Optimize Your Design for Manufacturing with LightTools (Part 2): Advanced Design with Global Optimization

Optical Solutions Editorial Team

Jun 17, 2024 / 3 min read

In a previous article, we described how you can use LightTools optimization features to help you quickly design illumination applications such as backlights, light guides, and freeform optics.

In this article, learn how you can use the LightTools global optimization engine, introduced in version 2024.03, to explore a broader solution space and benefit from this valuable addition to the local optimization engines. 

In this example, the goal is to focus the two fields onto the receiver. The starting design is shown on the left, and the final design obtained with LightTools global optimization is on the right. | Synopsys

In this example, the goal is to focus the two fields onto the receiver. The starting design is shown on the left, and the final design obtained with LightTools global optimization is on the right.

How Does It Work?

Designers use optimization to improve performance specifications and save time by automating the process of changing model parameters and evaluating the results. LightTools global optimization explores a large solution space to provide an advantageous starting point for local optimization -- saving development time. LightTools global optimization works in three easy steps.

Step 1: Choose your starting point.

Start by creating or opening a system model and defining at least one variable. Pick a merit function -- the position or orientation of an entity, receiver total power, or CIE coordinates, for example.

In optimization, variables are the parameters of a model (e.g., the height, length, or curvature of a component) that you allow LightTools to change in order to improve performance. Most of the input model parameters in LightTools can be used as optimization variables. LightTools provides intelligent defaults for upper and lower bounds on the value of each type of variable; this is useful to define where the global optimizer searches for a solution.

An optimization merit function sets the criteria for performance improvement. Sometimes it is referred to as an error function, since the smaller it gets during optimization, the closer it is to the target value. For example, to build a merit function, you can use:

  • Any input or calculated real number field
  • Data from illuminance and intensity results
  • Individual rays, ray fans, and ray grids
  • User-defined custom performance criteria

LightTools global optimization supports penalty constraints, which enable you to define a single-sided constraint that is evaluated as a part of merit function, such as a lens thickness being optimized to zero or less.

Step 2: Put global optimization to work.

Have LightTools global optimization start its search for designs. The feature can deliver a better starting point for further local optimization and sometimes the one global minimum.

Step 3: Review and refine the results to achieve your final design.

You can modify inputs and rerun the optimization to achieve your desired result. LightTools global optimization may find a new and improved configuration, or it may verify that your original configuration was the best after all.

When Should I Use Global Optimization?

The global optimization engine performs best when the parameter space is well defined. You should consider global optimization when:

  • The merit function contains multiple minima.
  • The dimensionality of the parameter space is high.
  • The parameter space is large and a general starting point to begin local optimization is unknown.

Global optimization is a powerful optimization method, but it is not a comprehensive solution for all system designs. It is not advisable to over-inflate the search space with a vast number of variables and merit functions and turn on global optimization in hopes of finding a good solution.

Instead, it is advisable to use global optimization for systems that have many variables that must be optimized simultaneously. Global optimization may also be beneficial when the merit function contains saddle points that have relative minima along the individual variable directions and instabilities in the joint variable directions.

Learn More About LightTools Global Optimization

LightTools version 2024.03 introduced an innovative global optimization engine designed to help users explore a broader solution space. This is a new optimization method that supplements the capabilities of the local optimization engines and is particularly beneficial when the starting design parameters are uncertain. This new engine can take a poorly defined starting point and transform it into a reasonable design, getting you closer to a good starting point for local optimization and speeding design development.

If you would like to try LightTools, submit a request to set up a 30-day evaluation.

Continue Reading