Quartz: Superoptimization of Quantum Circuits
summary
Quartz automatically generates and formally verifies symbolic circuit transformations and uses a cost-based backtracking optimizer to reduce quantum circuits. A generation phase called Rapture algorithm prunes ~15 million candidate circuits down to ~200,000 representatives, then an SMT solver (c3) checks parameterized equivalences. Choosing n=6,q=3 yields ~56,000 transformations in under 30 minutes; n=7 jumps to ~380,000 and ~3 hours. End-to-end results match or outperform prior optimizers on several hardware gate sets; optimizer runs have a 24-hour search timeout.
AI-written summary · about
chapters (11)
- 0:00Start
- 0:15Quantum Circuit Optimization
- 1:18Different Gate Sets
- 2:08Quartz: Superoptimization of Quantum Circuits
- 3:17Quartz Overview
- 3:33Transformation Generator
- 7:59Equivalence Verifier
- 9:20Circuit Optimization Example
- 10:07End-to-end Evaluation
- 10:56How to choose n and q?
- 12:11Conclusion
related talks
24mSynthesizing Quantum-Circuit Optimizers
19mEquality Saturation for Quantum Circuit Optimization
20mSynthetiq: Fast and Versatile Quantum Circuit Synthesis
22mHow to Synthesize Quantum-Circuit Optimizers
19mExtracting maximum value from today's quantum computers while prep
17mQuantum Applications with Classiq
19mQuarl: A Learning-Based Quantum Circuit Optimizer
20mCompilation of Qubit Circuits to Optimized Qutrit Circuits