LayerForge

Multilayer Material Design Dataset Generator

LayerForge is a tool made for generating comprehensive datasets of multilayer material designs and their corresponding optical properties. The project leverages the refractiveindex.info database for material properties and tmm-fast for accurate spectrum computations.

Project Structure

.
├── config.yaml           # Configuration file for materials and simulation parameters
├── requirements.txt      # Python dependencies
├── src/
│   ├── __init__.py
│   ├── material_generator.py  # Functions for generating random material sequences
│   └── spectrum_calculator.py # Functions for computing optical properties
├── tests/               # Unit tests for the project
├── data/                # Directory for storing generated datasets
└── notebooks/          # Jupyter notebooks for analysis and visualization

Setup

1. Create a virtual environment:

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

2. Install dependencies:

pip install -r requirements.txt

Usage

LayerForge utilizes tmm-fast to compute optical properties of multilayer materials. The workflow consists of:

1. Generating random material sequences with specified thicknesses
2. Computing their optical properties (absorption/reflection spectra)
3. Creating paired datasets for supervised learning

Configuration

Edit config.yaml to customize:

• Available materials
• Layer thickness ranges
• Number of layers
• Wavelength ranges
• Dataset parameters

Generating Material Sequences

from src.material_generator import MaterialGenerator

generator = MaterialGenerator()
sequence = generator.generate_sequence()
print(generator.sequence_to_string(sequence))  # Output format: "Au_10+Ag_20+SiO2_15"

Generating Datasets

Generate datasets using the main.py script:

python main.py

This will:

1. Generate random material sequences
2. Calculate their optical properties
3. Save the results in the data/ directory

Running Tests

The project includes comprehensive unit tests to ensure reliability. Run the tests using pytest:

pytest tests/

The tests cover:

• Material sequence generation
• Optical property calculations
• Configuration handling
• Error cases and edge conditions

Dataset Format

The generated dataset consists of pairs:

• X: Sequence of materials and thicknesses (e.g., "Au_10+Ag_20+SiO2_15")
• Y: Dictionary of wavelength-value pairs representing optical properties

License

MIT License