Inherent structures, derived by eliminating thermal fluctuations from complex trajectories, illuminate fundamental mechanisms underlying structural relaxation and dynamic heterogeneity in dense glassy systems. However, determining these structures in hard disk/sphere systems presents unique challenges due to the discontinuous nature of inter-particle potentials and resultant flat potential energy landscapes. To address this limitation, we introduce the Recursive Centroid of Free Area algorithm (ReCFA), a novel approach inspired by a steepest descent method, which computes inherent structure configurations in hard disk systems. We conducted comparative analyses between ReCFA, similar methods, and a conventional time-coarse-graining technique, focusing on string-like hopping motions in supercompressed binary hard disks that emulate supercooled liquid behavior. ReCFA demonstrated notable advantages in capturing entropic contributions. The configurations derived through ReCFA exhibited physically reasonable particle displacements analogous to inherent structures in soft particle systems, effectively identifying hopping motions between metastable basins in jammed states. This ReCFA-based analysis enhances our understanding of relaxation dynamics in highly compressed glassy systems, offering a robust analytical tool for investigating both dynamic and structural characteristics across hard and soft particle systems.

11 pages, 12 figures, to appear in J.Phys.Soc.Jpn