Physics / mechanism
Quantum compilers translate high-level quantum algorithms into hardware-executable gate sequences, managing qubit topology constraints, gate decomposition, and error-mitigation scheduling. The core problem is mapping abstract circuits onto physical connectivity graphs while minimising circuit depth and two-qubit gate count—both directly drive decoherence-limited fidelity. Key parameters: SWAP overhead (typically 3–5× circuit depth inflation on near-term devices), T-gate synthesis cost, and routing time. State of the art includes Qiskit Transpiler, tket (Quantinuum), and BQSKit; research compilers achieve 20–40% depth reduction over naive mapping on 100+ qubit devices via heuristic and ML-guided routing. Compilation time scales poorly with qubit count—exponential for optimal solutions, hence heuristic dominance.
Competitive landscape
Dominant stacks: Qiskit (IBM, open-source, device-specific), tket (Quantinuum, hardware-agnostic, commercially licensed), Pytket extensions, and PennyLane (Xanadu, differentiable/photonic focus). Microsoft Q# targets Majorana/topological hardware with distinct noise assumptions. Adjacent: error correction code compilers (surface code schedulers, lattice surgery compilers) are a distinct but converging layer. Pulse-level compilers (Qiskit Pulse, QubiC) operate below gate abstraction.
Companies using
Connected ideas
Sources
Frontier (open questions)
- To be added.