CIEM   05476
CENTRO DE INVESTIGACION Y ESTUDIOS DE MATEMATICA
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Numerical analysis of two Derivative Free Optimization Algorithms
Autor/es:
MARÍA B. AROUXÉT; NÉLIDA ECHEBEST; ELVIO A. PILOTTA
Lugar:
Córdoba
Reunión:
Congreso; I Congreso de Matemática Aplicada, Computacional e Industrial (MACI'2007) - XVI Congreso sobre Métodos Numéricos y sus Aplicaciones (ENIEF'2007); 2007
Institución organizadora:
Sección Argentina de SIAM - Asociación Argentina de Mecánica Computacional (AMCA)
Resumen:
Derivative free optimization methods are designed for solving nonlinear optimization problem where the derivatives of the objective function (and the constraints in the constrainedcase) are not available. Formally, we consider the problem of minimize a real valued function f subject to x in the n-dimensional space, where f is a smooth nonlinear objective function bounded below. We assume that the gradient of f(x) and the Hessian matrix of f(x) can not be computed for any x.Our purpose is compare numerically two methods for solving this problem: DFO and NEWUOA, both of them are based on trust-region strategy and multivariate polynomial interpolation. One of the main difference between them is DFO use Newton polynomials instead of Lagrange polynomials for the interpolation. This is an very important aspect because the geometry of the interpolation set and the linear algebra involved depends on such polynomials. We consider that having a deep knowledge of the implementation details of thecodes will enable us to be much more exacting when proposing a new code in the near future. Specifically, we compare the performance (eficiency and robustness) of the two codes by using a set of test problems, taking into account the many differentoptions of the two codes.