Dissertation > Excellent graduate degree dissertation topics show
Smoothing Newton Methods for Sloving e1 Norm Optimization Model
Author: JiaHongTao
Tutor: ZhangHongWei
School: Dalian University of Technology
Course: Operational Research and Cybernetics
Keywords: e1 norm Second-order cone constraints Smoothing Newton Method Global Convergence Local quadratic convergence
CLC: O224
Type: Master's thesis
Year: 2010
Downloads: 22
Quote: 0
Read: Download Dissertation
Abstract
|
In engineering applications , usually with multiple features of an object composed of high - dimensional vector to describe the objects . Linear normed space , the similarity between objects are often characterized by the l1 norm of the difference vector . The face recognition system , for example , the facial features of the people by the high -dimensional vector to represent , and then build the l1 norm optimization model was solved to achieve the purpose of face recognition . So the l1 norm optimization model for solving becomes an important issue . This paper studies a class of linear constraints l1 norm optimization problem . Through the introduction of the second-order cone constraints , the objective function can be transformed into smooth function which the original problem into a second order cone constrained optimization problem . Convenience, in order to study the original problem was constructed based on duality theory the dual problem , then the dual problem is solved . In the solution process , the introduction of the Fischer-Burmeister function complementary KKT system conditions to equivalent replacement will have the dual problem KKT system into a non - smooth equations and then with Armijo line search of the smoothing Newton method for solving . We to verify JΦ the non - singularity , the global and local quadratic convergence properties of the algorithm . Finally, write a MATLAB program for numerical solution .
|
Related Dissertations
- Trust Region Algorithms Based on the Conic Model,O224
- Unconstrained Nonlinear Conjugate Gradient Method,O224
- Studies on Efficient Algorithms for Finite-Dimensional Variational Inequality and Complementarity Problems,O242.23
- A New Superlinearly Convergent Algorithm of Combining QP Subproblem with System of Linear Equations for Constrained Optimization,O241.6
- Smoothing Newton Method for Nonlinear Programming Problem and SQP-Filter Method for Constrained Minimax Problem,O221.2
- Theory and Algorithm Study of Two Kinds of Nonlinear Bilevel Programming,O221.2
- Trust Region Method of New Conic Model for Nonlinearly Equality Constrained Optimization,O221.2
- Trust-Region Algorithm Using Two-dimen-sional Subspace Technique with New Conic Model,O221.2
- Research on Nonmonotone Trust Region Algorithms for Nonlinear Optimization,O224
- A Nonmonotone Smoothing Newton Algorithm for Solving Zero-one Nonlinear Integer Programming Problems,O221.4
- A Modified SQP Algorithm for Nonlinearly Inequality Constrained Optimization,O224
- The Improvement of Nonlinear Conjugate Gradient Methods,O224
- Study of Several Algorithms for Solving Nonlinear Optimization Problems,O221.2
- Two New Nonmonotone Line Search Methods,O224
- A Modified Quasi-Newton Method and It’s Convergence,O224
- Evolutionary algorithm based on immune neural evolution,TP18
- Improved genetic algorithm in nonlinear equations,O241.7
- Some Researches on the Semi-smoothing Asymptotically Newton Method for Complementarity Problems,O241.6
- Some Studies on Numerical Methods for the Complementarity Problems,O224
- Nonlinear complementarity problems Smoothing Newton Method,O224
- A Primal-Dual Interior-Point FS Algorithm and Its Global Convergence,O221.2
CLC: > Mathematical sciences and chemical > Mathematics > Operations Research > Optimization of the mathematical theory
© 2012 www.DissertationTopic.Net Mobile
|