Additional dataset for "Screening for wide-gap, high refractive index materials via frequency dependent dielectric function prediction aided by convolutional autoencoders"

## Contents This repository contains the data presented in the indicated paper. We will try to update this with more entries but this will depend on time, resources and willpower (all on short supply). If you want to contribute to expand the library of dielectric functions, a quick comparison with the PBE results of Grunert, Grossmann and Gross (PhysRevMaterials.8.L122201) which were computed with ``Quantum Espresso`` shows that, in general, in spite of being computed with diferent codes/pseudopotentials/geometries the results overlap well. This indicates that the database could be safely expanded using results from other sources without too strong restrictions (thus avoiding redundant calculations). The repository is divided into several folders - ``ae``: this folder contains the script defining the convolutional autoencoders and the pipeline for the ALIGNN+AE; - ``data``: this forlder contais the data generated in the paper. Essentially is the trace of the dielectric functions computed with both PBE and TASK, sampled in intervals of 0.1 eV in the range of 0 to 20 eV. We present the original data (i.e. transfered to the given mesh with spline interpolation) and a smoothed version for PBE (see details in the paper). - ``models``: the set of ALIGNN+AE parameters which should be loaded using the scripts in ``scripts``; ## Contributing Calculations where checked for convergence and strange results (e.g. clearly wrong dielectric functions) but silent error may have passed through. If you find any issues please let us know so they can be corrected. ## Authors and acknowledgment PB@CFisUC FNOG@CFisUC We acknowledge financing from FCT - Fundação para a Ciência e Tecnologia, I.P. in the framework of the projects UIDB/04564/2020 and UIDP/04564/2020 (with DOI identifiers 10.54499/UIDB/04564/2020 and 10.54499/UIDP/04564/2020, respectively). P.B. also acknowledges financig from FCT via contract 2020.04225.CEECIND. Computer resources provided by the Laboratory for Advanced Computing at University of Coimbra through project 2023.11003.CPCA.A2 and MareNostrum through project 2024.07090.CPCA.A3. This data was originally presented in: https://gitlab.com/pedromglb/cae_dielectric.git