I'm working on optical waveguide research in CST Studio Suite and for the current geometry I'm using, I need to detect a scattering parameter change on the order of 10^-9 due to different dielectric permittivity values. The interpolative frequency sweep threshold is set to 10^-9 on all S-parameters as is the adaptive tetrahedral mesh refinement. The solver accuracy is set to 10^-12 with a 3rd order solver. After running the simulation on a server with >120 GB of RAM, I kept seeing matrix factorization errors and the simulation doesn't converge. If anyone has experienced similar issues with high accuracy simulations I would greatly appreciate a second opinion. I can also reply with additional details as needed. I also apologize if this question is overly technical but I did not see a better community as of yet.

Edu ​​​​​​​CST Studio ​​​​​​​