Skip to main content
Search

# Dielectric spectrum inversion

1 related articles

Subscribe to updates
  1. arXiv

    A CNN trained only on synthetic multipole-Padé spectra infers plasmonic poles in one forward pass and transfers to first-principles and experimental dielectric spectra

    The work builds a one-dimensional convolutional neural network trained entirely on synthetic dielectric spectra generated by randomly sampling the analytical multipole-Padé expression, so that it directly inverts a spectrum into pole parameters (energy, broadening, spectral weight) without material-specific data; on synthetic validation spectra, RPA first-principles spectra (Al, Ca, Ti) and experimental spectra (Si, ZnO, Ta, Mo), a single forward pass reconstructs the dominant spectral structure, and an imaginary-part-only input still recovers the pole representation.

You’re all caught up