Processed Aerodynamic Surface-Field Learning Data and Evaluation Code for Physics-Guided Orthogonal Active-Subspace Learning

Processed aerodynamic surface-field learning data, fixed outer and repeat-specific inner partitions, aligned pressure and signed-friction predictions, panel-level figure-source tables, and NumPy evaluation code supporting a physics-guided orthogonal active-subspace study. The record contains 4207 parameterized geometry-flow cases and the exact partitions and archived predictions required to reproduce the reported evaluation metrics. Original solver case files, solver executables, model checkpoints, and external NASA validation data are not redistributed.