Divergence problem with anisotropic material


Calcit_ReflexionTest.fsp (4.0 MB)
Calcit_Pressling.lsf (1.3 KB)
create_Calcit.lsf (1.7 KB)


I have got a divergence problem which I do not have a clue about.
I need to simulate a reflectance spectrum of a pressed pellet of milled calcite, so I generated a bulk material consisting of little rectangles. To simulate a random orientation of the dielectric tensor function I assign a rotation attribute to every rectangle, and every rotation attribute gets another set of Euler angles. If every rotation attribute gets the same set of angles, everthing is fine, the autoshutoff decreases to the given value. If every rotation attribute gets a different set of Euler angles, the autoshutoff decreases to almost the given value and then suddenly increases, until the simulation stops due to divergence. Why is that? Where is the problem?
Attached are the FDTD fsp file, the Calcit_Pressling.lsf file to generate the bulk material and the create_calcite.lsf file to generate the anisotropic material.
Thanks for any kind of help,


Hi @sonja.hoefer1

I ran the attached simulation file and simulation finished fine. If you are seeing divergence problem, use a DFT monitor to see if divergence happens in the PML region (usually shown by a hot spot/area in the PML). If PML is causing it, try using different PML settings and number of layers.
I also recommend to extend the structures into the PML to avoid unwanted reflection from PML.

Hope this was helpful.


Dear bkhanaliloo,
with different monitors I found that there was some kind of reflection from and in the PML, adding up while the simulation was running. I increased the number of PML layers significantly, now the divergence problem is gone. Does that sound reasonable? Thank you very much for your help.
Best regards,


Hi @sonja.hoefer1

That sounds reasonable to me. You can change the number of layers and PML profile until simulations stops diverging. It is also a good idea to perform convergence testing: