Geobacter Simulation

A WebGL-based reaction-diffusion simulation of Geobacter metallireducens bacterial colonies, motivated by the search for life on Mars. Live App Demo

Overview

This application simulates the emergent formation of "leopard spot" patterns observed in the Jezero Crater on Mars. It models the lifecycle of bacteria as they consume iron minerals (Fe³⁺), produce waste (Fe²⁺), reproduce, and migrate towards food sources.

Mars 'Leopard Spots' in Jezero Crater. Credit: NASA/JPL-Caltech/MSSS

The simulation runs entirely on the GPU using GLSL shaders to solve a system of differential equations in real-time.

User Guide

The main canvas shows the bacterial colony evolving in real-time. Hovering over different layers in the top right allows you to inspect individual density fields.

• Red/Rust: Food (Ferric Iron / Fe³⁺).
• Yellow: Waste (Ferrous Iron / Fe²⁺).
• Green: Active bacteria.
• Blue: Dormant bacteria.

Simulation Parameters

• Fe3 Perlin Scale: Controls the scale of the initial food resource pattern.
• Fe3 Perlin Variation: Controls the intensity/variation of the initial food resources.
• Initial Active Seeds: Number of starting bacterial colonies.
• Seed Radius: Size of the initial bacterial colonies.
• Reproduction Rate: How fast bacteria divide.
• Death Rate: How fast bacteria die naturally.
• Fe3 Consumption: How fast bacteria consume resources.
• Gradient Bias: Strength of movement towards food.
• Active->Dormant Rate: Rate at which active bacteria go dormant when food is scarce.
• Dormant->Active Rate: Rate at which dormant bacteria reactivate when food is found.
• Time Step: Simulation speed.
• Bacterial Diffusion: Random movement representing water turbulence.
• Fe2 Diffusion: How fast waste products diffuse.

Technical Details

This project is built with:

• Vite + Svelte: Frontend framework.
• Three.js: WebGL rendering engine.
• GLSL: Shader running a system of ODEs for simulation (src/shaders/compute.frag).

Mathematical Models

The simulation defines spatial density fields for Food ($F_3$), Waste ($F_2$), Active Bacteria ($A$), and Dormant Bacteria ($D$). These fields change over time according to this flow diagram:

The flow diagram can be modeled mathematically with the following system of Ordinary Differential Equations (ODEs):

$$ \begin{aligned} \frac{dF_3}{dt} &= -r_c A F_3 \\ \frac{dF_2}{dt} &= +r_w A F_3 \\ \frac{dA}{dt} &= (r_r - r_d - r_{ad}(F_3)) A + r_{da}(F_3) D \\ \frac{dD}{dt} &= r_{ad}(F_3) A - (r_{da}(F_3) + r_d) D \end{aligned} $$

The stage transition functions scale linearly based on available resources:

$$ \begin{aligned} r_{ad}(F_3) &= \alpha_{ad}\left(1 - \frac{F_3}{K_F}\right) \\ r_{da}(F_3) &= \alpha_{da}\left(\frac{F_3}{K_F}\right) \end{aligned} $$

Project Structure

geobacter-sim/
├── src/
│   ├── components/                 # Svelte UI
│   │   ├── ControlPanel.svelte     # Parameter sliders
│   │   ├── InfoPanel.svelte        # Information section
│   │   ├── LayerControls.svelte    # Toggle the visibility of each layer
│   │   └── SimulationCanvas.svelte # Simulation canvas area
│   │
│   ├── shaders/                    # GLSL simulation code
│   │   ├── compute.frag            # ODE Models, Movement, and Diffusion
│   │   └── display.frag            # Visualization color gradients
│   │
│   └── simulation/                 # Non-GPU simulation code
│       ├── parameters.ts           # Parameter defaults
│       └── Simulation.ts           # GPU setup and lifecycle

Local Development

# Install dependencies
npm install

# Run dev server
npm run dev

# Build for production
npm run build

Future Work

• Custom Layer Colors: Allow users to define their own colors for the different simulation layers.
• Parameter Presets: A preset example buttons that load different initial parameter configurations.

Credits

• reaction-diffusion-playground by Jason Webb
• GPUComputationRenderer (Three.js examples)
• The Book of Shaders (Random functions)