Solid-angle nearest-neighbor method for size-disperse systems of spheres

Identifying nearest neighbors accurately is essential in particle-based simulations, from analyzing local structure to detecting phase transitions. While parameter-free methods, such as Voronoi tessellation and the solid-angle nearest-neighbor (SANN) algorithm, are effective in monodisperse systems, they become less reliable in mixtures with large size disparities. We introduce SANNR, a generalization of SANN that incorporates particle radii into the solid-angle criterion for robust, size-sensitive neighbor detection. We compare SANNR against Voronoi, Laguerre, and SANN in binary and size-disperse sphere mixtures. Using Wasserstein distance metrics, we show that SANNR closely matches size-aware Laguerre tessellation while preserving the geometric continuity of SANN. Applied to the crystallization of the complex AB13 phase, SANNR improves detection of local bond-orientational order and better captures the emergence of global symmetry. SANNR, thus, offers a smooth, parameter-free, and extensible framework for neighbor detection in polydisperse and multicomponent systems.