Paper Search Console

Home Search Page About Contact

Journal Title

Title of Journal: J Optim Theory Appl

Search In Journal Title:

Abbravation: Journal of Optimization Theory and Applications

Search In Journal Abbravation:

Publisher

Springer US

Search In Publisher:

DOI

10.1007/s10916-008-9250-9

Search In DOI:

ISSN

1573-2878

Search In ISSN:
Search In Title Of Papers:

Splitting Methods with Variable Metric for Kurdyka

Authors: Pierre Frankel Guillaume Garrigos Juan Peypouquet
Publish Date: 2014/09/16
Volume: 165, Issue: 3, Pages: 874-900
PDF Link

Abstract

We study the convergence of general descent methods applied to a lower semicontinuous and nonconvex function which satisfies the Kurdyka–Łojasiewicz inequality in a Hilbert space We prove that any precompact sequence converges to a critical point of the function and obtain new convergence rates both for the values and the iterates The analysis covers alternating versions of the forward–backward method with variable metric and relative errors As an example a nonsmooth and nonconvex version of the Levenberg–Marquardt algorithm is detailedThe authors thank H Attouch for useful remarks They would also like to thank the anonymous reviewer for his careful reading and constructive comments The second and third authors are partly supported by Conicyt Anillo Project ACT1106 ECOSConicyt Project C13E03 and Millenium Nucleus ICM/FIC P10024F The third author is also partly supported by FONDECYT Grant 1140829 and Basal Project CMM Universidad de ChileFor next results we introduce the following auxiliary property automatically fulfilled under the hypotheses of Theorems 31 and 32 which includes a stability of the sequence xk kin mathbb N with respect to the point x along with a sufficiently close initializationLet mathbf H 1 mathbf H 2 mathbf H 3 and mathbf Sxdelta rho hold Then xkin underlineGamma eta xrho for all k the sequence has finite length and converges to some overlinex Moreover liminf limits krightarrow infty Vert partial fxk Vert =0 and foverlinexle lim limits krightarrow infty fxk=fxCapture convergence and finite length follow from Lemma 72 and mathbf H 3 Next since b knotin ell 1 and sum k=1infty b k+1Vert partial fxk Vert le sum k=1infty Vert xk+1xkVert +sum k=1infty varepsilon k+1infty we obtain liminf krightarrow infty Vert partial fxk Vert =0 Finally observe that that lim krightarrow infty fxk exists because fxk is decreasing and bounded from below by fx and the lower semicontinuity of f implies foverlinexle lim krightarrow infty fxk If lim krightarrow infty fxk=beta fx the KŁ inequality and the fact that varphi is decreasing imply varphi beta fxVert partial fxk Vert ge varphi fxkfxVert partial fxk Vert ge 1 for all kin mathbb N which is impossible because liminf krightarrow infty Vert partial fxk Vert =0 whence beta = fx square Let xn krightarrow x with fxn krightarrow fx as krightarrow infty Since fxk is nonincreasing and admits a limit point we deduce that fxkdownarrow fx In particular we have fxle fxk for all kin mathbb N The function f satisfies the KŁ inequality on Gamma eta xdelta with desingularising function varphi Let K 0in mathbb N be sufficiently large so that fxKfxmin eta underlineadelta 2 and pick rho 0 such that fxKfxunderlineadelta rho 2Now we just need to check that the hypotheses mathbf H 3 are satisfied with our hypotheses on alpha k beta k and mu k Clearly mathbf H 3i holds since we have supposed that alpha k ge underlinealpha sigma fracsqrtpp +1 fracLrho Then mathbf H 3ii asks that b k notin ell 1 which is equivalent to frac1beta k +L notin ell 1 in our context This holds since we have supposed that frac1beta k notin ell 1 Hypothesis mathbf H 3iii is satisfied because fracbeta kalpha k+1 is supposed to be bounded Finally mathbf H 3iv asks the summability of fracbeta k mu kbeta k +L which is bounded by mu k in ell 1 square


Keywords:

References


.
Search In Abstract Of Papers:
Other Papers In This Journal:

  1. Incrementally Updated Gradient Methods for Constrained and Regularized Optimization
  2. Periodic Image Trajectories in Earth–Moon Space
  3. On Approximate KKT Condition and its Extension to Continuous Variational Inequalities
  4. A Note on Testing Axioms of Revealed Preference
  5. Duality Theorem for a Three-Phase Partition Problem
  6. Normality and Nondegeneracy for Optimal Control Problems with State Constraints
  7. Optimal Control for a Groundwater Pollution Ruled by a Convection–Diffusion–Reaction Problem
  8. Stability and Controllability Issues in Mathematical Modeling of the Intensive Treatment of Leukemia
  9. The Existence of Optimal Controls for Problems Defined on Time Scales
  10. Annealing a Genetic Algorithm for Constrained Optimization
  11. Convergent Bounds for Stochastic Programs with Expected Value Constraints
  12. Calmness and Exact Penalization in Constrained Scalar Set-Valued Optimization
  13. On Estimation in Interception Endgames
  14. Studies on Common Solutions of a Variational Inequality and a Fixed-Point Problem
  15. Closedness of the Solution Map in Quasivariational Inequalities of Ky Fan Type
  16. Distributed Stochastic Subgradient Projection Algorithms for Convex Optimization
  17. Harvesting Control for an Age-Structured Population in a Multilayered Habitat
  18. On Optimization over the Efficient Set in Linear Multicriteria Programming
  19. Some Existence Results of Solutions for General Variational Inequalities
  20. An Algorithm for Portfolio Optimization with Variable Transaction Costs, Part 1: Theory
  21. Sequential gradient-restoration algorithm for optimal control problems with bounded state
  22. Aerial Vehicle Search-Path Optimization: A Novel Method for Emergency Operations
  23. Generalized Qualification and Qualification Levels for Spectral Regularization Methods
  24. Optimality Conditions for Disjunctive Optimization in Reflexive Banach Spaces
  25. Global Optimality Conditions for Optimal Control Problems with Functions of A.D. Alexandrov
  26. Time-Optimal Control Problem for a Special Class of Control Systems: Optimal Controls and Approximation in the Sense of Time Optimality
  27. Directional Hölder Metric Regularity
  28. Existence and Uniqueness for a Linear Mixed Variational Inequality Arising in Electrical Circuits with Transistors
  29. The Effects of Adding Input Redundancies in Linear Quadratic Regulator Problems
  30. On Iterative Methods with Accelerated Convergence for Solving Systems of Nonlinear Equations
  31. Synchronization Criterion for Lur’e Systems via Delayed PD Controller
  32. An Optimal Strategy for Pairs Trading Under Geometric Brownian Motions
  33. On an Inequality of C. Sundberg: A Computational Investigation via Nonlinear Programming
  34. Convergence Analysis of Weighted Difference Approximations on Piecewise Uniform Grids to a Class of Singularly Perturbed Functional Differential Equations
  35. Social Interactions within a Dynamic Competitive Economy
  36. Game Analysis of Investment in a Group with Stickiness
  37. SI “Deterministic and Stochastic Variational Principles and Applications”. December 2015
  38. A New Approach to the Proximal Point Method: Convergence on General Riemannian Manifolds
  39. Infinite Dimensional Duality Theory Applied to Investment Strategies in Environmental Policy
  40. Approximate Functions in a Problem of Sets Separation
  41. Feasible Method for Generalized Semi-Infinite Programming
  42. On the Convergence of Adaptive Stochastic Search Methods for Constrained and Multi-objective Black-Box Optimization
  43. Stability of Solutions to Hamilton–Jacobi Equations Under State Constraints
  44. On Regularity for Constrained Extremum Problems. Part 2: Necessary Optimality Conditions
  45. A Generalization of the Classical α BB Convex Underestimation via Diagonal and Nondiagonal Quadratic Terms
  46. Mathematical Programs with Vanishing Constraints: Optimality Conditions, Sensitivity, and a Relaxation Method
  47. On the Finite Convergence of a Projected Cutter Method

Search Result: