Authors: Mihailo M. Martinovic (University of Arizona), Kristopher G. Klein (University of Arizona), Tereza Durovcova (Charles University, Prague), Benjamin L. Alterman (Southwest Research Institute)

Instabilities described by linear theory characterize an important form of wave-particle interaction in the solar wind. They arise when the plasma is sufficiently far from Local Thermodynamic Equilibrium, with these departures containing a considerable amount of free energy that can be emitted by the particles in the form of unstable wave modes. To diagnose the nature and predict the occurrence of unstable behavior for solar wind plasma between 0.3 and 1 au, we apply the Nyquist instability criterion to bi-Maxwellian fits of ~1.5M proton core, proton beam, and alpha particle Velocity Distribution Functions (VDFs) observed by Helios I and II. The variation of the fraction of unstable intervals with radial distance from the Sun is linear, signaling a gradual decline in the activity of unstable modes, while for solar wind velocity and Coulomb number we obtain more extreme, exponential trends. Instability growth rates demonstrate similar behavior, and significantly decrease with Coulomb number. Even though very accurate, the Nyquist analysis for a data set of this size is very computationally expensive; to expedite the process for future use, we use the same data set to train a Logistic Regression algorithm that predicts if any given set of ion VDFs is stable or unstable with accuracy of over 90%. We also classify the types of the unstable modes in various types of physical and phase spaces, and find that the ion cyclotron instability dominates the collisionally young solar wind, gradually fading as beam-driven and parallel firehose take over as the most abundant unstable modes.