Reducing the time that TRM requires to solve systems of nonlinear equations
Hasan H. Dwail, Mushtak A. K. Shiker
Abstract
Abstract The trust region method is one of the important and effective ways to solve optimization problems due to its robust and accuracy in its convergence. One of its disadvantages is that its algorithm needs much time to solve its subproblems especially in the case of the problems that have a large-scale data. In this work, a new algorithm is suggested to solve this problem specifically subspace TRM. The initial TRR will be adequate to the subproblems which can be solved with a simple subspace, it will be simpler to solve than the classic TRM. The global convergence of the new method is investigated. The numerical results indicated that the new approach is promising.
Topics & Concepts
Convergence (economics)Subspace topologySimple (philosophy)Mathematical optimizationTrust regionNonlinear systemComputer scienceScale (ratio)MathematicsArtificial intelligenceEpistemologyPhysicsPhilosophyRADIUSEconomicsComputer securityQuantum mechanicsEconomic growthAdvanced Optimization Algorithms ResearchSparse and Compressive Sensing TechniquesIterative Methods for Nonlinear Equations