Evading the Predator: Using Genetic Algorithm to Optimize Shoaling Fish Survival

Description

This project explores the optimization of survival strategies in mosquitofish (Gambusia holbrooki) against a predator using a novel agent-based model (ABM). The model incorporates attraction and repulsion behaviors based on the nearest neighbor, which has proven effective in smaller shoals. The primary aim is to use a genetic algorithm to evolve the behavioral parameters of the fish, enhancing their survival rate in larger shoals. The approach is benchmarked against a random search algorithm for comparative analysis.

Installation

To run this project, ensure you have the following dependencies installed:

• Python 3.x
• NumPy
• Matplotlib

You can install the required packages using pip:

pip install numpy matplotlib

Usage

To use this project, follow these steps:

1. Clone the repository to your local machine:

git clone https://github.com/yourusername/your-repo-name.git

2. Navigate to the project directory:

cd your-repo-name

3. Run the Python scripts:

• For the genetic algorithm simulation with 10 runs:

python "fish abm model with predator. 3D Generations - 10 runs_GA.py"

• For the genetic algorithm simulation with varying shoal size:

python "fish abm model with predator. 3D varying shoal size_GA.py"

• For genetic algorithm standard deviation analysis:

python "vanilla stddev.py"

• For the random search algorithm simulation with 10 runs:

python "fish abm model with predator. 3D Generations - 10 runs_RS.py"

• For the random search algorithm simulation with varying shoal size:

python "fish abm model with predator. 3D varying shoal size_RS.py"

• For random search standard deviation analysis:

python "random search stddev.py"

Project Structure

• fish abm model with predator. 3D Generations - 10 runs_GA.py: Script to run the genetic algorithm for 10 generations.
• fish abm model with predator. 3D varying shoal size_GA.py: Script to run the genetic algorithm with varying shoal sizes.
• vanilla stddev.py: Script for vanilla standard deviation analysis.
• fish abm model with predator. 3D Generations - 10 runs_RS.py: Script to run the random search for 10 generations.
• fish abm model with predator. 3D varying shoal size_RS.py: Script to run the random search with varying shoal sizes.
• random search stddev.py: Script for random search standard deviation analysis.

Features

Model Setup

• Fish Agent: Simulates the behavior of fish using 6 parameters: perception_radius, attraction_dist, repulsion_dist, max_speed, speed_boost, and acc_throttle.
• Predator Agent: Simulates the predator with 5 parameters: PREDATOR_PERCEPTION_RADIUS, PREDATOR_MAX_SPEED, PREDATOR_CATCH_RANGE, PREDATOR_REFLEX, and PREDATOR_VELCITY_SCALE.

Algorithms

• Genetic Algorithm: Optimizes fish behavior parameters to increase survival rates.
• Random Search Algorithm: Benchmarks performance against the genetic algorithm.

Metrics

• Cohesion: Average distance of all fish from their collective center of mass.
• Separation: Average distance between each fish and its nearest neighbor.
• Alignment: Average alignment of each fish with its nearest neighbor.

Results

The project compares the outcomes of the genetic algorithm and random search algorithm, highlighting the differences in fish behaviors and survival rates under different shoal sizes and parameter optimizations.

Contributing

If you would like to contribute to this project, please follow these guidelines:

• Fork the repository.
• Create a new branch with a descriptive name.
• Make your changes and commit them with clear and concise messages.
• Push your changes to your forked repository.
• Create a pull request detailing the changes you have made.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Contact

For questions or inquiries, please me at saqib.majumder01@gmail.com.