1: \begin{abstract}

2: We propose two novel unbiased estimators of the integral

3: $\int_{[0,1]^{s}}f(u)\du$ for a function $f$, which depend on a smoothness

4: parameter $r\in\mathbb{N}$. The first estimator integrates exactly the

5: polynomials of degrees $p<r$ and  achieves the optimal error $n^{-1/2-r/s}$

6: (where $n$ is the number of evaluations of $f$) when $f$ is $r$ times

7: continuously differentiable. The second estimator  is also optimal in term of convergence rate and has the advantage to be computationally cheaper, but it is restricted to functions that vanish on the boundary of $[0,1]^s$. The construction of the two estimators relies on a combination of cubic stratification and control variates based on numerical derivatives.  We provide numerical evidence that they show good performance even for moderate values of $n$.

8: \end{abstract}

9: