Software for large-scale linearly constrained minimization problems

• Grant number: 08/04519-2
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: August 01, 2008
• End date: May 31, 2009
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computational Mathematics
• Principal Investigator: José Mário Martinez Perez
• Grantee: Marina Andretta
• Host Institution: Instituto de Matemática, Estatística e Computação Científica (IMECC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 06/53768-0 - Computational methods of optimization, AP.TEM

Abstract

Linearly constrained minimization problems are important not only because of the problem itself, that appears in many areas, but because it appears as a subproblem of more generic problem, such as nonlinear programming. In the candidate's doctoral thesis, a efficient software for solving medium sized linearly constrained problems was developed. Now the goal is to develop a software for solving large-scale linearly constrained problems, that will be a part of the Tango Project (www.ime.usp.br/~egbirgin/tango), which has several softwares to solve different kind of problems.The software to be developed should use a feasible point methodology, an active-set strategy, spectral projected gradient strategy to leave thefaces and several internal algorithms, that the main algorithm will choose accordingly to the characteristics of each subproblem. The software will also use parallelism, whenever it is necessary.The work will be developed with collaboration of professor Ernesto G. Birgin, from the department of Computer Science of the University of São Paulo (USP). The candidate intends to work for some period with professor Michael Saunders, from the department of Management Science and Engineering of the Stanford University.