Machine-readable data files for the modelling in Figs.1-5, together with supplementary figures. The data consists of 5 folders. All the figures for the main text can be obtained by running “main.py”. The QNM data was generated with COMSOL 5.0

angle_resolved_spectra - This folder contains .npy files with both high and low angle experimental spectra of NPoM, as well as extracted 10, 11 and 20 positions in separate .npy files. It also contains all .py files needed to generate plots with experimental data in the main text and SI.

mapping_and_regression - This folder contains QNM eigenvalues (eigs.csv) for each shape of facet. In each shape’s folder there is also the z-component of the near-fields of the QNMs extracted in the center of the NPoM gap in the folder “Ez”. These are also .csv. The labels for these QNMs were assigned with “label_modes.py” and exist in “labels.yml”. This data can be condensed for each shape to large matrices of data, or “datrices”, using “make_datrices.py”. The files needed to make the web API are generated with “imag_regression_API.py” and “real_regression_API.py”.

mode_reconstruction - This folder relates to Fig. S1. and has .csv of QNM data.

psuedomode - Also .csvs of QNM data.

rounding - similar to 2, with the facet rounding being investigated instead