Data for "Separating Geometry From Interference in Constrained Quantum Optimization"
收藏资源简介:
This repository contains the numerical data, and analysis scripts supporting the manuscript “Separating Geometry From Interference in Constrained Quantum Optimization.” The manuscript develops a framework that separates three components of constrained quantum optimization algorithms: mixer-induced transport geometry on an encoded product space; coherent interference among transported amplitudes; problem-dependent classical maps applied to measured samples. For complete-graph block-(XY) mixers acting on one-hot product spaces, the code implements the exact shell-transfer coefficients, the finite radial recursion, and the normalized stochastic shell kernel derived in the manuscript. It also contains numerical diagnostics comparing different (XY)-mixer geometries and evaluates the preservation of shell-resolved transport under product-formula implementations of the complete-graph mixer. The archived material supports the following results and figures: evaluation of the one-block complete-graph (XY)-mixer amplitudes; construction of the shell-transfer coefficients (T_{r,t}(\beta)); iteration of the exact radial path-mass recursion; construction of the normalized radial Markov kernel; evaluation of radial drift, variance, and stationary shell distributions; comparison of complete, cycle, path, star, and sparse (XY)-mixer geometries; calculation of one-site absolute transfer factors; calculation of Hamming-shell and geometry-refined lumpability defects; generation of shell-sampling profiles; simulation of Trotterized complete-graph (XY) mixers; calculation of local quotient defects and shell-kernel errors; evaluation of certified Trotter step counts and logical exchange depths; generation of the numerical data used in the manuscript figures and tables. The repository is intended to make the numerical results reproducible and to provide a reusable implementation of the shell-transport formalism for constrained quantum optimization on product spaces. The manuscript should be cited together with this archive when the code or data are used.



