OntheConvergenceofPrior-GuidedZeroth-Order OptimizationAlgorithms