SpectroTranslator: a deep-neural network algorithm to homogenize spectroscopic parameters

The emergence of large spectroscopic surveys requires homogenising on the same scale the quantities they measure in order to increase their scientific legacy. We developed the SpectroTranslator, a data-driven deep neural network algorithm that can convert spectroscopic parameters from the base of one survey to another. The algorithm also includes a method to estimate the importance that the various parameters play in the conversion from base A to B. As a showcase, we apply the algorithm to transform effective temperature, surface gravity, metallicity, Mg/Fe and los velocity from the base of GALAH into the APOGEE base. We demonstrate the efficiency of the SpectroTranslator algorithm to translate the spectroscopic parameters from one base to another using parameters directly by the survey teams, and are able to achieve a similar performance than previous works that have performed a similar type of conversion but using the full spectrum rather than the spectroscopic parameters, allowing to reduce the computational time, and to use the output of pipelines optimized for each survey. By combining the transformed GALAH catalogue with the APOGEE catalogue, we study the distribution of Fe/H and Mg/Fe across the Galaxy, and we find that the median distribution of both quantities present a vertical asymmetry at large radii. We attribute it to the recent perturbations generated by the passage of a dwarf galaxy across the disc or by the infall of the Large Magellanic Cloud. Although several aspects still need to be refined, in particular how to deal in an optimal manner with regions of the parameter space meagrely populated by stars in the training sample, the SpectroTranslator already shows its capability and promises to play a crucial role in standardizing various spectroscopic surveys onto a unified basis.