Algorithms for numerical minimization.
This package provides classes and interfaces for finding a minimum of functions.
To find minimum, you should choose between methods that require only evaluations of the function and
methods that also require evalutaions of first derivative. In multidimensional case derivative is the gradient.
Methods that using derivative are more powerfull, but slowly in calculation.
For one-dimensional case, minimization is done without calculating the derivative of function( bracket
the minimum and then use Brent's routine to find minimum).