QuantumSpeedupsofOptimizingApproximately ConvexFunctionswithApplicationstoLogarithmic RegretStochasticConvexBandits