What do you want to model?  Just the output from it?  Or propagation through it? 

Hi,
My understanding is that you’re looking at using POP. Defining a square shaped input beam then being the central question for this use case.
You actually have several options to achieve this, using different POP input beams definition options:

• using the “file” option, either text-based or binary using for instance a Python Script. Sample codes for the latter option is available here: Python Reading Writing Binary Files (ZRD, ZBF, DAT, SDF) | Zemax Community
• using the “file” option, with another approach. You could use the available images and convert this image to a zbf file as illustrated here: Python Converting an image into a ZBF file | Zemax Community
• using the “DLL” option, to analytically define the desired shape. The benefit of using this approach would be flexibility as you could define your own parameters such a the square size, and why not the BPP metric from the manufacturer. I gave a quick look at the attached datasheet and it seems that the BPP combines the beam waist * divergence. For a typical Gaussian beam those parameters sound redundant but for a square beam, it might not be the case. I’d double check this with the manufacturer.

I hope this helps!