Collaborative Paper Published: Interpretable Thermal Transport Prediction in Amorphous Graphene

Summary

In this paper, we demonstrate that the thermal conductivity of two-dimensional amorphous graphene can be predicted with high accuracy using persistent homology, a topological data analysis method. By performing an inverse analysis of the regression results, we identify key local atomic-structure features—distorted hexagonal and triangular motifs—that are strongly associated with reduced thermal conductivity. These features spatially correlate with low-frequency localized vibrational modes, supporting a physical interpretation in which they suppress heat transport. This work is a collaborative study with Prof. Minamitani’s group at Osaka University.