Authors: Qin Li (New Jersey Institute of Technology), Bo Shen (New Jersey Institute of Technology), Haodi Jiang (Sam Houston State University), Vasyl B. Yurchyshyn (Big Bear Solar Observatory), Taylor Baildon (Big Bear Solar Observatory), Kangwoo Yi (New Jersey Institute of Technology), Wenda Cao (Big Bear Solar Observatory), Haimin Wang (New Jersey Institute of Technology)

We present MEInversionPINN, a Physics-Informed Neural Network (PINN) approach for solving the Milne–Eddington (ME) inversion problem, applied to high-resolution spectropolarimetric observations from BBSO/NIRIS at the Fe I 1.56 µm line (MEInversionPINN). Traditional ME inversion methods, while widely used, are often limited by computational cost, sensitivity to noise, and difficulty capturing profile asymmetries caused by gradients in magnetic and velocity fields. Our MEInversionPINN framework embeds the ME solution into the network, enabling the network to infer magnetic field parameters with improved accuracy, efficiency, and physical consistency. The model performs well across a range of polarization conditions, including weak and asymmetric Stokes profiles. Comparisons with SDO/HMI and Hinode/SP magnetograms yield correlation coefficients exceeding 0.88, validating MEInversionPINN’s reliability. This method paves the way for fast, physics-aware inversions suitable for high-cadence solar data pipelines and next-generation space weather monitoring.