Higher Order Verified Inclusions of Multidimensional Systems by Taylor Models
Abstract
Different from floating point computations, interval methods provide
rigorous enclosures of functions, however the limitation of the
methods is the overestimation mostly caused by the lack of information
on functional dependency. The first cure to the problem is to use a
smaller domain, but when a function is complicated, as it often is for
practical problems, the number of subdivisions becomes quite large. In
case of multidimensional systems, the computational expense by simple
interval methods increases astronomically. A new approach, the Taylor
model method, models a function by a higher order polynomial which
keeps the majority of the functional dependency, and an interval which
contains the small remaining error. The method naturally suppresses
the dependency problem, and proves particularly effective for the
treatment of complicated multidimensional systems.
Note: The scan of the paper loses some interesting
fine detail in the figures due to limited resolution.
Below are higher resolution versions of the figures.
Plots in Fig. 1
Left top,
left bottom,
middle top,
middle bottom,
right top,
right bottom
Plots in Fig. 2
Left top,
left bottom,
right top,
right bottom
K. Makino, M. Berz, Nonlinear Analysis 47 (2001) 3503-3514
Download
Click on the icon to download the corresponding file.
Download Adobe PDF version (565141 Bytes).
Go Back to the reprint server.
Go Back to the home page.
This page is maintained by Kyoko Makino. Please contact her if there are any problems with it.