Ecosystem Simulation Project

An interactive ecosystem simulation with evolutionary behaviors, natural disasters, and environmental dynamics using the Mesa framework.

Overview

This project simulates an ecosystem with:

• Herbivores: Consume vegetation and avoid predators
• Carnivores: Hunt herbivores for sustenance
• Vegetation: Grows and regrows when consumed
• Natural Disasters: Fires and tornadoes that disrupt the environment
• Weather System: Affects vegetation growth rates

The simulation includes evolving agent behaviors through:

• Genetic inheritance
• Reinforcement learning

Features

• Genetic Evolution: Agents inherit and mutate traits affecting movement, foraging efficiency, metabolism, and reproduction
• Reinforcement Learning: Agents can learn optimal behaviors through Q-learning
• Dynamic Environment: Weather patterns and natural disasters create a changing landscape
• Interactive Visualization: Real-time visualization of agent populations and evolving traits

Agent Types

Animals

Animals inherit their traits through a genetic system and can learn optimal behaviors. Each animal has:

• Herbivores

  • Consume vegetation patches for energy
  • Avoid predators based on risk aversion genes
  • Make movement decisions balancing food needs and safety
  • Reproduction when energy reserves are high enough

• Carnivores

  • Hunt and consume herbivores for energy
  • Use foraging efficiency genes to determine hunting success
  • Track prey through the environment
  • Reproduction when energy reserves are high enough

Plants

• Vegetation
  • Fixed agents that grow and regrow at rates affected by weather
  • Provide energy to herbivores when consumed
  • Regrow after a set time period determined by environmental conditions

Disasters

• Fire

  • Spreads through the environment at a defined rate
  • Destroys vegetation in its path
  • Can cause damage to animals
  • Burns out after a set duration

• Tornado

  • Mobile disaster that moves through the environment
  • Can damage or kill animals in its path
  • Destroys vegetation
  • Dissipates after a set duration

Genetic Traits

All animals possess genes that control:

• Movement speed and directedness
• Foraging efficiency and detection range
• Reproduction thresholds
• Metabolism rates
• Risk aversion (particularly important for herbivores)

Requirements

• Python 3.x
• Mesa (Experimental features)
• Solara (for visualization)

Installation

1. Set up a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

2. Install the required packages:

   pip install mesa solara
   

Usage

Copy the "ecosystem_project" folder somewhere.

Run the simulation !from its parent directory! with:

python -m solara run ecosystem_project.app

The web interface will allow you to:

• Adjust simulation parameters (population sizes, mutation rates, etc.)
• Toggle features (weather, disasters)
• Observe population dynamics and evolutionary trends in real-time

Project Structure

• agents.py: Defines all agent types (animals, vegetation, disasters)
• model.py: Contains the main ecosystem model
• reinforcement_learning.py: Q-learning implementation
• app.py: Visualization and UI components

Simulation Parameters

• Grid Size: Controls the environment dimensions
• Population Size: Initial number of herbivores and carnivores
• Reproduction Rates: Probability of reproduction per step
• Energy Values: Energy gained from food and initial energy
• Mutation Rate: Controls how quickly genes evolve
• Evolution Method: Choose between genetic or reinforcement learning approaches
• Weather & Disasters: Toggle and adjust environmental effects