3-D Topology Optimization


I tried to use 3D layered version of the topology optimization method available in Lumerical. While program calculates the gradient of the adjoint and forward field, it stays in that process indefinitely. There is also screenshot of it in the link. What could be the reason for this?



Even the provided example file experiences the same problem if parameters in the .lsf script file slightly tweaked.

During this time, I actually wrote my procedure for topology optimization. I realized when there is a mismatch between the sizes (not dimension) of average pooled E-field dataset and topology optimization area grid, gradient calculation becomes problematic. Maybe this could be the reason for the error presented in this post as well.

Kind Regards

Hello @oyesilyu,

This is a problem with the python matrix solve which causes the optimization to end with failed line search. The best way to correct this is to ensure your monitors are exactly aligned with the grid.

Could you provide more information on the second issue you are having.