Status: Submitted for publication
GitHub: github.com/rylanmalarchick/QubitPulseOpt
Paper: arXiv:2511.12799
Overview
QubitPulseOpt is a professional-grade Python framework for quantum gate optimization using GRAPE (Gradient Ascent Pulse Engineering) with realistic noise modeling. The project demonstrates that high-fidelity quantum gate operations can be achieved through systematic pulse optimization, reaching 99.14% X-gate fidelity in closed-system simulations—a 77× error reduction compared to Gaussian baseline pulses.
Key Results
| Metric | Value |
|---|---|
| X-gate Fidelity (closed system) | 99.14% |
| Gaussian Baseline Fidelity | 33.40% |
| Error Reduction Factor | 77.17× |
| Optimization Time | 12.64s |
| Unit Tests | 822 |
| Code Coverage | 74% |
Project Architecture: Three Pillars
The project has evolved into a Quantum Reinforcement Learning Laboratory with three pillars:
Pillar I: The Brain (RL)
- Gymnasium environment for quantum feedback control
- PPO agent via stable-baselines3
- Autonomous quantum control exploration
Pillar II: The Environment (Physics)
- GRAPE and Krotov optimization algorithms
- Lindblad master equation for T1/T2 decoherence
- Stochastic master equation for trajectory simulation
Pillar III: The Watchdog (Analysis)
- Criticality detection and phase slip analysis
- Filter functions for noise characterization
- Robustness testing and benchmarking
Technical Features
Core Physics Engine
| Module | Description |
|---|---|
| GRAPE Optimizer | ~1250 lines, gradient ascent with line search, momentum, amplitude constraints |
| Krotov's Method | Monotonically convergent alternative to GRAPE |
| Lindblad Master Equation | T1/T2 decoherence, Ramsey experiments, thermal states |
| Stochastic Master Equation | Quantum trajectory simulation for feedback control |
Pulse Library
- Gaussian, DRAG, Blackman, Cosine pulses
- Pulse scaling and area normalization for π-pulses
- Smooth pulse shaping for leakage suppression
Hardware Integration
| Feature | Status |
|---|---|
| IQM Backend Manager | Complete (~933 lines) |
| IQM Translator | Qiskit-to-IQM circuit conversion |
| Async Job Manager | Asynchronous job submission |
| Hardware Characterization | Parameter extraction |
Hardware Connectivity: IQM Garnet 20-qubit superconducting processor (API verified)
Analysis & Visualization
- Bloch sphere animation
- Fidelity dashboards
- Robustness testing
- Filter functions
- Benchmarking suite
- Report generation
Verified Results
Results verified and documented in verified_results/PROVENANCE.md:
| Metric | Value |
|---|---|
| GRAPE X-gate Fidelity | 99.14% |
| Gaussian Baseline | 33.40% |
| Error Reduction | 77.17× |
| Optimization Time | 12.64s |
Important Note: GRAPE optimization performed in closed quantum system (unitary evolution). Open-system GRAPE with decoherence is documented as future work.
Technology Stack
| Category | Technology |
|---|---|
| Quantum Simulation | QuTiP 5.0+, NumPy, SciPy |
| Optimization | GRAPE, Krotov, L-BFGS |
| RL | Gymnasium, stable-baselines3 |
| Hardware | IQM Resonance API |
| Testing | pytest (822 tests), GitHub Actions CI/CD |
Recent Work
- Comprehensive test suite expansion (108+ new tests)
- Hardware integration test suite
- arXiv preprint finalized with verified results
- AI assistance disclosure added for ethical transparency
- Barrier options and convergence analysis
Future Integration
This project integrates with:
- quantum-circuit-optimizer: Circuit compilation
- qco-integration: End-to-end fidelity analysis
- QubitOS: Real-time quantum control kernel
Links
Citation
Malarchick, R. (2025). "GRAPE Pulse Optimization for Quantum Gates
with Hardware-Representative Noise." Submitted. arXiv:2511.12799.