general tweaks

config/constants.py

@@ -44,7 +44,7 @@ SELECTION_THRESHOLD = 3  # pixels
 
 # Simulation settings
 FOOD_SPAWNING = True
-FOOD_OBJECTS_COUNT = 100
+FOOD_OBJECTS_COUNT = 500
 RANDOM_SEED = 0
 
 # Vector visualization settings
@@ -56,10 +56,10 @@ ANGULAR_TIP_SIZE = 2.5
 DIRECTION_TIP_SIZE = 3
 
 # Cell physics settings
-MAX_ACCELERATION = 0.1
+MAX_ACCELERATION = 0.125
 MAX_ANGULAR_ACCELERATION = 0.25
-MAX_VELOCITY = 0.5
-MAX_ROTATIONAL_VELOCITY = 3
+MAX_VELOCITY = 1
+MAX_ROTATIONAL_VELOCITY = 3 # degrees per tick
 
 KEYMAP_LEGEND = [
     ("WASD", "Move camera"),

core/input_handler.py

@@ -76,7 +76,7 @@ class InputHandler:
         elif event.key == pygame.K_r:
             self.camera.reset_position()
         elif event.key == pygame.K_RSHIFT:
-            self.sprint_mode = True  # Enter sprint mode
+            self.sprint_mode = not self.sprint_mode  # Enter sprint mode
 
         return running
 
@@ -84,8 +84,8 @@ class InputHandler:
         """Handle keyup events."""
         if event.key == pygame.K_LSHIFT:
             self.tps = self.default_tps
-        if event.key == pygame.K_RSHIFT:
-            self.sprint_mode = False  # Exit sprint mode
+        # if event.key == pygame.K_RSHIFT:
+        #     self.sprint_mode = False  # Exit sprint mode
 
     def _handle_mouse_down(self, event):
         """Handle mouse button down events."""

core/simulation_engine.py

@@ -3,6 +3,7 @@ import time
 import random
 import sys
 
+from world.base.brain import CellBrain, FlexibleNeuralNetwork
 from world.world import World, Position, Rotation
 from world.objects import FoodObject, DefaultCell
 from world.simulation_interface import Camera
@@ -38,14 +39,21 @@ class SimulationEngine:
         world = World(CELL_SIZE, (CELL_SIZE * GRID_WIDTH, CELL_SIZE * GRID_HEIGHT))
         random.seed(RANDOM_SEED)
 
+        half_width = GRID_WIDTH * CELL_SIZE // 2
+        half_height = GRID_HEIGHT * CELL_SIZE // 2
+
         if FOOD_SPAWNING:
             for _ in range(FOOD_OBJECTS_COUNT):
-                x = random.randint(-100, 100)
-                y = random.randint(-100, 100)
+                x = random.randint(-half_width, half_width)
+                y = random.randint(-half_height, half_height)
                 world.add_object(FoodObject(Position(x=x, y=y)))
 
-        for _ in range(20):
-            world.add_object(DefaultCell(Position(x=random.randint(-100, 100), y=random.randint(-100, 100)), Rotation(angle=0)))
+        for _ in range(300):
+            new_cell = DefaultCell(Position(x=random.randint(-half_width, half_width), y=random.randint(-half_height, half_height)), Rotation(angle=0))
+
+            new_cell.behavioral_model = new_cell.behavioral_model.mutate(3)
+
+            world.add_object(new_cell)
 
         return world
 

world/base/brain.py

@@ -1,6 +1,8 @@
 import numpy as np
 import random
 from copy import deepcopy
+
+from config.constants import MAX_VELOCITY, MAX_ROTATIONAL_VELOCITY
 from world.behavioral import BehavioralModel
 
 
@@ -10,7 +12,7 @@ class FlexibleNeuralNetwork:
     Supports variable topology with cross-layer connections.
     """
 
-    def __init__(self, input_size=2, output_size=2):
+    def __init__(self, input_size=2, output_size=2, empty_start=True):
         self.input_size = input_size
         self.output_size = output_size
 
@@ -18,8 +20,11 @@ class FlexibleNeuralNetwork:
         # Each neuron is represented by its connections and bias
         self.layers = []
 
-        # Initialize with just input and output layers (no hidden layers)
-        self._initialize_basic_network()
+        # Initialize network based on empty_start parameter
+        if empty_start:
+            self._initialize_empty_network()
+        else:
+            self._initialize_basic_network()
 
         self.network_cost = self.calculate_network_cost()
 
@@ -46,6 +51,24 @@ class FlexibleNeuralNetwork:
 
         self.layers = [input_layer, output_layer]
 
+    def _initialize_empty_network(self):
+        """Initialize an empty network with no connections or biases."""
+        # Input layer (no actual neurons, just placeholders)
+        input_layer = [{'type': 'input', 'id': i} for i in range(self.input_size)]
+
+        # Output layer with no connections and zero bias
+        output_layer = []
+        for i in range(self.output_size):
+            neuron = {
+                'type': 'output',
+                'id': f'out_{i}',
+                'bias': 0.0,
+                'connections': []  # Empty connections list
+            }
+            output_layer.append(neuron)
+
+        self.layers = [input_layer, output_layer]
+
     def _remove_duplicate_connections(self):
         """Remove duplicate connections and keep only the last weight for each unique connection."""
         for layer in self.layers[1:]:  # Skip input layer
@@ -99,14 +122,19 @@ class FlexibleNeuralNetwork:
                     continue  # Skip input neurons in hidden layers
 
                 # Calculate weighted sum of inputs
-                weighted_sum = neuron['bias']
+                weighted_sum = 0.0  # Start with 0 instead of bias
 
-                for source_layer, source_neuron, weight in neuron['connections']:
-                    if source_layer < len(activations):
-                        if source_neuron < len(activations[source_layer]):
-                            weighted_sum += activations[source_layer][source_neuron] * weight
+                # Only add bias if neuron has connections
+                if 'connections' in neuron and len(neuron['connections']) > 0:
+                    weighted_sum = neuron['bias']
+
+                    for source_layer, source_neuron, weight in neuron['connections']:
+                        if source_layer < len(activations):
+                            if source_neuron < len(activations[source_layer]):
+                                weighted_sum += activations[source_layer][source_neuron] * weight
 
                 # Apply activation function (tanh for bounded output)
+                # If no connections and no bias applied, this will be tanh(0) = 0
                 activation = np.tanh(weighted_sum)
                 layer_activations.append(activation)
 
@@ -451,7 +479,7 @@ class CellBrain(BehavioralModel):
         super().__init__()
 
         # Define input and output keys
-        self.input_keys = ['distance', 'angle']
+        self.input_keys = ['distance', 'angle', 'current_speed', 'current_angular_velocity']
         self.output_keys = ['linear_acceleration', 'angular_acceleration']
 
         # Initialize inputs and outputs
@@ -461,7 +489,9 @@ class CellBrain(BehavioralModel):
         # Set input ranges for normalization
         default_ranges = {
             'distance': (0, 50),
-            'angle': (-180, 180)
+            'angle': (-180, 180),
+            'current_speed': (-MAX_VELOCITY, MAX_VELOCITY),
+            'current_angular_velocity': (-MAX_ROTATIONAL_VELOCITY, MAX_ROTATIONAL_VELOCITY)
         }
         self.input_ranges = input_ranges if input_ranges is not None else default_ranges
 

world/objects.py

@@ -322,6 +322,8 @@ class DefaultCell(BaseEntity):
         input_data = {
             "distance": distance_to_food,
             "angle": angle_between_food,
+            "current_speed": math.sqrt(self.velocity[0] ** 2 + self.velocity[1] ** 2),
+            "current_angular_velocity": self.rotational_velocity,
         }
 
         output_data = self.behavioral_model.tick(input_data)
@@ -384,7 +386,7 @@ class DefaultCell(BaseEntity):
 
         movement_cost = abs(output_data["angular_acceleration"]) + abs(output_data["linear_acceleration"])
 
-        self.energy -= (self.behavioral_model.neural_network.network_cost * 0.01) + 1 + (0.2 * movement_cost)
+        self.energy -= (self.behavioral_model.neural_network.network_cost * 0.01) + 1 + (0.5 * movement_cost)
 
         return self