QAOA — Unified Quantum Optimization Playground

Summary: A multi-framework QAOA research hub combining Qiskit, CUDA-Q, PennyLane, QSim, Qulacs, Qibo, QuTip, Qudora and classical baselines.

Tech: Qiskit, CUDA-Q, PennyLane, Python, Qulacs, QSim, Qibo, QuTip, Qudora

Tags: quantum, qaoa, optimization, research

Repository

This repository is a single, unified workspace for my long-term QAOA research.
It brings together multiple quantum SDKs, consistent toy cases (e.g., Max-Cut, JSSP), reproducible workflows, and backend comparisons into one coherent structure.
The goal is to provide a clean, scalable platform for experimentation, benchmarking, and teaching QAOA across different frameworks.

Features

• Unified multi-SDK structure (Qiskit, CUDA-Q, PennyLane, Qulacs, QSim)
• Clear toy_cases for Max-Cut and JSSP
• Reproducible workflows for CPU/GPU simulation (will be updated)
• Shared utilities for logging, reproducibility, and experiment design
• Consistent project structure across frameworks
• Ready-to-use notebooks for end-to-end examples

Results

• Verified QAOA pipelines implemented in multiple frameworks
• Clean, comparable structure for running the same problem across SDKs
• Working JSSP and Max-Cut toy cases
• Qiskit 2.x primitives supported (will be updated)
• CUDA-Q and PennyLane backends functional
• Explorer setup for benchmarking and scaling behaviour

Next steps

• Backend comparison benchmarks (performance, depth, stability)
• GPU acceleration workflows for tensor network simulators (will be updated)
• Warm-start variants and advanced mixer designs
• More real-world optimization instances
• Expanded documentation and diagrams
• Automatic visualization of energy landscapes and schedules