DICS question - documentation is inconsistent

Hello,

Iā€™m trying to understand some ambiguity in the mne.beamformer.make_dics documentation (Version 1.3). Different parts of the documentation for this function differ on the default value for the ā€˜inversionā€™ parameter, and Iā€™d like to know which is the most proper to use.

In the function definition, the inversion argument defaults to ā€˜matrix.ā€™ In the description of parameters, the documentation says:
While inversion=ā€™singleā€™ is more stable, inversion=ā€™matrixā€™ is more precise. See section 5 of [4]. Defaults to ā€˜matrixā€™
! Changed in version 0.21: Default changed to ā€˜matrixā€™

However, if you scroll down to the bottom, you find this:
The DICS beamformer is very similar to the LCMV beamformer and many of the parameters are shared. However, make_dics() and make_lcmv() currently have different defaults for these parameters which were settled on separately through extensive practical use case testing ā€¦
The default setting reproduce the DICS beamformer as described in [4]:
Inversion=ā€™singleā€™ ,weight_norm=None, depth=1

The first description says the default is ā€˜matrixā€™ and references [4], while the second description says the default is ā€˜singleā€™, and also references [4].

From reference 4:
ā€œIn our approach, we treat dipoles with different orientations as separate sources, even if their locations are the same, and consequently compute the beamformer filter for each dipole individually. In this case, L(r)T C-1 L(r) reduces to a scalar value, which avoids having to compute the inverse of another rank deficient matrixā€

Given this conflicting information, Iā€™m unsure what to choose.

Thanks!

Allison