Easy-hard phase transition in parameter estimation for optical waveguides.

Area of Science:

• Optics and Photonics
• Computational Physics
• Materials Science

Background:

• Determining optical waveguide parameters, like fiber diameters, typically involves inverse evaluation of scattering intensities.
• The inverse problem is solved by minimizing the mismatch between measured and simulated scattering patterns, forming an energy landscape.

Purpose of the Study:

• To investigate the structure of the energy landscape for optical waveguide parameter determination.
• To analyze how complex refractive indices influence the landscape's topology and computational difficulty.

Main Methods:

• Studied the energy landscape's structure for varying complex refractive indices in 2- and 3-layer optical fibers.
• Analyzed the presence of global vs. local minima in the mismatch function.

Main Results:

• Identified two distinct energy landscape types: one with a global minimum (easy phase) and one with many local minima (hard phase).
• Discovered sharp phase-transition lines separating these easy and hard phases, characterized by a critical exponent.
• Found the critical exponent also describes the number of local minima's dependence on diameters, mirroring combinatorial optimization problems.

Conclusions:

• The study demonstrates the existence of easy-hard transitions in a real-world optimization problem relevant to optical fiber technology.
• Findings provide insights into the computational complexity of determining optical waveguide parameters and suggest a universal behavior in optimization problems.