Iterative difference three step method with $1+\sqrt 2$ convergence rate (in Ukrainian)
Ivan Franko National University of Lviv
The new three step method for solving unconstrained minimization problems is proposed. The method use the idea of building of three-step methods and is based on a method with the rate of convergence $1+\sqrt 2$. The rate of convergence for the new method is investigated. Numerical investigation is conducted on the test functions. The result of numerical experiments shows that three step method is more effective in sense of amount of calculations. The efficiency of the method is growing with increasing of function's dimension.
tree step method; minimization problem
1. Vasiljev F.P. Numerical methods for solving extremal problems. M.: Nauka, 1988. 552 p. (in Russian)
2. Pshenichnyj B.N., Danilin Yu.M. Numerical methods in extremal problems. M.: Nauka, 1975. 319 p. (in Russian)
3. Bartish M.Ya., Shcherbyna Yu.M. On a difference method for solving nonlinear operator equations// Dop. AN USSR., ser. ΐ. 1972. Ή7. P. 579582. (in Ukrainian)
4. Beyko I.V., Zinko P.M., Nakonechnyj O.G. Problems methods and algorithms optimization. VGC.: Kyiv. Univ., 2012. 800 p. (in Ukrainian)
5. Bartish M.Ya., Kovalchuk O.V., Ogorodnyk N.P. Three-step methods for soving unconstrained minimization problems// Visnyk Lviv. Univ., ser. Appl. Math. Inf. 2007. Ή13. P. 310. (in Ukrainian)
6. Koko J. A conjugate gradient method with quasi-Newton approximation// Aplicationes mathematicae. 2000. Ή27. P. 153165.
|Full text of paper|
|Table of content of issue|