MazeF1

Author: Norbert Kozlowski

.editorconfig

@@ -15,6 +15,9 @@ indent_style = space
 indent_size = 4
 max_line_length = 79
 
+[*.maze]
+insert_final_newline = false
+
 # Matches the exact files either package.json or .travis.yml
 [*.yml]
 indent_style = space

alcs/environment/maze/Maze.py

@@ -89,8 +89,8 @@ def insert_animat(self, pos_x: int = None, pos_y: int = None) -> None:
             possible_coords = []
             for x in range(0, self.max_x):
                 for y in range(0, self.max_y):
-                    if self.matrix[x][y] == self.mapping['path']['value']:
-                        possible_coords.append((x, y))
+                    if self.matrix[y][x] == self.mapping['path']['value']:
+                        possible_coords.append((y, x))
 
             starting_position = random.choice(possible_coords)
             self.animat_pos_x = starting_position[1]

alcs/environment/maze/MazeMapping.py

@@ -26,4 +26,4 @@ def find_value(self, symbol: str) -> int:
             if v['symbol'] == symbol:
                 return v['value']
 
-        raise ValueError('No mapping for symbol ' + symbol)
+        raise ValueError('No mapping for symbol [{}]'.format(symbol))

examples/helpers/visualization.py

@@ -14,30 +14,30 @@ def plot_performance(**kwargs):
 
     ax1 = fig.add_subplot(221)
     ax1.plot(time, total_classifiers, 'b')
-    ax1.plot(time, _filter(total_classifiers, f_reward), 'ro', linewidth=2.0)
+    ax1.plot(time, _filter(total_classifiers, f_reward), 'r.')
     ax1.set_title('Total Classifiers')
     ax1.set_xlabel('Time')
     ax1.set_ylabel('Macro-classifiers')
     ax1.grid(True)
 
     ax2 = fig.add_subplot(222)
     ax2.plot(time, avg_quality, 'g')
-    ax2.plot(time, _filter(avg_quality, f_reward), 'ro', linewidth=2.0)
+    ax2.plot(time, _filter(avg_quality, f_reward), 'r.')
     ax2.set_title('Quality')
     ax2.set_xlabel('Time')
     ax2.set_ylabel('Quality per classifier')
     ax2.grid(True)
 
     ax3 = fig.add_subplot(223)
     ax3.plot(time, avg_fitness, 'y')
-    ax3.plot(time, _filter(avg_fitness, f_reward), 'ro', linewidth=2.0)
+    ax3.plot(time, _filter(avg_fitness, f_reward), 'r.')
     ax3.set_title('Fitness')
     ax3.set_xlabel('Time')
     ax3.set_ylabel('Fitness per classifier')
     ax3.grid(True)
 
     plt.tight_layout()
-    plt.savefig('learning_progress.png')
+    plt.savefig('learning_progress.png', dpi=200)
 
 
 def _filter(metrics: list, mask: list) -> list:

examples/run_acs2.py

@@ -19,13 +19,13 @@
 if __name__ == '__main__':
 
     # Load environment
-    env = maze.Maze('mazes/m1.maze')
+    env = maze.Maze('mazes/MazeF1.maze')
 
     # Initialize agent
     acs2 = ACS2(env)
 
     # Evaluate simulation
-    metrics = acs2.evaluate(500)
+    metrics = acs2.evaluate(50)
 
     # Plot results
     plot_performance(**metrics)

mazes/MazeF1.maze

@@ -0,0 +1,8 @@
+4
+6
+####
+#.$#
+#.##
+#..#
+#.##
+####