A calibration refines a tool application for a specific purposes. The combination of an existing application and a calibration can be dynamically defined as a new application on its own, making a calibrated tool available under a unique name. A calibration is represented as a pre-processor for an application's input parameter vector that maps parameter settings on tool invocation to calibrated settings. In the simplest case, the calibration only supplies values for a set of calibration parameters accessible in the tool that do not depend on the input parameter values (static calibration). I.e., the calibration represents a specific set of default values for a given purpose (cf. Figure 4.16).
Figure 4.16: Static calibration vs. matrix calibration.
In the general case, all input parameter values of the application depend on all parameter settings values; the calibration represents a multiplication of the invocation settings vector with a calibration matrix (matrix calibration). In both cases, the actual calibration parameters and factors can be derived using a procedure outlined in Section 8.8. Instead of a matrix, it is also possible to use a response surface model for mapping input parameters to calibrated parameters if a linear transformation is not sufficient.