10.2.2.2.1. TEgoKBModeler

The TEgoKBModeler has a constructor without argument and 2 user methods:

• setModel defines the model that will be used in optimization. It has 3 parameters: a const char* defining the model to use ("matern7/2" for example); another const char* defining the trend ("const" for example); a double defining a regularization (1.e-8, use 0.0 for no regularization).

• setSolver define how the model will be constructed. It has 4 parameters: a const char* defining the objective to minimize ("ML" for maximum likelihood); a const char * defining the NLopt solver to use ("Bobyqa" for example); an int defining the size of the preliminary screening; an int defining the maximum evaluation number for optimization.

Take a look at The kriging method for details on possible parameter values.