Classification of Magnetohydrodynamic Simulations Using Wavelet Scattering Transforms

Abstract The complex interplay of magnetohydrodynamics, gravity, and supersonic turbulence in the interstellar medium (ISM) introduces a non-Gaussian structure that can complicate comparison between theory observation. In this paper, we show wavelet scattering transform (WST), combination with linear discriminant analysis (LDA), is sensitive to 2D ISM dust maps. WST-LDA classifies magnetohydrodynamic (MHD) simulations up 97% true positive rate our testbed 8 varying sonic Alfvénic Mach numbers. We present side-by-side two other methods for characterization, reduced (RWST) three-point correlation function (3PCF). also demonstrate 3D-WST-LDA, apply it classification density fields position–position–velocity (PPV) space, where correlations be studied using velocity coherence as proxy. robust common observational artifacts, such striping missing data, while being enough extract net magnetic field direction sub-Alfvénic turbulent fields. include brief effect point-spread functions image pixelization on 2D-WST-LDA applied fields, which informs future goal applying or 3D all-sky maps hydrodynamic parameters interest.