46334

Автор(ы): 

Автор(ов): 

2

Параметры публикации

Тип публикации: 

Пленарный доклад

Название: 

Newton method with adaptive step-size for under-determined systems of equations

Наименование конференции: 

  • 8th International Conference on Optimization Methods and Applications “OPTIMIZATION AND APPLICATINS” (OPTIMA-2017)

Наименование источника: 

  • Proceedings of the 8th International Conference on Optimization Methods and Applications “OPTIMIZATION AND APPLICATINS” (OPTIMA-2017)

Город: 

  • Петровац

Издательство: 

  • CEUR-WS

Год издания: 

2017

Страницы: 

474-480 http://ceur-ws.org/Vol-1987/paper68.pdf
Аннотация
Newton method is a well-known tool for solving finite-dimensional systems of equations. Pure Newton-Raphson method has at least quadratic convergence rate, but its convergence radius is often limited. Damped version of Newton method uses smaller step-size with same direction, with larger convergence ball but linear convergence rate. We propose mixed step-size choice strategy, incorporating both quadratic convergence rate and wide (global in some cases) convergence radius. The method can be used in cases of under-determined equations and Banach-space equations. We present a modification of proposed method requiring no a-priori knowledge of problem constants as well. The method may be also used for solving a class of non-convex optimization problems.

Библиографическая ссылка: 

Поляк Б.Т., Тремба А.А. Newton method with adaptive step-size for under-determined systems of equations / Proceedings of the 8th International Conference on Optimization Methods and Applications “OPTIMIZATION AND APPLICATINS” (OPTIMA-2017). Петровац: CEUR-WS, 2017. С. 474-480 http://ceur-ws.org/Vol-1987/paper68.pdf.