Design of Photonic Crystal Cavities by Genetic Algorithms and Numerical Optimization Techniques:

This was my Senior Thesis, done in my Junior and Senior year in Professor Jelena Vuckovic's research group. The group's webpage can be accessed here. I investigated various algorithms for one-dimensional Photonic Crystal (DBR stack) design, such as Genetic Algorithms, Gradient Descent methods and Convex Optimization routines.

We found that Genetic Algorithms allowed us a great deal of flexibility in performing photonic crystal design, and worked well in both simple and more complex problems. The figures below show how our Genetic Algorithm successfully found an optimal solution in a conventionally "difficult" mutipeak optimization landscape.

Fig 1. Graph of true optimization profile. The central peak is higher (but has a smaller width) than the rest (barely visible to the naked eye).

Fig 2. Figures of Genetic Algorithm as it progresses. The black dots represent the individual "organisms", which evolve over the generations to localize on the final "true" maximum on the central peak. This took 50 generations of evolution.

---