Process Integration and Design Optimization

Process Integration and Design Optimization

Optimization in Lumerical FDTD ‏

    • Kessem Zamir
      Subscriber
      Dear Sir/ Madam,
      I'm currently working on a periodic structure consisting of two waveguides with grating, in FDTD using Bloch boundary condition.
      I want to do an optimization on the waveguides' and gratings' parameters in order to get some feature in the dispersion relation.
      I would like that my cost function will be a function of the eigenvectors of the transfer matrix of a unit cell of the structure (which depends on the waveguides parameters). I wasn't able to find how to do a simulation with optimization that its cost function is related to the eigenvector. Is there a feasible approach to accomplish this task?
      I appreciate your help,
      Kessem
    • Guilin Sun
      Ansys Employee

      FOM (Figure of Merit) can only be a scalar, one value from the built-in particle swam optimization. eigen vector is a not  a scale factor. So you will need to manuliplate the eigenvector to form a FOM.

      For more than one FOM value, other third tools may have multi-targets algorithms that you might use. Currently Lumerical can use Matlab and Python. You may explore them.

       

      • Kessem Zamir
        Subscriber

        Dear Guilin,

        Thank you for your answer. Indeed I intend to manuliplate the eigenvector so my cost function (FOM) will be a scalar, sorry if it wasn't clear.

        I am encountering difficulties in translating this concept into code. Specifically, I am unsure about the implementation details regarding the extraction of eigenvectors and subsequent optimization within the FDTD framework.

        Could you possibly provide me with an example or point me towards a similar simulation that demonstrates the manipulation of eigenvectors and subsequent optimization? Any guidance or resources would be greatly appreciated.

        Thank you for your assistance.

         

    • Guilin Sun
      Ansys Employee

      I guess you know how to get the eigen vector: https://optics.ansys.com/hc/en-us/articles/360034925793-eig-Script-command

      You will need to get the eigen vector in an analysis group; in "model"; or in Custom FOM if possible: 

      https://optics.ansys.com/hc/en-us/articles/360034922953-Optimization-utility

      examples:

      Inverse design of y-branch

      Inverse design of waveguide crossing

       

Viewing 2 reply threads
  • The topic ‘Optimization in Lumerical FDTD ‏’ is closed to new replies.