Supplementto" Sample-EfficientReinforcement LearningforLinearly-ParameterizedMDPs withaGenerativeModel "