Why is FRED Optimum’s hybrid optimization different than lens design optimization? FRED’s new hybrid general optimization algorithm is non-sequential, allows for multiple targets, has fractional weighting capability to link variables and utilizes several built-in merit functions plus a user-defined scripted merit function for unusual tasks. The hybrid algorithm has the full capability to optimize surfaces created directly in FRED as shown in the figure below or imported from CAD as NURBS. This optimization scheme gives the user complete control over variables, merit functions and optimization algorithms (1D or Downhill Simplex) to solve the toughest illumination design problems.

FRED Optimum’s menus are easy to use: tabbed, built-in spreadsheets that define the parameters used in an optimization. There are three tabbed windows to define variables, the merit functions, and the optimization methods. In the Variables tab the user defines the entities to vary, the parameter of the entity to vary, fractional pickup information to link one variable to another and the current, initial, lower and upper values for this variable. Adding variables to the sheet is as simple as clicking on the bottom blank variable line and adding data.
The merit function spreadsheet uses any of the four available merit functions: RMS Spot Size, Encircled Spot Radius, Total Power on the Surface or a User Scripted function. The unique hybrid optimization algorithms in FRED allow the user to select multiple target surfaces with different weights during optimization. This multiple target functionality creates a multitude of scenarios where the user can optimize on different targets using different criteria as shown in the figure below.

The optimization tab selects between 1D minimization and Downhill Simplex methods as well as convergence criteria, number of iterations and other options.