Remarks on optimized predictions for the Drell-Yan process