Active-particle flows can be defined as streams of interacting self-propelled particles moving collectively. Their properties have typically been studied though agent-based simulations and generalized hydrodynamic equations. The agent-based approach has mainly considered groups of self-propelled particles with velocity-based interactions, where each agent tends to align its velocity to that of its local neighbors. The hydrodynamic approach has focused on unveiling the features of what has been assumed to be a universal linear hydrodynamic description of active systems.

In this work, we show that active-particle flows can be more diverse than anticipated since there are at least two different classes: one occurring when particles have velocity-based interactions and the other when they have position-based interactions. Each displays a distinct phase diagram characterized by different velocity and density field structures. They also show different correlations, waves, and information propagation properties beyond the linear hydrodynamic theory, even when placed in equivalent macroscopic hydrodynamic regimes.