bsdf file and actual geometry

edit

The easiest way to use this (and most) BSDFs for daylighting analysis is to apply the BSDF distribution to a plane representing the window in a 3-phase workflow. In this case, you'd use Radiance to create a daylight matrix (relates flux between the window and the sky dome) and a view matrix (relates flux from the window to the calculation point or view pixel), and the BSDF is the third matrix in between those two, managing the transfer of flux through the Lightlouver. You'd place a single polygon just inside the window and trace your rays from that, in each direction. The Radiance tool rfluxmtx handles all of this for you, as long as you set up the window geometry file header(s) correctly. (Incidentally, OpenStudio handles all of this for you automatically and includes a BSDF for Lightlouver as an example daylight redirection device.)

When you do it this way, you can use the 3-phase method to fairly quickly create an annual daylight distribution dataset of building geometry that uses specular redirection devices. This is cool. Note I said this is the easy way. If you want to use BSDFs for glare evaluation or in a 5-phase workflow to calculate ASE, you need to obtain a variable-resolution tensor tree BSDF for your device. Such BSDFs do exist for Lightlouver and other products, but you are on your own using them in an annual simulation context.

I'll leave it to other experts to pick up your question on using the geometry info in the BSDF; there is an example of this in the slide deck you cited, but I have no practical experience with that methodology. That's generally used more for qualitative analysis and glare evaluation of specific views (renderings).