QGML: Quantum Geometric Machine Learning

Quantum Geometric Machine Learning (QGML) is a comprehensive framework that combines quantum geometric structures with machine learning. Features dual JAX and PyTorch backends for optimal performance across different computational environments.

Key Features

Quantum Geometric Analysis

• Berry curvature computation and topological invariants
• Quantum metric tensor and geometric loss functions
• Quantum phase transition detection

Dual Backend Support

• JAX Backend: XLA compilation, automatic differentiation, TPU support
• PyTorch Backend: Dynamic graphs, extensive ecosystem, GPU optimization
• Seamless switching: Change backends with single function call

Specialized Applications

• Genomics analysis with chromosomal instability detection
• High-dimensional manifold learning
• Quantum computing algorithm implementation

Production Ready

• Comprehensive testing across both backends
• Extensive documentation and examples
• Performance benchmarks and optimization guides

Quick Start

Basic Usage

import qgml

# Set computational backend
qgml.set_backend("pytorch") # or "jax"

# Create quantum geometric trainer
trainer = qgml.geometry.QuantumGeometryTrainer(
    hilbert_dim=8,
    feature_dim=4,
    backend="auto" # Uses current backend
)

# Analyze quantum geometric properties
analysis = trainer.analyze_complete_quantum_geometry(
    data_points,
    compute_berry_curvature=True,
    compute_chern_numbers=True
)

Backend Comparison

import qgml

# Compare performance across backends
results = qgml.utils.compare_backends(
    data=my_dataset,
    models=["supervised", "geometric"],
    metrics=["speed", "memory", "accuracy"]
)

print(results.summary())

Backend Performance

Feature	JAX Backend	PyTorch Backend
Compilation	XLA (fast)	JIT (moderate)
Memory	Efficient	Standard
GPU/TPU	Excellent	GPU excellent, no TPU
Ecosystem	Scientific	ML/DL focused
Debugging	Functional	Imperative

Documentation

View Live Documentation

• Installation Guide
• Quickstart Tutorial
• API Reference
• Visualization Gallery
• Examples

️ Installation

Basic Installation

pip install qgml

With specific backend

# PyTorch backend (default)
pip install qgml[pytorch]

# JAX backend
pip install qgml[jax]

# Both backends
pip install qgml[full]

Development

git clone https://github.com/jasonlarkin/qgml.git
cd qgml
pip install -e .[dev]

Research Applications

• Genomics: Chromosomal instability analysis
• Physics: Quantum phase transitions and topological states
• Finance: High-dimensional manifold learning for risk analysis
• Quantum Computing: Algorithm design and quantum advantage analysis

Contributing

We welcome contributions! Please see CONTRIBUTING.md.

Citation

@software{qgml2024,
  title={QGML: Quantum Geometric Machine Learning with Dual Backend Support},
  author={Jason Larkin},
  year={2024},
  url={https://github.com/jasonlarkin/qgml}
}

License

MIT License - see LICENSE for details.