Effects of the nature of the starting population on the properties of Rugate filters designed with the genetic algorithm

The genetic algorithm optimization technique for optical filter design is applied to two starting populations, an inverse Fourier transform population and a random population. The refractive index profiles after convergence, and the transmittance of the filter outside the region of support, are markedly different in the two cases. The Fourier filter has lower sidelobes and fails gracefully outside the region of support whereas the random filter fails catastrophically in this wavelength region. The ripple in the passband is higher for the random filter. Furthermore, the average value of refractive index profile and the excursion in refractive index are much larger for the filter generated with the random starting population. However, most of the drawbacks of the random starting population are eliminated by chromosome manipulation in the spatial frequency domain. The fitter properties after this intervention approximate those of the Fourier filter.