fix merge

app/app.py

@@ -58,7 +58,7 @@ control_optimizer = ControlOptimizer(cfg)
 # %% Init mcts parameters
 
 # Hyperparameters mctss
-iteration_limit = 20
+iteration_limit = 50
 
 # Initialize MCTS
 searcher = mcts.mcts(iterationLimit=iteration_limit)

app/rule_extention.py

@@ -185,8 +185,8 @@ def init_extension_rules():
     rule_vocab.create_rule("InitMechanism_3_R_A", ["ROOT"], ["F", "SML", "SMRPA","SMRMA","EM","EM","EM"], 0 , 0,[(0,1),(0,2),(0,3),(1,4),(2,5),(3,6)])
     rule_vocab.create_rule("InitMechanism_3_L", ["ROOT"], ["F", "SMLP","SMLM", "SMR","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(1,4),(2,5),(3,6)])
     rule_vocab.create_rule("InitMechanism_3_L_A", ["ROOT"], ["F", "SMLPA","SMLMA", "SMR","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(1,4),(2,5),(3,6)])
-    #rule_vocab.create_rule("InitMechanism_4", ["ROOT"], ["F", "SMLP","SMLM", "SMRP","SMRM","EM","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(0,4),(1,5),(2,6),(3,7),(4,8)])
-    #rule_vocab.create_rule("InitMechanism_4_A", ["ROOT"], ["F", "SMLPA","SMLMA", "SMRPA","SMRMA","EM","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(0,4),(1,5),(2,6),(3,7),(4,8)])
+    rule_vocab.create_rule("InitMechanism_4", ["ROOT"], ["F", "SMLP","SMLM", "SMRP","SMRM","EM","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(0,4),(1,5),(2,6),(3,7),(4,8)])
+    rule_vocab.create_rule("InitMechanism_4_A", ["ROOT"], ["F", "SMLPA","SMLMA", "SMRPA","SMRMA","EM","EM","EM","EM"], 0 , 0, [(0,1),(0,2),(0,3),(0,4),(1,5),(2,6),(3,7),(4,8)])
     rule_vocab.create_rule("FingerUpper", ["EM"], ["J", "L","EM"], 0 , 2, [(0,1),(1, 2)])
 
     rule_vocab.create_rule("TerminalFlat1", ["F"], ["F1"], 0 , 0)

rostok/graph_generators/graph_environment.py

@@ -214,16 +214,28 @@ class GraphVocabularyEnvironment(GraphEnvironment):
         return list(possible_actions)
 
     def getReward(self):
+        if len(seen_graphs)>0:
+            seen_graphs_t = list(zip(*seen_graphs))
+            i = 0
+            for graph in seen_graphs_t[0]:
+                if graph == self.graph:
+                    self.reward = seen_graphs_t[1][i]
+                    self.movments_trajectory = seen_graphs_t[2][i]
+                    reporter.add_reward(self.state, self.reward, self.movments_trajectory)
+                    print('seen reward:', self.reward)
+                    return self.reward
+                i += 1
 
         result_optimizer = self.optimizer.start_optimisation(self.graph)
         self.reward = -result_optimizer[0]
         self.movments_trajectory = result_optimizer[1]
-
+        seen_graphs.append([deepcopy(self.graph), self.reward, deepcopy(self.movments_trajectory)])
         reporter.add_reward(self.state, self.reward, self.movments_trajectory)
         if self.reward > reporter.best_reward:
             reporter.best_reward = self.reward
             reporter.best_control = self.movments_trajectory
             reporter.best_state = self.state
+
         print(self.reward)
         return self.reward
 

rostok/graph_grammar/node.py

@@ -323,9 +323,19 @@ class GraphGrammar(nx.DiGraph):
 
         return list(dfs_preorder_nodes(self, self.get_root_id()))
 
-    def __eq__(self, __o) -> bool:
-        if isinstance(__o, GraphGrammar):
-            is_node_eq = __o.nodes == self.nodes
-            is_edge_eq = __o.edges == self.edges
-            return is_edge_eq and is_node_eq
+    def __eq__(self, __rhs) -> bool:
+        if isinstance(__rhs, GraphGrammar):
+            self_dfs_paths = self.graph_partition_dfs()
+            self_dfs_paths_lbl = []
+            for path in self_dfs_paths:
+                self_dfs_paths_lbl.append([self.get_node_by_id(x).label for x in path])
+
+            self_dfs_paths_lbl.sort(key = lambda x: "".join(x))
+            rhs_dfs_paths = __rhs.graph_partition_dfs()
+            rhs_dfs_paths_lbl = []
+            for path in rhs_dfs_paths:
+                rhs_dfs_paths_lbl.append([__rhs.get_node_by_id(x).label for x in path])
+
+            rhs_dfs_paths_lbl.sort(key = lambda x: "".join(x))
+            return self_dfs_paths_lbl == rhs_dfs_paths_lbl
         return False