On shape optimization of optical waveguides using inverse problem techniques

Abstract

Optical waveguides are the basis of the optoelectronics and telecommunications industry. These comprise optical fibres and the integrated optical components which manipulate, filter and dispatch incoming optical signals. A taper is a generic kind of optical waveguide with a cross section that varies continuously along its length z. Tapers are used to couple light from a waveguide into another with different cross sectional profile. It is well known that the power lost through the taper side walls decreases for increasing taper lengths. For practical reasons however, it is desirable to keep the taper length as short as possible. The aim of this study is to develop a formulation to minimize the power loss by varying the taper profile of a given fixed length. It turns out that this shape optimization problem exhibits ill-posed behaviour which would slow down the convergence of traditional optimization routines. We show how these problems can be overcome by reformulating the shape optimization problem as a nonlinear inverse problem, which can then be solved using established inverse problem regularization techniques. Numerical results presented here show that this new approach can lead to robust optimization algorithms less sensitive to large discretization refinements.

Efficient nonadiabatic planar waveguide tapers

“…In [6], the coupling between a fiber and a silica ridge waveguide is improved by 2.6 dB using a genetic optimization. A more mathematical approach to optimize the shape of a coupler can be found in [8], where the problem is presented as a nonlinear inverse problem. Also building on inverse problem theory is the optimization of a 3-dB Y-splitter in [9].…”

Holographic planar lightwave circuit for on-chip spectroscopy

“…The solution that we designed consists of an optical circuit that connects the entire device and expands its functionalities. Light is introduced from a single mode fiber to the chip by means of a taper 33 to reduce coupling losses by matching the mode profiles in the horizontal direction. The input light is then split by directional couplers 34,35 according to its spectral band ( Figure 3) and directed through two distinct ridge waveguides (RWGs) to the two holograms, which are working in the separate spectral bands 630-694 nm (436 channels) and 766-850 nm (490 channels) respectively for a total bandwidth of 148 nm and 0.15 nm spectral resolution.…”

Optical Design Challenges of Thermally Assisted Magnetic Recording Heads

“…We developed all of five homemade modeling programs: 1) Optical beam propagation method (optical-BPM) for waveguides, 2) optical finite difference time domain method (optical-FDTD) for the PG, 3) thermal finite difference method (thermal-FDM) for media heat diffusion, 4) electromagnetic field finite element method (magnetic-FEM) for magnetic head field, and 5) micromagnetic-FDM model for TAMR recording process. 0018-9464/$26.00 © 2010 IEEE Furthermore we utilized Rosenbrock's method (RBM) [4] for shape optimization technique used in waveguide and PG modeling [5]. The shape is formed of many free nodes and they modulate in each RBM step, resulting in having different shape, to search better throughput efficiency.…”