1: \begin{abstract}

2: We show that a very simple randomised algorithm for numerical integration

3: can produce a near optimal rate of convergence for integrals of

4: functions in the $d$-dimensional weighted Korobov space. This

5: algorithm uses a lattice rule with a fixed generating vector and the only

6: random element is the choice of the number of function evaluations.

7: For a given computational budget $n$ of a maximum allowed number of

8: function evaluations, we uniformly pick a prime $p$ in the range $n/2

9: < p \le n$. We show error bounds for the randomised error, which is

10: defined as the worst case expected error, of the form $\calO(n^{-\alpha -

11: 1/2 + \delta})$, with $\delta > 0$, for a Korobov space with smoothness $\alpha > 1/2$ and

12: general weights. The implied constant in the bound is

13: dimension-independent given the usual conditions on the weights. We

14: present an algorithm that can construct suitable generating vectors

15: \emph{offline} ahead of time at cost $\calO(d n^4 / \ln n)$ when

16: the weight parameters defining the Korobov spaces are so-called product

17: weights. For this case, numerical experiments confirm our theory that the

18: new randomised algorithm achieves the near optimal rate of the randomised

19: error.

20: \end{abstract}