1: \begin{abstract}

2:     Turbulence in the protoplanetary disks induces collisions between dust

3:     grains, and thus facilitates grain growth. We investigate the two fundamental

4:     assumptions of the turbulence in obtaining grain collisional velocities --

5:     the kinetic energy spectrum and the turbulence autocorrelation time -- in the

6:     context of the turbulence generated by the magneto-rotational instability

7:     (MRI). We carry out numerical simulations of the MRI as well as driven

8:     turbulence, for a range of physical and numerical parameters.

9:     We find that the convergence of the turbulence $\alpha$-parameter does not

10:     necessarily imply the convergence of the energy spectrum.

11:     The MRI turbulence is largely solenoidal, for which

12:     we observe a persistent kinetic energy spectrum of $k^{-4/3}$. The same is

13:     obtained for solenoidal driven turbulence with and without magnetic

14:     field, over more than 1 dex near the dissipation scale. This power-law

15:     slope appears to be converged in terms of numerical

16:     resolution, and to be due to the bottleneck effect.

17:     The kinetic energy in the MRI turbulence peaks 

18:     at the fastest growing mode of the MRI.

19:     In contrast, the magnetic energy peaks at the dissipation scale.

20:     The magnetic energy spectrum in the MRI turbulence 

21:     does not show a clear power-law range, and is almost constant over

22:     approximately 1 dex near the dissipation scale.

23:     The turbulence autocorrelation time is nearly constant at large scales,

24:     limited by the shearing timescale, and shows a power-law drop close to $k^{-1}$

25:     at small scales, with a slope steeper than that of the 

26:     eddy crossing time. The deviation 

27:     from the standard picture of the Kolmogorov turbulence with the injection scale

28:     at the disk scale height can potentially

29:     have a significant impact on the grain collisional velocities.

30: \end{abstract}

31: