Algorithm
As local optimizers a Trust Region Framework, the Nelder Mead Simplex method and a Interpolated Quasi Newton algorithm are implemented. The Trust Region Framework, which is the most modern of those, and the Quasi Newton Method are very fast due to the support of interpolation of primary data. The advantage of the Simplex approach is that no gradient information has to be approximated, which can be expensive if the number of variables gets bigger. Three global optimization techniques, CMA Evolutionary Strategy, a Genetic and a Particle Swarm optimizer, are also available. By using a statistical model CMA Evolutionary Strategy improves it's performance without sacrificing it's global optimization approach. More information about the optimization strategies can be found at the Optimizer - Settings and the Global Algorithm Settings page.
Parameters
You can select the parameters for optimization within the Optimizer - Settings property page. Furthermore, you must specify a bounding interval and a number of samples N for each selected parameter. Thus, the specified interval is divided into N-1 sections. The primary data is interpolated within each section.
Goals
The definition of at least one goal is required. The goals are evaluated while the optimizer is running. The goal value is calculated for each parameter configuration that the optimizer sets during the optimization process. The goal value affects further parameter configurations of the optimization process. Goals can be defined within the Optimizer - Goals property page.
Information
While the optimizer is running, information about the course of the parameter settings and the evaluated goal values are displayed by the Optimizer - Info property page.
How to start the optimizer
The optimizer can be used in two ways:
As standalone optimizer which operates on all
defined simulation
tasks of your model. You can launch the optimizer dialog box of the
standalone optimizer by choosing 
Home:
Simulation 
Optimizer.
As an optimization task which operates on all sub-tasks of itself. You can launch the optimizer dialog box of an optimization task by double clicking on the task tree item, or by choosing Properties&ldots; of the task tree item context menu.
The optimization task is more flexible than the standalone version: By defining a sequence or even a hierarchy of tasks you can realize specialized workflows, such as an optimization nested in a parameter sweep. This could be used to sweep one geometrical parameter of an antenna in the outer parameter sweep task, while optimizing another geometrical parameter or a matching circuit in the inner optimization task.
The standalone optimizer is better suited if you quickly want to set up an optimization for an existing task, since you do not need to move or copy the task.
See also
Optimizer: Settings, Goals, Info, Optimizer - Global Algorithm Settings, Simulation Task Overview