A Note on a Globally Convergent Newton Method for Solving Monotone Variational Inequalities
نویسندگان
چکیده
R esum e. Il est bien connu que la m ethode de Newton, lorsqu'appliqu ee a un probl eme d'in equation variationnelle fortement monotone, converge localement vers la solution de l'in equation, et que l'ordre de convergence est quadratique. Dans cet article nous mon-trons que la direction de Newton constitue une direction de descente pour un objectif non dii erentiable et non convexe, et ceci m^ eme en l'abscence de l'hypoth ese de monotonie forte. Ce r esultat permet de modiier la m ethode et de la rendre globalement convergente. De plus, sous l'hypoth ese de forte monotonie, les deux m ethodes sont localement equivalentes: il s'ensuit que la m ethode modii ee h erite des propri et es de convergence de la m ethode de Newton: identiication implicite des contraintes actives a la solution (sous l'hypoth ese de stricte compl ementarit e) et ordre de convergence quadratique. Abstract. It is well-known (see Pang and Chan 7]) that Newton's method, applied to strongly monotone variational inequalities, is locally and quadratically convergent. In this paper we show that Newton's method yields a descent direction for a nonconvex, nondiierentiable merit function, even in the abscence of strong monotonicity. This result is then used to modify Newton's method into a globally convergent algorithm by introducing a linesearch strategy. Furthermore, under strong monotonicity (i) the optimal face is attained after a nite number of iterations (ii) the stepsize is eventually xed to the value one, resulting in the usual Newton step. Computational results are presented.
منابع مشابه
A Note on Globally Convergent Newton Method for Strongly Monotone Variational Inequalities
Newton’s method for solving variational inequalities is known to be locally quadratically convergent. By incorporating a line search strategy for the regularized gap function, Taji et al. (Mathematical Programming, 1993) have proposed a modification of a Newton’s method which is globally convergent and whose rate of convergence is quadratic. But the quadratic convergence has been shown only und...
متن کاملA globally convergent descent method for nonsmooth variational inequalities
We propose a descent method via gap functions for solving nonsmooth variational inequalities with a locally Lipschitz operator. Assuming monotone operator (not necessarily strongly monotone) and bounded domain, we show that the method with an Armijo-type line search is globally convergent. Finally, we report some numerical experiments.
متن کاملA hybrid Newton method for solving the variational inequality problem via the D-gap function
The variational inequality problem (VIP) can be reformulated as an unconstrained minimization problem through the D-gap function. It is proved that the D-gap function has bounded level sets for the strongly monotone VIP. A hybrid Newton-type method is proposed for minimizing the D-gap function. Under some conditions, it is shown that the algorithm is globally convergent and locally quadraticall...
متن کاملA Class of Globally Convergent Algorithms for Pseudomonotone Variational Inequalities
We describe a fairly broad class of algorithms for solving variational inequalities, global convergence of which is based on the strategy of generating a hyperplane separating the current iterate from the solution set. The methods are shown to converge under very mild assumptions. Specifically, the problem mapping is only assumed to be continuous and pseudomonotone with respect to at least one ...
متن کاملGlobal convergence of descent processes for solving non strictly monotone variational inequalities
This paper addresses the question of global convergence of descent processes for solving monotone variational inequalities deened on compact subsets of R n. The approach applies to a large class of methods that includes Newton, Jacobi and linearized Jacobi methods as special cases. Furthermore, strict monotonicity of the cost mapping is not required.
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 1987