Many fundamental machine learning tasks can be formulated as min-max optimization. This
motivates us to design effective and efficient first-order methods that provably converge to …