docNumFunc
|
|
Centre for Vision, Speech & Signal Processing |
Numerical Functions Class Structure and Usage
NumFuncC provides a base class which defines an interface for any numerical function of the form Y=f(X), where Y and X are n and m dimensional vectors respectively. It serves as a base class for many derived classes which are evaluated given an input vector and provide an output vector. Examples of this are models, classifiers, cost functions, probability density function estimators, vector magnitude estimators and feature vector reduction methods.
What all these methods have in common is a method for evaluating the object or function given an input vector. This can be performed using either the Evaluate or operator() member functions. A list of vectors can be evaluated using an overloaded form of the Evaluate member function. In addition, the Jacobian (dY/dX) at X0 can be evaluated by calling Jacobian with parameter X0. Note that if an analytical solution for the Jacobian is not provided for the particular function you are working with then it will be estimated numerically automatically.
Here are some possible examples for their use:
- A constant function NumFuncConstantC is provided which gives the same output Y regardless of the input vector X. This is more as a test and example of how to use the NumFuncC class than any particular usefulness.
- It can be used for implementing an empirical model of a process given that all the parameters are known. If the parameters need to be estimated then a NumModelC would be more appropriate.
|
Author: Robert Crida, Created: 12/6/1998, Generated by DocCreate: March 20, 2001 |