Constraint Mixers for Optimization Problems on Circuit Model Quantum Computing
summary
Claims that using XY mixers with k-hot constraint preservation and Dick state initialization keeps the QAOA search inside the feasible subspace, reducing infeasible outputs. Implements mixer unitaries via Trotterized exponentials and uses a linear-anneal parameter schedule instead of full parameter optimization. Tests on a portfolio problem (global k-hot, many mixer terms) show degraded performance due to large Trotter error, while a paint‑shop instance (local pairwise constraints) achieves much higher optimal-solution probability with few layers.
AI-written summary · about
chapters (9)
- 0:00Quantum optimization research
- 1:47Industrial vs MaxCut
- 3:31Handling constraints in QAOA
- 6:44Constraint mixers and XY mixers
- 10:23Modified QAOA with XY mixer
- 11:49Case studies: Portfolio & paint shop
- 15:24Implementation challenges
- 17:50Choosing S for XY mixers
- 19:15Conclusion
related talks
52mThe Virtues of the Quantum Approximate Optimization Algorithm
24mOptimal synthesis into fixed XX interactions
01h 02mProduct Mixing, Quantum Separation, and Comfortable Juntas
20mMarQSim: Reconciling Determinism and Randomness in Compiler Optimization for Quantum(…)
28mConcentration bounds for quantum states and limitations on the QAOA from polynomial approximations
20mOptimal compilation of parametrised quantum circuits
27mA Haskell Adiabatic DSL: Solving Classical Optimization Problems on Quantum Hardware
15mAssertion-Based Optimization of Quantum Programs