Unsupervised Identification of Coherent Structures: Linking Acceleration Mechanisms to Radiative Signatures in Relativistic Turbulence

Some of the most powerful objects in the universe, such as the jets launched by supermassive black holes, the winds blown off rapidly spinning neutron stars, and the explosions known as gamma-ray bursts, produce enormous amounts of light and accelerate particles to energies far beyond anything achievable on Earth. Understanding the physical processes driving these extreme environments has direct bearing on our ability to interpret astronomical observations, develop next-generation space telescopes and detectors, and train the scientists and engineers who will build them. This award supports research into the fundamental plasma physics underlying these cosmic accelerators and also funds educational programs that engage students at the high school and undergraduate levels, broadening participation in science and technology fields. This award supports a combined simulation and machine learning study of particle acceleration in relativistic magnetized turbulence. State-of-the-art three-dimensional particle-in-cell simulations are used to model turbulent pair plasmas across a systematic parameter space in magnetization and turbulence amplitude, including multiple radiative cooling regimes. A GPU-accelerated unsupervised machine learning tool, aweSOM, is used to identify and catalog the coherent plasma structures, including current sheets, vorticity sheets, and shear-velocity regions, that are responsible for injecting and accelerating particles to non-thermal energies. High-cadence particle tracking is combined with a recently developed energetic particle scattering framework to decompose individual particle energization histories into contributions from distinct physical mechanisms, including parallel electric field acceleration, magnetic mirror reflection, and curvature drift. Synthetic spectra, light curves, and polarization signatures are computed via post-processing with the 3DPol radiative transfer code, enabling direct comparison with multi-wavelength observations of relativistic outflows. Together, these efforts establish a statistically grounded, physically motivated framework linking kinetic-scale plasma processes to macroscopic observables. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2607293 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Colby Haggerty | Institution: University of Hawaii, HONOLULU, HI | Award Amount: $686,392 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2607293 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2607293.html