parameter_name: str) -> float:
     def parameter_str(self) -> List[str]:
         return self._parameter_str
 
-    def __call__(self, x: List[float], graph: "CartesianGraph") -> None:
-        output_str = str(self._def_output)
-
-        if len(self._parameter_names) > 0:
-            raise RuntimeError("can not call a node that uses parameters")
-
-        for input_name in self._input_names:
-            idx = self._extract_index_from_input_name(input_name)
-            output_str = output_str.replace(input_name, f"{graph[self._input_nodes[idx]].output}")
-
-        exec_str = f"self._output = {output_str}"
-        exec(exec_str)
-
     def _replace_input_names(self, output_str: str, graph: "CartesianGraph") -> str:
         for input_name in self._input_names:
             idx = self._extract_index_from_input_name(input_name)

test/test_cartesian_graph.py

@@ -7,28 +7,6 @@
 from cgp.genome import ID_INPUT_NODE, ID_NON_CODING_GENE, ID_OUTPUT_NODE
 
 
-def test_direct_input_output():
-    params = {"n_inputs": 1, "n_outputs": 1, "n_columns": 3, "n_rows": 3, "levels_back": 2}
-    primitives = (cgp.Add, cgp.Sub)
-    genome = cgp.Genome(
-        params["n_inputs"],
-        params["n_outputs"],
-        params["n_columns"],
-        params["n_rows"],
-        params["levels_back"],
-        primitives,
-    )
-    genome.randomize(np.random)
-
-    genome[-2:] = [0, ID_NON_CODING_GENE]  # set inputs for output node to input node
-    graph = cgp.CartesianGraph(genome)
-
-    x = [2.14159]
-    y = graph(x)
-
-    assert x[0] == pytest.approx(y[0])
-
-
 def test_to_func_simple():
     primitives = (cgp.Add,)
     genome = cgp.Genome(2, 1, 1, 1, 1, primitives)

test/test_node.py

@@ -24,19 +24,16 @@ def test_inputs_are_cut_to_match_arity():
     assert node.input_nodes == input_nodes[:2]
 
 
-def _test_graph_call_and_to_x_compilations(
+def _test_to_x_compilations(
     genome,
     x,
     y_target,
     *,
-    test_graph_call=True,
     test_to_func=True,
     test_to_numpy=True,
     test_to_torch=True,
     test_to_sympy=True,
 ):
-    if test_graph_call:
-        _test_graph_call(genome, x, y_target)
     if test_to_func:
         _test_to_func(genome, x, y_target)
     if test_to_numpy:
@@ -47,11 +44,6 @@ def _test_graph_call_and_to_x_compilations(
         _test_to_sympy(genome, x, y_target)
 
 
-def _test_graph_call(genome, x, y_target):
-    graph = cgp.CartesianGraph(genome)
-    assert graph(x) == pytest.approx(y_target)
-
-
 def _test_to_func(genome, x, y_target):
     graph = cgp.CartesianGraph(genome)
     assert graph.to_func()(x) == pytest.approx(y_target)
@@ -107,7 +99,7 @@ def test_add():
     x = [5.0, 1.5]
     y_target = [x[0] + x[1]]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_sub():
@@ -141,7 +133,7 @@ def test_sub():
     x = [5.0, 1.5]
     y_target = [x[0] - x[1]]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_mul():
@@ -175,7 +167,7 @@ def test_mul():
     x = [5.0, 1.5]
     y_target = [x[0] * x[1]]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_div():
@@ -209,7 +201,7 @@ def test_div():
     x = [5.0, 1.5]
     y_target = [x[0] / x[1]]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_pow():
@@ -243,7 +235,7 @@ def test_pow():
     x = [5.0, 1.5]
     y_target = [x[0] ** x[1]]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_constant_float():
@@ -273,7 +265,7 @@ def test_constant_float():
     x = [1.0, 2.0]
     y_target = [1.0]  # by default the output value of the ConstantFloat node is 1.0
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_parameter():
@@ -292,7 +284,7 @@ def test_parameter():
     x = [1.0]
     y_target = [1.0]  # by default the output value of the Parameter node is 1.0
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target, test_graph_call=False)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_parameter_w_custom_initial_value():
@@ -316,7 +308,7 @@ class CustomParameter(cgp.Parameter):
     x = [1.0]
     y_target = [initial_value]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target, test_graph_call=False)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_parameter_two_nodes():
@@ -353,7 +345,7 @@ def test_parameter_two_nodes():
     # hence the sum of two Parameter nodes is 2.0
     y_target = [2.0]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target, test_graph_call=False)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_parameter_w_random_initial_value(rng_seed):
@@ -486,15 +478,15 @@ def test_if_else_operator():
 
     x_0 = [1.0, 10.0, -20.0]
     y_target_0 = [10.0]
-    _test_graph_call_and_to_x_compilations(genome, x_0, y_target_0)
+    _test_to_x_compilations(genome, x_0, y_target_0)
 
     x_1 = [0.0, 10.0, -20.0]
     y_target_1 = [10.0]
-    _test_graph_call_and_to_x_compilations(genome, x_1, y_target_1)
+    _test_to_x_compilations(genome, x_1, y_target_1)
 
     x_2 = [-1.0, 10.0, -20.0]
     y_target_2 = [-20.0]
-    _test_graph_call_and_to_x_compilations(genome, x_2, y_target_2)
+    _test_to_x_compilations(genome, x_2, y_target_2)
 
 
 def test_raise_broken_def_output():
@@ -591,7 +583,7 @@ class MyScaledAdd(cgp.node.OperatorNode):
     x = [5.0, 1.5]
     y_target = [2 * (x[0] + x[1])]
 
-    _test_graph_call_and_to_x_compilations(genome, x, y_target)
+    _test_to_x_compilations(genome, x, y_target)
 
 
 def test_custom_node_with_custom_atomic_operator():
@@ -635,9 +627,7 @@ class MyScaledAdd(cgp.node.OperatorNode):
     x = [5.0, 1.5]
     y_target = [2 * (x[0] + x[1])]
 
-    _test_graph_call_and_to_x_compilations(
-        genome, x, y_target, test_graph_call=False, test_to_sympy=False
-    )
+    _test_to_x_compilations(genome, x, y_target, test_to_sympy=False)
 
 
 def test_custom_node_with_custom_atomic_operator_with_external_library():
@@ -683,6 +673,4 @@ class MyScaledAdd(cgp.node.OperatorNode):
     x = [5.0, 1.5]
     y_target = [scipy_const.golden_ratio * (x[0] + x[1])]
 
-    _test_graph_call_and_to_x_compilations(
-        genome, x, y_target, test_graph_call=False, test_to_sympy=False
-    )
+    _test_to_x_compilations(genome, x, y_target, test_to_sympy=False)