The problem of non convergence in lumopt forward adjoint simulation

- May 1, 2025 at 3:06 am
  
  fachienchong
  Subscriber
  
  hello
  Firstly, let me introduce the basic setup of my simulation, which involves customizing the z-axis negative direction input structure of the light source and converting it into waveguide mode. The frequency is 3.4 THz. The index of waveguide and design area medium is 3.6, and the index of background air is 1.
  Firstly, reverse engineering requires a forward simulation, which does not converge.
  I made some attempts to adjust all boundaries to metal, but still did not converge. I further reduced the time step to 0.5 but still did not converge. May I ask how I should adjust next? The following is the log of forward simulation, which shows that it has been oscillating at the level of 1e-1
  
  Thank you for your reply
  fachien
- May 6, 2025 at 2:24 pm
  
  Kirill
  Ansys Employee
  
  This topic appears to be a duplicate of 有关lumopt正向伴随仿真不收敛的问题 and will be closed.