Humans have ma d he glasses for thousands of years, but have yet to understand why the glass transition occurs and how solids are formed. Indeed, the physical mechanism behind their rigidity is an enduring question central to Condensed Matter. The relevance of this problem extends to fields such as Computer Science, Mathematics, and Optimization, as glass formation relates to the geometric question of how complex, rough energy landscapes are explored dynamically. A central component of various theories of the glass transitions is the existence of groups of particles that transition to a new local minimum of the energy, and participate in structural relaxation. These objects are called “local excitations”. Yet, their number and nature have remained an elusive question. To make progress in this field, now methods must be designed to extract excitations at the relevant energy scale, so as to quantify them precisely and test or build new theoretical descriptions.

Here, we introduced an algorithm that extracts the elementary excitations of glasses numerically up to an energy scale never reached before. We measure excitations systematically to characterize their density, geometry, and location and find that these quantities are remarkably predictive of structural relaxation. The geometry of excitations reveals the presence of a dynamical transition, supporting that the latter governs structural relaxation even at low temperatures.

Inviter: Yu-Liang Jin