A General Algorithm for Simultaneous Estimation and
Nonlinear Optimization of Constrained Problems
Giacomo Patrizi
Dipartimento di Statistica, Probabilità e Statistiche Applicate
Università degli Studi di Roma "La Sapienza"
Wednesday, January 26, 2005
4:45 - 5:45 PM
Terman Engineering Center, Room 453
Abstract:
Challenging problems in computational mathematical programming stem from empirical
work. From a given data set, a functional form, satisfying all the required statistical
conditions must be determined and an optimal control problem must be solved. Both
optimization problems must be solved simultaneously for nonlinear problems, as the
parameter space and the control space are mutually dependent.
The aim of this paper is to present an algorithm for general nonlinear constrained
simultaneous estimation and optimization problems. Convergence conditions will be derived.
The optimal solution will furnish the values of the parameters of the functional form, that
satisfy the required statistical conditions, and the optimal values of the control variables.
Thus statistically correct optimal control solutions are obtained.
Typical applications in financial optimization problems, in therapeutic control problems
in medicine, in supply chain management will be presented, requiring in certain cases, the
inclusion of binary variables, disjunctive constraints and other special structures, which can
be handled in this general context.
Operations Research Colloquia: http://or.stanford.edu/oras_seminars.html