Deep-Learning the Landscape

Theoretical physics now firmly resides within an Age wherein new physics, new mathematics and new data coexist in a symbiosis which transcends interdisciplinary boundaries and wherein concepts and developments in one field are evermore rapidly enriching another. String theory has spearheaded this vision for the past few decades and has, perhaps consequently, become a paragon of the theoretical sciences. That she engenders the cross-fertilization between physics and mathematics is without dispute: interactions on an unprecedented scale have commingled fields as diverse as quantum field theory, general relativity, condensed matter physics, algebraic and differential geometry, number theory, representation theory, category theory, etc. With the advent of increasingly powerful computers, from this fruitful dialogue has also arisen a plethora of data, ripe for mathematical experimentation. This emergence of data in some sense began with the incipience of string phenomenology [1] where compactification of the heterotic string on Calabi-Yau threefolds (CY3) was widely believed to hold the ultimate geometric unification.