Dielectric Function of Semiconductors and Insulators under the Independent Particle Approximation

This dataset contains the frequency dependent dielectric function of 9915 semiconductors and insulators calculated using Quantum Espresso and YAMBO under the Independent Particle Approximation. For details, especially the selection of materials and calculation parameters, see [1] .
database_100 contains the spectra calculated using a broadening of 100 meV, database_300 contains the spectra calculated using a broadening of 300 meV.
Each .json contains a ComputedStructureEntry from pymatgen, for details on how to load them check the pymatgen documentation or the code in https://doi.org/10.5281/zenodo.14066644.
The parameters field contains the following parameters (for the remaining parameters, see the documentation of the Alexandria database (https://alexandria.icams.rub.de/)):
pw_conv_k: The converged kppa density in inverse Angstrom for the ground-state calculationpw_conv_cutoff: The converged plane-wave cutoff in Rydberg for the ground-state calculationipa_eps_kppa: The converged kppa density in inverse Angstrom for the IPA calculationipa_nbands: The number of bands used in the IPA calculationipa_broad: The broadening in meV used for the IPA calculationThe data field contains the following parameters:
ipa_epsI_0: The xx-component of the imaginary part of the dielectric function, starting from 0 eV sampled in 10 meV stepsipa_epsR_0: The xx-component of the real part of the dielectric function, starting from 0 eV sampled in 10 meV stepsipa_direct_gap: The direct gap in eV as obtained on the fine grid used for the IPA calculationipa_indirect_gap: The indirect gap in eV as obtained on the fine grid used for the IPA calculationipa_similarity: The SC between the final and the penultimate step of the k-grid convergence for the dielectric function (only for database_100). For more details, see [1].Optionally, ipa_epsI_1, ipa_epsR_1, ipa_epsI_2, ipa_epsR_2: The yy/zz-component of the imaginary/real part of the dielectric function, starting from 0 eV sampled in 10 meV steps. Only present if they differ from ipa_epsI_0 or ipa_epsR_0 due to lack of symmetry[1]: M. Grunert, M. Großmann and E. Runge, Deep learning of spectra: Predicting the dielectric function of semiconductors, Phys. Rev. Materials (2024)