1: \begin{abstract}

2: We numerically investigate the statement that local random quantum circuits acting on $n$ qubits composed of polynomially many nearest neighbour two-qubit gates form an approximate unitary $\poly(n)$-design [F.G.S.L. Brand\~ao et al., arXiv:1208.0692]. Using a group theory formalism, spectral gaps that give a ratio of convergence to a given $t$-design are evaluated for a different number of qubits $n$ (up to $20$) and degrees $t$ ($t=2,3,4$ and $5$), improving previously known results for $n=2$ in the case of $t=2$ and $3$. Their values lead to a conclusion that the previously used lower bound that bounds spectral gaps values may give very little information about the real situation and in most cases, only tells that a gap is closed. We compare our results to the another lower bounding technique, again showing that its results may not be tight. 

3: %This implies that the size of a circuit that forms an $\varepsilon$-approximate $t$-design, calculated according to the bounds, in principle, is very different from %that predicted according to the convergence of the values of spectral gaps.

4: \end{abstract}

5: