Qiskit Qubit Mapping Toolkit

transpiler-plugin
circuit-optimization
graph-theory
Graph-theoretic layout and routing heuristics, benchmarked against LightSABRE
Published

August 17, 2026

STATUS · Qiskit Ecosystem Member · PyPI v0.1.1

What it does

Graph-theoretic initial layout and routing heuristics for mapping logical Qiskit circuits onto hardware coupling maps: IsomorphismLayout and WalkBasedLayout for initial placement, BaselineSwapRouter and LookaheadSwapRouter for routing. Registered as Qiskit transpiler plugins (prefixed qqm_ to avoid colliding with Qiskit’s own built-in plugin names), so they drop into an existing transpile pipeline rather than requiring a separate workflow.

Why graph theory for this problem

Initial layout and routing are graph problems — matching a logical circuit’s interaction graph against a hardware coupling graph, and finding low-cost paths through it when a direct edge isn’t available — and treating them as such (isomorphism search, walk-based cost heuristics) rather than as transpiler-specific special cases is what the underlying math background this portfolio’s other work draws on (graph theory, walk algorithms) is actually good for here.

Benchmarked, not just implemented

Every heuristic here is benchmarked directly against Qiskit’s own production default, LightSABRE (SabreLayout/SabreSwap, Qiskit ≥1.2), rather than presented as a improvement without a baseline comparison. This project’s claim is about where these heuristics land relative to LightSABRE on specific circuit/coupling-map combinations, not a general “faster than Qiskit” claim.