BornAgain up to and including version 1.19 generates wrong intensities if either the source or the detector is below the horizon (z<0, alpha<0).
The erroneous intensity diverges with the depth of buried particles. More dangerous is the case of shallow particles as simulations may look unsuspicious.
Many thanks to Karolina Mothander of Lund University for pointing us to this issue.
The bug has been corrected in our ‘develop’ branch, tag ‘repair_transmission’ of 22jul21.
Unfortunately, unrelated other changes in this branch prevent us from releasing a new BornAgain version any soon.
For the time being, we can only advise you to make sure that always alpha_i>0 and alpha_f>0.
If you urgently need to simulate transmission GISAS, or worse, if you already published BornAgain simulations for that case, then please contact us.
There is no issue with BornAgain for SAS; the bug only concerns multi-layer samples under grazing incidence.
Sorry for these bad news, Joachim
July 22, 2021