Dissertation > Excellent graduate degree dissertation topics show

Study of the Parallel Variable Distribution Algorithms for Solving Optimization Problems

Author: XuFangFang
Tutor: HeGuoPing
School: Shandong University of Science and Technology
Course: Applied Mathematics
Keywords: PVD Algorithm "Forget-me-not" term SQP Type PVD Algorithm Maratos Effect Inexact PVD Algorithm Mix-integer Nonlinear Optimization Problem
CLC: O224
Type: Master's thesis
Year: 2010
Downloads: 46
Quote: 0
Read: Download Dissertation

Abstract


Parallel Variable Distribution algorithm(PVD algorithm) is a kind of parallel algorithm in the whole structure. The main difference from other algorithms is its "forget-me-not" term. Each processor has the primary responsibility for updating its block of variables while allowing the remaining "secondary" variables to change in a restricted fashion along some easily computable directions, which enhances robustness and flexibility of the algorithm. This thesis explains the important meaning of PVD algorithm by "forget-me-not" term.In order to get close to the reality, we need to improve the basic framework of PVD algorithm. The thesis gives the improvement direction of PVD algorithm, and puts forward two new PVD algorithms. Firstly, my thesis improves the existing SQP type PVD algorithm, and gives a new FSQP type PVD algorithm, whose search direction is the combination of descent direction, feasible direction and second-order revised direction. This new algorithm is very effective in preventing the occurrence of maratos effect. The theoretical analysis shows that global and superlinear convergence can be induced under some suitable conditions. Secondly, we give a new inexact PVD algorithm for general optimization problem and the justification of its global convergence. In the algorithm, we choose projected gradient residual function as PVD direction, and we replace the minimization problem with a kind of sufficient descent condition.This thesis makes use of PVD algorithm to solve mix-integer nonlinear optimization problem. Its constraints are divided into:separable constrains and global constraints, then we compromise the global constrains to target function using penalty function method. So the mix-integer nonlinear optimization problem becomes separable constrained optimization problem, which can be solved by PVD algorithm of separable constrained optimization problem.

Related Dissertations

  1. Filter Line Search Methods for Nonlinear Optimization,O224
  2. Filter method for solving nonlinear optimization problem,O221.2
  3. Research on Some Parallel Algorithms for Optimization Problems,O246
  4. Study of Parallel Optimization Algorithms,O221
  5. Study of the Parallel Variable Distribution Algorithms of Quadratic Programming,O246
  6. The optimal scale of investment of private equity funds,O224
  7. A Smoothing Method for Solving Model under WCVarR,O224
  8. Mining resources based on genetic algorithm optimization model of,O224
  9. Global Optimization for Sum-of-Ratios Problems,O224
  10. Research on Cultural Algorithm and Its Application in Constrained Optimization Problems,O224
  11. Hysteresis -based optimization of vehicle routing problem,O224
  12. Problem Solving Generalized Geometric Programming Two global optimization methods,O224
  13. Research on Traffic Assignment Problems Based on Intelligent Optimization,O224
  14. A Semi-Smooth Newton Method for Non-Smooth Equations,O224
  15. Trust Region Algorithms Based on the Conic Model,O224
  16. Some Researches on Newton-type Algorithm for Solving Unconstrained Optimization Problems,O224
  17. Unconstrained Nonlinear Conjugate Gradient Method,O224
  18. Conditions for the Superlinear Convergence of Quasi-Newton Methods on Degenerate Solutions,O224
  19. Big Bang Search (BBS) Intelligence Optimization and Its Improvement,O224
  20. Nomonotone Trust Region Algorithms for Unconstrained Optimization,O224

CLC: > Mathematical sciences and chemical > Mathematics > Operations Research > Optimization of the mathematical theory
© 2012 www.DissertationTopic.Net  Mobile