Asymptotic expression of the linear discrete best ℓpℓp-approximation

Let hphp, 11p>1, where γp∈Rnγp∈Rn with ∥γp∥=o(ap/p)∥γp∥=oap/p.