On the Convergence Rate of Dual Ascent Methods for Linearly Constrained Convex Minimization | Synapse