The optical and infrared are connected @CCA

Galaxies are conventionally thought of as composites of physically distinct components, making the components separable. One consequence of this assumption is that different wavelength ranges carry little information about each other, since the different components dominate emission at different wavelengths. This common assumption of separability is implicitly embedded in galaxy spectral energy distribution (SED) fitting codes, which are broadly used to derive the physical parameters of galaxies. In this paper, we show that optical spectra can be used to predict near- and mid-infrared photometry to $\chi^2_N \approx 1$, in strong contrast with SED fitting codes. We also demonstrate that derived quantities such as the bolometric luminosity of active galactic nuclei (AGN) can empirically be tightly constrained from the optical, just as dust parameters such as $\mathrm{q_{PAH}}$. We investigate the shortcomings of the separability assumption in SED fitting codes, finding that it leads to biases and imprecision in parameter inference. We construct a tool for determining the most important areas of improvement for SED models.