Synthesizing Quantum-Circuit Optimizers
summary
Synthesizes a device-specific quantum-circuit optimizer from a gate-set spec in 1.2 minutes and outputs 7,001 symbolic rewrite rules for IBM’s gate set, giving better two-qubit gate reduction than Qiskit and TKET on about 85% of benchmarks. Represents gate semantics with path-sum algebra, reduces equivalence to polynomial identity testing, and uses a polynomial identity filter plus beam search to extract and apply rules. Reported synthetic failure probability is around 10^-7 and evaluation emphasizes two-qubit gate reductions.
AI-written summary · about
chapters (25)
- 0:00Quantum circuit optimizers intro
- 0:21Quantum computing dream and NISQ
- 1:46Noise and errors in circuits
- 2:50Hardware-specific gate challenges
- 3:55QUESO optimizer introduction
- 5:00Classical vs quantum optimizations
- 5:46Quantum computing fundamentals
- 7:10Simple quantum rewrite rules
- 8:14Complex symbolic rewrite rules
- 8:35Naive optimizer synthesis
- 9:20QUESO polynomial-based approach
- 10:02Feynman path integrals for semantics
- 11:46Symbolic circuits transformation
- 13:16Polynomial equivalence checks
- 15:01Polynomial filter and equivalence
- 16:26Evaluation results and failures
- 17:10Benchmark vs state-of-the-art
- 17:50Comparison with Quartz
- 18:33Future work and enhancements
- 19:17Q&A
- 19:39Evaluation metrics and apps
- 20:19Algorithm optimizations and cases
- 21:02Qubits vs gate count trade-offs
- 22:29Scalability of polynomial step
- 23:12Closing remarks
related talks
22mHow to Synthesize Quantum-Circuit Optimizers
20mSynthetiq: Fast and Versatile Quantum Circuit Synthesis
17mOptimization and Synthesis of Quantum Circuits with Global Gates
19mExtracting maximum value from today's quantum computers while prep
19mEquality Saturation for Quantum Circuit Optimization
15mAssertion-Based Optimization of Quantum Programs
15mModular Component-Based Quantum Circuit Synthesis
19mAn Industry Perspective on the Path Forward in Quantum Optimization