A hadamard-stable extension of courant's sequential method for convex extremal problems (Q3697680)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3921353
Language Label Description Also known as
default for all languages
No label defined
    English
    A hadamard-stable extension of courant's sequential method for convex extremal problems
    scientific article; zbMATH DE number 3921353

      Statements

      A hadamard-stable extension of courant's sequential method for convex extremal problems (English)
      0 references
      0 references
      1985
      0 references
      convex extremal problem
      0 references
      exterior penalty method
      0 references
      regularization
      0 references
      method
      0 references
      minimum norm solution
      0 references
      Let X be a closed convex subset of a Hilbert space, Y be a normed space, \(C\subset Y\) be a closed convex cone, \(f: X\to {\mathbb{R}}\) be continuous and convex, and \(g: X\to Y\) be continuous and concave with respect to the partial order on Y induced by C. The author considers the problem \[ (P)\quad \min imize\quad f(x)\quad subject\quad to\quad x\in X,\quad g(x)\in C \] and assumes the existence of a Lagrange multiplier for this problem. Now let \(\{t_ n\}\) be a real positive null sequence, \(p\in [1,+\infty [\) be arbitrary but fixed, and let \[ F_ n(x):=f(x)+t_ n\| x\|^ 2+t_ n^{-p}d(g(x),C)^ p\quad (n=1,2,...) \] where d(\(\cdot,C)\) denotes the distance function of C. The corresponding sequence of programs is now \[ (P_ n)\quad \min imize\quad F_ n(x)\quad subject\quad to\quad x\in X. \] The main result is the following. Let (P) have a solution. Then: (i) for every n, problem \((P_ n)\) has a unique solution \(x_ n\); (ii) the sequence \(\{x_ n\}\) converges strongly to the minimum norm solution \(x_ 0\) of (P); (iii) \(d(g(x_ n),C)=o(t_ n^{1+1/p})\) and \(f(x_ n)=f(x_ 0)+o(t_ n)\) as \(t_ n\to +0\).
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references