PETR KRYSL
Department of Computer Science
California Institute of Technology
MS 256-80
Pasadena, CA 91125, USA
http;//www.multires.caltech.edu/~pkrysl
pkrysl@cs.caltech.edu
Abstract:
Approximation of the dynamics of nonlinear time-dependent finite
element mechanics models through a small number of `modes' is
presented as a general methodology with applications, among others, to
control of dynamical systems, design optimization, and adaptive
computations. The reduction preserves the mechanical structure of the
original models, and has been implemented for both explicit and
implicit time integrators. The optimal basis is the Karhunen-Loeve
expansion of the response statistics in the configuration space. A
number of examples document the interesting features and the
computational benefits.