NanoPhotoNet-MPM Dataset

Copyright © 2026 Islam I. Abdulaal and Omar A. M. Abdelraouf. Dataset artifacts are released under CC BY 4.0. Paper: forthcoming.

This dataset contains the Finite Difference Eigenmode (FDE) tables, measured dispersion-index files, and consolidated vector-field tensors used by NanoPhotoNet-MPM, a physics-informed neural surrogate for modal phase-matched SPDC inverse design in monoclinic NbOCl2 ridge waveguides.

The paired model release is available at DOI 10.57967/hf/9068. The archived code release is available at DOI 10.5281/zenodo.18653064.

---

Files

File	Description
data/NbOCl2x-measured-index.txt	Measured refractive indices for monoclinic NbOCl2 over the modeled wavelength range
data/SiO2-measured-index.txt	Measured refractive indices for the SiO2 substrate
data/neff_775.mat	Effective-index table for pump modes at 775 nm
data/neff_1550.mat	Effective-index table for signal/idler modes at 1550 nm
data/fde_profiles_128x128.npy	Consolidated full-vector electric-field profile tensor
data/fde_profiles_mask.npy	Spatial mask tensor for boundary/crop handling during training

---

Schema

data/fde_profiles_128x128.npy stores a NumPy tensor with shape (280, 6, 16384):

Axis	Meaning
280	Paired waveguide geometry samples
6	Real and imaginary parts of Ex, Ey, and Ez
16384	Flattened 128 x 128 transverse grid

The .mat files contain effective-index grids over swept waveguide widths and heights. The measured-index text files provide dispersion inputs for material interpolation.

---

Loading Example

import numpy as np
import scipy.io as sio

neff_775 = sio.loadmat("data/neff_775.mat")
neff_1550 = sio.loadmat("data/neff_1550.mat")
profiles = np.load("data/fde_profiles_128x128.npy")
mask = np.load("data/fde_profiles_mask.npy")

print("neff_775 keys:", sorted(neff_775.keys()))
print("neff_1550 keys:", sorted(neff_1550.keys()))
print("profiles shape:", profiles.shape)
print("mask shape:", mask.shape)

---

Generation Summary

The dataset was generated with a full-vectorial FDE workflow over ridge-waveguide width and height sweeps from 100 nm to 1000 nm. Pump samples target 775 nm quasi-TE modes, while signal/idler samples target 1550 nm quasi-TM modes. The consolidated tensor is intended for supervised transverse-field training and downstream physics-informed validation.

---

Limitations

The dataset is scoped to the simulated NbOCl2 ridge-waveguide design space and the material-index inputs included here. Extrapolation outside the documented geometry, wavelength, or material range should be validated with an independent electromagnetic solver.

---

Citation

@misc{abdulaal2026nanophotonet_dataset,
  author    = {Islam I. Abdulaal},
  title     = {{NanoPhotoNet-MPM} Dataset},
  year      = {2026},
  publisher = {Hugging Face},
  doi       = {10.57967/hf/9066},
  url       = {https://doi.org/10.57967/hf/9066}
}