Multilayer Perceptron and Chebyshev Polynomials Based Neural Network for Solving Emden–Fowler Type Initial Value Problems

Present study is devoted to solve singular non-linear differential equations and system of singular non-linear differential equations of Emden–Fowler type using the Multilayer perceptron and Chebyshev polynomials based functional link neural network techniques. We are emphasizing on obtaining accurate solutions of such problems with less computation time than the earlier neural network techniques. Comparisons between Multilayer perceptron and Chebyshev polynomials based network in aspects of approximation abilities and computational efficiency are presented with several examples.