Method to create lognormal fields differs from references

[zms8gyy949]

While looking into the sourcecode for the creation of Lognormal galaxy mocks, I noticed that the method used is very different from the one in the given references, i.e. the Coles & Jones 1991 paper and the Lognormal generator by Agrawal et al. Specifically, the original gaussian density field is created to have the unbias linear power spectrum P_lin instead of having the power spectrum corresponding to the transformed correlation function ln[1+\xi_lin(r)]. Further, the tranformation into the lognormal field is done using a bias parameter \delta --> const* exp[b_L*\delta] where b_L = 1-b_lin is called the 'Lagrangian bias'.

I was wondering if someone could provide me with some paper or other reference detailing this approach, since clearly it does work. I believe this question also overlaps with hganjoo's issue #678.

Thank you so much for any help!