Merging galaxies play a key role in galaxy evolution , and progress in our understanding of galaxy evolution is slowed by the difficulty of making accurate galaxy merger identifications .
We use GADGET-3 hydrodynamical simulations of merging galaxies with the dust radiative transfer code SUNRISE to produce a suite of merging galaxies that span a range of initial conditions .
This includes simulated mergers that are gas poor and gas rich and that have a range of mass ratios ( minor and major ) .
We adapt the simulated images to the specifications of the SDSS imaging survey and develop a merging galaxy classification scheme that is based on this imaging .
We leverage the strengths of seven individual imaging predictors ( Gini , M _ { 20 } , concentration , asymmetry , clumpiness , Sérsic index , and shape asymmetry ) by combining them into one classifier that utilizes Linear Discriminant Analysis .
It outperforms individual imaging predictors in accuracy , precision , and merger observability timescale ( > 2 Gyr for all merger simulations ) .
We find that the classification depends strongly on mass ratio and depends weakly on the gas fraction of the simulated mergers ; asymmetry is more important for the major mergers , while concentration is more important for the minor mergers .
This is a result of the relatively disturbed morphology of major mergers and the steadier growth of stellar bulges during minor mergers .
Since mass ratio has the largest effect on the classification , we create separate classification approaches for minor and major mergers that can be applied to SDSS imaging or adapted for other imaging surveys .