This is where we select and customize the objective function that the optimizer will minimize. Choosing a good objective function for any given task is itself a complex and unresolved research problem, and making a satisfactory choice in any specific circumstance typically requires jugdement, experience, and empirical tests.
By default, a Generator computes an error value as "the square norm between the data and dependent variable treated as continuous curves," i.e. the sum of squared errors between the data and the function. Other methods are available, as discussed in the help files under the keyword MulRunFitter, but let's try the default method first. We will also leave the vertical blue lines, which mark the range of the independent variable over which the error is computed, in their default locations (spanning the entire range of the independent variable).
That's all we have to do about the objective function, at least for this example.
Next step : optimize!
Copyright © 2003 by N.T. Carnevale and M.L. Hines, All Rights Reserved.