TheoreticallyBetterandNumericallyFaster DistributedOptimizationwith Smoothness-AwareQuantizationTechniques