Constraining the Milky Way's Ultraviolet to Infrared SED with Gaussian Process Regression

Improving our knowledge of the global properties of the Milky Way (MW) is critical for connecting the detailed measurements only possible within our own galaxy to our understanding of the broader galaxy population. To do this we utilise Gaussian Process Regression (GPR), a method for making predictions from sparsely-sampled, multi-dimensional datasets that can capture both local and large-scale trends. We train GPR models to map from galaxy properties that are well-measured for both the MW and external galaxies to broadband fluxes in a wide variety of photometric bands. The galaxy properties we use to predict photometric characteristics are stellar mass, apparent axis ratio, star formation rate, bulge-to-total ratio, disk scale length, and bar vote fraction. We use these models to estimate the global UV (GALEX $FUV/NUV$), optical (SDSS $ugriz$) and IR (2MASS $JHKs$ and WISE $W1/W2/W3/W4$) photometric properties for the MW as they would be measured from outside, resulting in a full UV-to-IR spectral energy distribution (SED). We show for the first time that the MW must lie in the star-forming regime in standard UV and IR diagnostic diagrams, in contrast to its position in the green valley in optical colour-mass diagrams. This is characteristic of the population of red spiral galaxies, suggesting that the MW may be a member of that class. Although each GPR model only predicts one band at a time, we find that the resulting MW UV-IR SED is in good agreement with SEDs of local spirals with characteristics broadly similar to the MW, suggesting that we can combine the individual band fluxes with confidence. Our MW UV-IR SED will serve as a valuable tool for reconstructing the Milky Way's star formation history via the same tools used for external galaxies, allowing for comparisons of results from $\textit{in situ}$ measurements to those from the methods used for extra-galactic objects.