README   |   Architecture   |   License

AetherOpt

Production-Grade Quantum-Inspired Optimization Platform

AetherOpt is an end-to-end framework and API for mapping hard combinatorial problems to Quadratic Unconstrained Binary Optimization (QUBO) and solving them using classical, quantum-inspired, and quantum solvers.

It is designed with stability, asynchronous execution, and rigorous data validation to serve as a robust backend for production workloads, integrating Cryptography and Data Science pipelines directly into the optimization workflows.

---

Interactive Web Interface

---

Features

• Asynchronous Execution: Safely run long combinatorial solves in the background with a thread-safe SQLAlchemy integration.
• Strict Data Validation: Comprehensive Pydantic models for all problem domains (Portfolio, Max-Cut, Routing, Scheduling) ensuring fail-fast validation. Support for multi-objective & PUBO schemas.
• Multiple Solvers:
  • Simulated Bifurcation: High-performance quantum-inspired solver for dense QUBOs.
  • QAOA (Local): Local quantum circuit simulation for Exact QAOA algorithms.
  • Hybrid Quantum-Classical Pipeline: Seed classical refinement (like SA or HiGHS) with states found by quantum/quantum-inspired solvers.
  • Correlation Reduction: Freeze variables dynamically based on spin correlation to reduce search space.
  • Quantum-Inspired SA: Simulated annealing enhanced with barrier tunneling probability bounds.
  • Classical SA & HiGHS: Fast, traditional baselines and exact MIP solving.
• Cryptography Layer (Secure QUBO): Submit blind optimization jobs. Matrix coefficients are scalar-blinded before reaching the solver to ensure mathematical privacy.
• Data Science Layer: Automated experiment tracking (SQLite/MLflow concepts) to save run configurations, energies, and metadata across experiments.

---

Getting Started

1. Installation

Clone the repository and ensure you have uv installed (if not: pip install uv).

uv sync

2. Run the Server

Start the application with both the frontend and API exposed on port 8000:

uv run aetheropt

3. Interactive Web UI

Visit http://localhost:8000 to access the interactive web interface, where you can define matrices and submit jobs seamlessly.

---

Interactive API Usage

Note

The X-API-Key is configured to secret_key for local development by default. When deploying to production, enforce a secure token by setting the API_KEY environment variable and changing AETHEROPT_ENV=production.

You can also submit jobs directly via curl. Try out this Portfolio Optimization problem!

Show cURL Example

curl -X 'POST' \
  'http://localhost:8000/api/v1/jobs/' \
  -H 'accept: application/json' \
  -H 'X-API-Key: secret_key' \
  -H 'Content-Type: application/json' \
  -d '{
  "problem_type": "portfolio",
  "data": {
    "expected_returns": [0.1, 0.2],
    "covariance_matrix": [[0.04, 0.01], [0.01, 0.05]],
    "k": 1,
    "risk_aversion": 0.5
  },
  "solvers": [
    "simulated_bifurcation",
    "quantum_warmstart",
    "highs"
  ],
  "config": {
    "num_reads": 10,
    "num_steps": 1000,
    "secure": true
  }
}'

After receiving a Job ID, poll the results:

curl -H 'X-API-Key: secret_key' http://localhost:8000/api/v1/results/{YOUR_JOB_ID}

---

Architecture

Curious about how AetherOpt processes jobs and maps domains to QUBOs with crypto layers? Check out the Architecture Overview.

Testing

Execute the comprehensive test suite locally:

uv run pytest