Homework 5: Algorithmic Bias and Agent-based Modeling (EXTRA CREDIT)#

CMSE 201 – Spring 2025#

Part 2: Modeling Traffic Flow with Agent-based Modeling (20 points)#

In this problem, you will be exploring how we can model traffic flow using Agent-based models (and what we can learn from it!). Below, there are three research questions. You must choose one of them to explore for this section. Make sure you answer all of the questions (beginning below the working basic model).

RQ1: How does the density of cars on a road affect the likelihood of traffic jams forming?

RQ2: What is the optimal speed limit that maximizes traffic flow while minimizing congestion?

RQ3: How do different braking behaviors impact overall traffic flow?

import numpy as np
import matplotlib.pyplot as plt

def create_road(length=100, n_cars=20):
    road = np.zeros(length)  # 0 = empty, 1 = car
    car_positions = np.random.choice(length, n_cars, replace=False)
    car_speeds = np.zeros(n_cars)
    
    for pos in car_positions:
        road[pos] = 1
        
    return road, car_positions, car_speeds

def update_car_speed(current_speed, road_ahead, max_speed=5):
    if road_ahead:
        return max(0, current_speed - 1)
    return min(max_speed, current_speed + 1)

def move_cars(road, positions, speeds):
    new_positions = []
    new_speeds = []
    
    for i, (pos, speed) in enumerate(zip(positions, speeds)):
        # Check if there's a car ahead
        road_ahead = any(abs(p - pos) < speed + 1 for p in positions if p > pos)
        
        # Update speed and position
        new_speed = update_car_speed(speed, road_ahead)
        new_pos = (pos + new_speed) % len(road)
        
        new_positions.append(new_pos)
        new_speeds.append(new_speed)
    
    # Update road state
    new_road = np.zeros_like(road)
    for pos in new_positions:
        new_road[int(pos)] = 1
        
    return new_road, np.array(new_positions), np.array(new_speeds)

# Initialize the simulation
road_length = 50  # Smaller road for easier visualization
num_cars = 10
road, positions, speeds = create_road(road_length, num_cars)

# Function to run simulation for n steps
def run_simulation(n_steps=20):
    current_road = road
    current_positions = positions
    current_speeds = speeds
    
    for step in range(n_steps):
        # Display current state
        print(f"Step {step}")
        print(''.join(['X' if x == 1 else '_' for x in current_road]))
        
        # Update simulation
        current_road, current_positions, current_speeds = move_cars(
            current_road, current_positions, current_speeds
        )
        
        # Optional: add a small delay to see the animation
        plt.pause(0.5)

# Run the simulation
run_simulation()

Step 0
_____________X__X_XX_____X____X___X_X____X___X____
Step 1
______________X__X_XX_____X____X___X_X____X___X___
Step 2
________________X__X__X_____X____X___X_X____X___X_
Step 3
_X_________________XX_X__X_____X____X_X___X____X__
Step 4
_X___X_______________XX_X____X_____X__X_X_____X___
Step 5
_X__X_____X___________XX___X______X___XX___X______
Step 6
_____X__X______X______X__X_____X______X_XX_____X__
Step 7
__X_____X____X______X__X____X_______X____X__X_____
Step 8
_______X____X_____X_____XX______X_______X__X____X_
Step 9
___X_______X_____X_____X___XX________X_____X__X___
Step 10
X_______X_______X_____X____X_X__X_________X__X____
Step 11
_____X_______X_______X____X___X_X____X________X_X_
Step 12
__X_______X_______X______X___X__X___X_____X______X
Step 13
___X___X_______X_______X____X__X___X_____X_____X__
Step 14
__X___X_____X_______X______X__X___X____X______X___
Step 15
_X____X___X______X_______X____X__X____X_____X_____
Step 16
_____X___X_____X______X______X__X____X_____X_____X
Step 17
____X___X____X______X______X____X__X______X_____X_
Step 18
___X____X___X_____X______X_____X__X____X_______X__
Step 19
__X____X___X_____X_____X______X___X__X______X_____

# put any code here

Part 3: Making your Own (simple) Agent-based Model (25 points)#

You now have two examples of Agent-based models. Your task in this section is to develop your own Agent-based model that is different from the traffic flow model and the forest fire (though feel free to re-use code!). There is an activity on HiTA to help you navigate this problem! Navigate to HiTA and select the Activity Labeled “Homework 5 (optional): Part 3. You can also try this direct link.

Broadly, you will need to decide a context (e.g. disease spread, game of life) for your model. You will then do your best to code the basic model and get it to run for multiple time steps. This is a challenging problem that also leaves space for creativity! To get you started, here are the approximate steps for what you will need to do for your idea:

1. Decide what kind of model you are going to try to code (disease spread, game of life, ants, etc.). If you are having trouble coming up with an idea to try, go back to NetLogo for some additional ideas!

2. Define how you will set up your board (how big should it be? how will you define the initial state?)

3. Visualize your initial set up

4. Determine the rules for advancing your board (feel free to use OnBoard() and getNeighborValues from the Forest-fire model directly!)

5. Determine how long your model should run

6. Demonstrate what your model does over multiple time steps

Note: The grading is as follows for this problem

• 25 out of 25 points for a model that runs and includes an implementation for all of the above steps

• 20 out of 25 points for a model that runs but is missing one of the steps

• 10 out of 25 points for a model that runs but is missing multiple steps

• 5 out of 25 points for an attempt that does not run and/or has multiple bugs

# put any code here (feel free to make additional cells!)

✎ Put any explanatory text here!