If the minimum-norm objective is adopted with the non-negative condition, Q(x)=0, as additional constraints, then a convex approximant is obtained; however, numerical tests reveal that the basis functions are continuous but not continuously differentiable in the interior of Q.