✦ LIBER ✦

On the role of orthonormality of sensitivity functions in parameter optimization problems

✍ Scribed by S.P. Bingulac

Publisher: Elsevier Science
Year: 1969
Tongue: English
Weight: 332 KB
Volume: 5
Category: Article
ISSN: 0005-1098
DOI: 10.1016/0005-1098(69)90113-7

No coin nor oath required. For personal study only.

✦ Synopsis

This paper compares the application of the two algorithms, the classical gradient and the Gauss-Newton, for parameter optimization, and it investigates the influence of the form of sensitivity functions on the convergency of these algorithms. It is shown that the cases where the classical gradient algorithm fails corlespond to the cases where the adjustable parameters are selected in such a manner that the sensitivity functions, used in optimization, approach linearly dependent functions. Hence, a suitable scalar measure of the "degree" of linear dependence of sensitivity functions is defined. By investigating the value of this scalar quantity, it is possible in each particular case to conclude in advance whether the convergency of the classical gradient algorithm will be satisfactory or not.

📜 SIMILAR VOLUMES

Role of Different Parameters in the Opti

Role of Different Parameters in the Optimization of Hydrodynamic Cavitation

✍ P. Braeutigam; M. Franke; Z.-L. Wu; B. Ondruschka 📂 Article 📅 2010 🏛 John Wiley and Sons 🌐 English ⚖ 443 KB 👁 1 views

Stability and uniqueness of a catalyst p

Stability and uniqueness of a catalyst particle problem: Parameter optimization in Liapunov functions

✍ Ching-Tien Liou; Henry C. Lim; William A. Weigand 📂 Article 📅 1973 🏛 American Institute of Chemical Engineers 🌐 English ⚖ 590 KB

On the use of orthogonal functions in fl

On the use of orthogonal functions in fluid-dynamic gradient-based optimization

✍ L. A. Catalano; A. Dadone; V. S. E. Daloiso 📂 Article 📅 2008 🏛 John Wiley and Sons 🌐 English ⚖ 652 KB

On the Problem of More Than One Kurtosis

On the Problem of More Than One Kurtosis Parameter in Multivariate Analysis

✍ H.S. Steyn 📂 Article 📅 1993 🏛 Elsevier Science 🌐 English ⚖ 743 KB

When a multivariate elliptical distribution is used as the basis in multivariate analysis all fourth-order cumulants are expressed in terms of a single kurtosis parameter. This and other well-known properties place unrealistic restrictions on the distribution of the covariance matrix. In this paper

On the solution of the decoupling proble

On the solution of the decoupling problem in lqg optimal systems

✍ Bor-Sen Chen; Chung-Chun Kung 📂 Article 📅 1986 🏛 John Wiley and Sons 🌐 English ⚖ 610 KB

On the problem of modelling for paramete

On the problem of modelling for parameter identification in distributed structures

✍ Mark A. Norris; Leonard Meirovitch 📂 Article 📅 1989 🏛 John Wiley and Sons 🌐 English ⚖ 809 KB

Structures are often characterized by parameters, such as mass and stiffness, that are spatially distributed. Parameter identification of distributed structures is subject to many of the difficulties involved in the modelling problem, and the choice of the model can greatly affect the results of the