|
Activity Number:
|
21
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
| Abstract - #303570 |
|
Title:
|
A Surrogate-Assisted Method Using Adaptive Multi-Accurate Function
|
|
Author(s):
|
Ray-Bing Chen*+ and Weichung Wang and Hsu Chia-Lung
|
|
Companies:
|
National University of Kaohsiung and National Taiwan University and National University of Kaohsiung
|
|
Address:
|
Institute of Statistics, , ,
|
|
Keywords:
|
Computer experiments ; surrogate-assisted method ; iterative function value evaluation ; dynamic controlled iterative process ; multiple accuracy ; sublevel set searching
|
|
Abstract:
|
In many computer experiments, surrogates are determined by response function values evaluated by costly iterative processes. Thus the computational burdens may prevent regular surrogate-assisted methods from being efficient. Instead of computing the fully convergent response function values, we propose dynamically controlling the function evaluation iterations to save on function evaluation processes without downgrading the overall performance. Our algorithms adaptively determine whether each of the function evaluation iterations should be paused, kept running, or re-started; we then use the approximate function values with various levels of accuracy to construct the surrogates. The numerical results show that our algorithms achieve significant savings for sublevel set searching problems that involve identifying positive Lyapunov exponents of a dynamical system.
|
