Rethinking the concept of the paper I' like to conclude as follows:

Research question
-----------------
Reacting to Squire et al. our research question is

How are the dynamo capabilities of an MHD background turbulence modified in comparison
to a corresponding purely hydrodynamic background turbulence?

in particular

Can a mean-field dynamo, observed in the presence of MHD background turbulence, be explained
as a shear-current effect dynamo on the basis of mean-field coefficients measured by the testfield
method?

Approach
--------
As we cannot rely upon a small-scale dynamo, we have to force magnetically
To minimize the number of system parameters, purely magnetic forcing would be preferable.
Using both forcings would be similar to vary Pm - maybe not needed.

As a basic task, kinematic growth rates of mean fields and the mean-field coefficents corresponding
to these kinematic stages must be determined. Then it has to be checked whether the growth rates can be reproduced from
the dispersion relation for the coherent effects, equipped with the measured coefficients. As a backup in case
of failure, the incoherent effects should be considered.

- To get clear exponential growth, the kinematic stage must be long enough, so has to be started with very small
  B amplitudes. 

- Timestep must be chosen conservatively, optimally a convergence test is needed. 

- The order of magnitude of the mean magnetic fields due to leakage has to be checked by accompanying purely magnetic 
  runs with purely magnetic forcing. Non-negligible amplitudes are likely to indicate too long timesteps and need to
  be suppressed. Suppression of k_y=0 forcing wavneumbers is mandatory, while high precision Pi values for box dimensions
  and removal of means in the "0-equations" of TFM could provide additional help.

- TFM can be employed with the same run, but the duration of the kinematic stage might be too short to obtain reliable
  statistics. Hence, it could be favourable to do the main run without TFM and employ TFM in separate runs without main run.

This approach would guarantee that the flavors of the TFM can not differ. Likewise, simultaneous and non-simultaneous
usage of TFM can also not yield different results.

Entering the nonlinear stage of the mean-field dynamo is not necessary for answering the research question, so should be
limited to very few exemplary cases. Two main difficulties:

- The turbulence is no longer homogeneous, so z-dependent mean-field coeffcients had to be considered. As a consequence,
  no analytical dispersion relation is available any longer.

- The flavors of the TFM will in general differ while we have no direct way to determine which is true, if any. 

Given that I see a mean-field growing with moderate shear (S=-0.1) and low Rm (1.6), the non-dynamo cases of Table 1 should be
reconsidered. Problems of my findings:

- low resolution (64^3)
- no clear exponential stage, more like an initial jump.