Diverging simulation error



I am trying to simulate bullseye structure on silver film and facing divergence simulation error, I tried all option mentioned on the website still facing the same problem. I am attaching the file, please take a look and suggest the possible solution.silver bullseye copy.fsp (273.8 KB)


Hi, @abhishek.dubey !

In your simulation file I changed the PML profile to stabilized, following one of the recommendations from the knowledge base (https://kb.lumerical.com/en/layout_analysis_diverging_simulations.html), and it made the simulation converging.
Also, I increased the number of PML layers to make reflection value lesser than unity.

silver bullseye_converged.fsp (275.5 KB)

Mind that the simulation still needs to be brushed up to obtain quantitative results.
For this, I recommend to explore further the page: https://kb.lumerical.com/en/layout_analysis_diverging_simulations.html

Please, keep us informed here on how your task is going.


Hi @abhishek.dubey,

In addition to @msaygin post (thanks a lot for your contribution! :slight_smile: ), I would have a few comments about your simulations:

  • Material fit: the settings seem a bit odd; the maximum number of coefficient is very high (120), there is no need for such value. Using a large number of coefficient requires more memory and time, and can sometimes lead to instability. Also, the material data stop at 1um, so for higher wavelength, the fit might not correspond to the actual material properties. It would be better to use data that cover the whole wavelength range.
  • PML: the distance between the PML and the structure is rather short, it can possibly affect the simulation. A good starting point would be to leave at least half of the largest wavelength. Then you can also check if changing this distance changes the results.
  • Mesh: you set the mesh accuracy to 6, that’s very high. It sets the mesh to be very fine, but then you added a mesh override region that set a larger mesh. It would be better to keep the mesh accuracy to a smaller value (2 or 3) and use the mesh override region to set a finer mesh where needed, over the rings.
  • Sources: I’m not too sure why you are using 2 sources. Can you eventually explain?

Lastly, you can find information about modifying the material fit here, and information about the parameters that affects the FDTD accuracy and how to test the convergence of the results here.