Joint Research Conference
June 24-26, 2014
In design augmentation there is often substantial information about the model. It may be desirable to obtain more model information while simultaneously choosing design points that match some specified value of the response.
Using a physical analogy to design points as charged particles makes it possible to accomplish the above goals using an compound optimization approach that trades statistical efficiency for response matching.
Example scenarios include maximizing and minimizing the predicted response as well as matching a response contour.