DesignofExperiments forStochasticContextualLinearBandits