Machine Learning from Cosmological Simulations to Identify Distant Galaxy Mergers

I describe efforts to blend cosmological simulations with surveys of distant galaxies. In particular, I will discuss our work to create and interpret millions of synthetic images derived from the Illustris project, a recent large hydrodynamic simulation effort. Recently, we showed that because galaxies assembled so rapidly, distant mergers are more common than the simplest arguments imply. Further, we improved image-based merger diagnostics by training many-dimensional ensemble learning classifiers using the simulated images and known merger events. By applying these results to data from the CANDELS multi-cycle treasury program, we measured a high galaxy merger rate in the early universe in broad agreement with theory, an important test of our cosmological understanding.