Method for nonlinear transfer function approximation

The application of the genetic algorithm to the approximation of nonlinear transfer functions is considered. It is shown that the GA approximation method gives better accuracy than the classical Chebyshev approximation, which is sometimes considered to be the best available method for the minimax criterion. Other advantages of the proposed method include the ability to carry out a global search for the optimal solution and easy implementation of various approximation criteria.