🧪 Quantum Experiment Agent

Hybrid quantum optimization powered by IBM Granite

Design, optimize, validate, and visualize hybrid quantum experiments through a dedicated A2A service.

Warning

This agent is in active development. The currently implemented execution workflow is p=1 QAOA Max-Cut for graphs containing 2–8 nodes. VQE and error-mitigation workflows are planning capabilities and are not yet executed automatically.

What works today

Capability	Result
QAOA Max-Cut	Optimizes a p=1 variational circuit with deterministic grid search
Exact validation	Compares QAOA against the exact classical optimum
Sampling	Produces reproducible shot distributions and top partitions
Experiment Canvas	Shows graph partition, convergence, approximation ratio, and outcomes
Optional QPU evaluation	Sends the optimized final circuit to Quantum Computing Agent through A2A
Research planning	Uses IBM Granite 4.2 8B for VQE, mitigation, and experimental-design requests

Architecture

Quantum Lab Agent :8000
          │ A2A
          ▼
Quantum Experiment Agent :8004
     ├── local QAOA optimizer
     ├── exact Max-Cut validator
     ├── Canvas dashboard
     └── Quantum Computing Agent :8003 ──► IBM Quantum QPU

Quick start

git clone https://github.com/BrUn3y/quantum-experiment-agent.git
cd quantum-experiment-agent
uv sync
ollama pull granite4.2:8b
cp .env.example .env
./start.sh

The A2A agent card is available at:

http://127.0.0.1:8004/.well-known/agent-card.json

Try it

Use QAOA to solve Max-Cut on a 5-node graph using the local simulator

Run QAOA for a 4-node graph with edges (0,1), (1,2), (2,3), (3,0) using the local simulator

Compare a 5-node QAOA Max-Cut baseline with real IBM Quantum hardware

Simulator requests stay local. Explicit hardware comparison requests send the optimized final circuit once to the Computing Agent, which creates a fresh IBM Quantum job.

Test

uv run python -m unittest discover -s tests -v
uv run python -m compileall -q src

Related public agents

Repository	Role
Quantum Lab Agent	Main orchestrator
Quantum Developer Agent	Code generation
Quantum Status Agent	Backend and job monitoring
Quantum Computing Agent	Circuit execution

Roadmap

• Multi-layer and optimizer-pluggable QAOA
• VQE execution for small molecular Hamiltonians
• Error-mitigation comparisons
• Automatic simulator-versus-QPU reports
• Multi-job convergence tracking