Information Technology Reference
In-Depth Information
the model (5.27) designed with a fixed set of numerical constants. More
formally, the model (5.28a) can be expressed by the following C++ function:
double apsModel(double d[])
{
double dblTemp = 0.0;
dblTemp = (83/(pow((d[0]-d[1]),(24-21))+11));
dblTemp += sqrt(d[2]);
dblTemp += (73/(12+pow(d[2],d[1])));
dblTemp += (((13/((exp(d[0])-d[2])+sqrt(d[2])))-
d[0])+(23-d[2]));
dblTemp += (pow(sqrt(sqrt(sqrt((d[0]*40)))),sqrt(d[2]))/6);
dblTemp += 56;
return dblTemp;
}
(5.28b)
Note how profusely the GEP-RNC algorithm uses the random numerical
constants at its disposal. Note also that, structurally, this program resembles
the model (5.27) designed with the fixed set of numerical constants, and that
both of them are very different from the model (5.26) created without nu-
merical constants.
So, in conclusion, before starting the modeling process, it is advisable to
inquire if numerical constants are essential to design good models. When the
answer is negative, then the simpler system without numerical constants
should be used as it allows a much faster evolution and, consequently, al-
lows the creation of more and better programs in record time. But if the
answer is affirmative, then the GEP-RNC algorithm should be used as it
provides the ideal tools for designing extremely accurate models with nu-
merical constants.
