Title: Artificial kagome spin ice magnetic phase recognition from the initial magnetization curve

Abstract: Artificial spin ices (ASIs) are designable arrays of interacting nanomagnets that span a wide range of magnetic phases associated with a number of spin lattice models. Here, we demonstrate that the phase of an artificial kagome spin ice can be determined from its initial magnetization curve. As a proof of concept, micromagnetic simulations of these curves were performed starting from representative microstates of different phases of the system. We show that the curves are characterized by phase-specific features in such a way that a pattern recognition algorithm predicts the phase of the initial microstate with good reliability. This achievement represents a new strategy to identify phases in ASIs, easier and more accessible than magnetic imaging techniques normally used for this task.