HNO Helmholtz Scattering Dataset

Dataset associated with the manuscript "FFT-free Neural Operators for Helmholtz Scattering via Adaptive Basis Modulation" by Ju O Kim and Deokwoo Lee (Keimyung University), submitted to Applied Sciences (MDPI), Special Issue on "Physics-Informed Learning: Applications in Physics-Informed Neural Networks and Machine Learning". The dataset contains 2D Helmholtz scattering solutions on a 128x128 grid, generated by a 5-point finite-difference solver with Perfectly Matched Layer (PML) boundary conditions, organized into 4 complexity levels (single scatterer to high-contrast OOD scenarios). Total: ~497 MB across 4 HDF5 files. Each HDF5 file contains the refractive-index field n^2(x) and the complex wave field (real and imaginary parts) for multiple samples at k0 = 20. See README.md for the full schema, Python usage example, and train/test split convention used in the paper.