The essence of OSLO is its optimization capability, which minimizes an error function using user-assigned variables. Although extremely powerful, OSLO optimization is also easy to use, largely because the program is very interactive in its design, and you can easily monitor what is going on during the optimization process.
The highest level of OSLO (OSLO SIX) contains several of optimization methods providing local or global optimization of multi-configuration systems, using derivative-based or non-derivative-based methods. In this document, however, only two optimization methods are described: manual optimization, and damped-least-squares optimization. These are available in all levels of OSLO. Each is very widely used, and suitable for a wide variety of optical design problems.
Optimization in optical design includes three main aspects: variables, error functions, and iteration. Variables are simply the constructional parameters (curvatures, thicknesses, etc.) that are to be changed to optimize the system. An error function (also called a merit function) is a single number that is the weighted sum of squares of quantities that we call operands (also referred to as aberrations or targets). The operands are functions of the variables. Typically the operands might be ray displacements, aberration coefficients, or even system data that are to be adjusted as close as possible to desired values, consistent with performance. Iteration describes the process of changing the values of the variables so that the error function is minimized. Because the error function is a non-linear function of the variables, it is normal to require several cycles of iteration to optimize a system.
Manual, or "hand" optimization refers to the process of changing the values of variables manually, and observing the effects of the changes on performance. Manual optimization doesn't require the construction of an error function. The designer is free to use whatever evaluation methods are relevant. As a simple example of manual optimization, we show the design of a YAG laser that includes an intra-cavity lens.
Although manual iteration was extensively used before the advent of computers, today it is often looked upon as an unsophisticated technique used only by those who don't know much about optical design. In fact, almost the opposite is often the case. The complexity of some of today's optical systems is so great that it is not possible to characterize them by a single number. The wisdom of an experienced optical designer (or often, anyone with common sense!) can sometimes lead to better solutions than can be achieved by the most powerful computers. In actual practice, the best solution is usually the result of a cooperative relationship between a human designer and interactive software.
To study practical optimization, we consider a simple system that might be used to make a laser communications link. This example is typical of the type of problem that can be readily solved manually using catalog lenses together with OSLO's interactive design windows. Alternately, a custom lens for this application could be designed using the damped-least-squares method in OSLO, optimizing either the aberration coefficients or exact ray data. All these methods are described in detail, so you can follow the exact steps using your version of OSLO.
Problem statement and initial setup
Singlet solution - manual optimization
Doublet/meniscus solution - Interactive design window
Air-spaced doublet - Aberration-based SCP optimization
Air-spaced doublet - Ray-based SCP optimization
Modifying an SCP error function
Optimizing in OSLO PRO and OSLO SIX
Copyright © 1997 Sinclair Optics Inc. All rights reserved.
Page last updated 19970601