Generating Compilers for Qubit Mapping and Routing
summary
A system generates qubit mapping and routing compilers from a compact specification and a device graph, using an Amara language frontend and a parametric solver backend that runs simulated annealing over a greedy scheduling template. Evaluation covers seven device models and 243 circuits; generated compilers match IBM Qiskit swap counts in roughly half the benchmarks, improve estimated success probability in about half, and beat a baseline for a proposed neutral atom + surface code design in 93% of cases, with a Google use case reported.
AI-written summary · about
chapters (15)
- 0:00Quantum compiler generation intro
- 0:25Quantum hardware landscape
- 1:48Quantum compilers importance
- 3:55Automating compiler development
- 4:41Qubit mapping challenges
- 7:33Mapping and routing unification
- 8:39QPU state machines
- 11:53Amara language
- 14:25Amara program semantics
- 15:29Solver design with annealing
- 17:15Problem properties analysis
- 18:40Benchmarks and evaluation
- 20:51Handcrafted vs solver compilers
- 21:53Future solver improvements
- 22:58Q&A
related talks
14mDependency-Aware Compilation for Surface Code Quantum Architectures
01h 10mThe Rise & Collapse of a Quantum State
39mBuilding a Compiler for Quantum Computers
46mUnlocking quantum potential with software and compilers
22mSimuQ: A Framework for Programming Quantum Hamiltonian Simulation with Analog Co...
17mQGAT: A Generate-and-Test Paradigm for Quantum Circuits
20mCompilation of Qubit Circuits to Optimized Qutrit Circuits
15mCompositional Quantum Control Flow with Efficient Compilation in Qunity