MarQSim: Reconciling Determinism and Randomness in Compiler Optimization for Quantum(…)
summary
MarQSim builds a Markov-chain HTT graph of Hamiltonian terms, turns transition-matrix design into a constrained minimum-cost flow problem that encodes gate-cancellation costs, and produces transition matrices by solving MCF and combining small random perturbations via convex combinations. Evaluation on molecular and ionic Hamiltonians (8–14 qubits) yields ~25% CNOT reduction for the gate-aware matrix and ~27% for the combined scheme versus QDrift, plus an 8.3% reduction in result variance, and faster mixing from spectral tightening.
AI-written summary · about
chapters (8)
- 0:00MarQSim introduction
- 1:49Quantum Hamiltonian background
- 3:54Deterministic vs randomized compilers
- 7:30HTT graph IR introduction
- 10:02Gate cancellation with min-cost flow
- 12:52Convergence via random perturbations
- 13:54Experimental results
- 17:05Q&A
related talks
22mSimuQ: A Framework for Programming Quantum Hamiltonian Simulation with Analog Co...
22mClassical Simulation of Quantum Circuits with Partial Interference Effects
39mBuilding a Compiler for Quantum Computers
24mSynthesizing Quantum-Circuit Optimizers
26mGenerating Compilers for Qubit Mapping and Routing
19mExtracting maximum value from today's quantum computers while prep
19mEquality Saturation for Quantum Circuit Optimization
01h 02mQuantum Computing Concepts