1: \begin{abstract}

2: We study the scaling of the convergence of several statistical properties of

3: a recently 

4: introduced random unitary circuit ensemble towards their limits given by

5: the circular unitary ensemble (CUE). Our study includes the full

6: distribution of the absolute square of a matrix element, moments of that

7: distribution up to order eight, as well as correlators containing up to 16

8: matrix elements in a given column of the unitary matrices. Our numerical

9: scaling analysis shows that all of these quantities can be reproduced

10: efficiently, with a number of random gates which scales at most as

11: $n_q\log (n_q/\epsilon)$ with the

12: number of qubits $n_q$ for a given fixed precision $\epsilon$. This suggests

13: that quantities which require an exponentially large number of gates are of

14: more complex nature.

15: 

16: \end{abstract}

17: