Kriging Models That Are Robust With Respect to Simulation Errors

A.Y.D. Siem, D. den Hertog

Research output: Working paperDiscussion paperOther research output

254 Downloads (Pure)

Abstract

In the field of the Design and Analysis of Computer Experiments (DACE) meta-models are used to approximate time-consuming simulations. These simulations often contain simulation-model errors in the output variables. In the construction of meta-models, these errors are often ignored. Simulation-model errors may be magnified by the meta-model. Therefore, in this paper, we study the construction of Kriging models that are robust with respect to simulation-model errors. We introduce a robustness criterion, to quantify the robustness of a Kriging model. Based on this robustness criterion, two new methods to find robust Kriging models are introduced. We illustrate these methods with the approximation of the Six-hump camel back function and a real life example. Furthermore, we validate the two methods by simulating artificial perturbations. Finally, we consider the influence of the Design of Computer Experiments (DoCE) on the robustness of Kriging models.
Original languageEnglish
Place of PublicationTilburg
PublisherOperations research
Number of pages28
Volume2007-68
Publication statusPublished - 2007

Publication series

NameCentER Discussion Paper
Volume2007-68

Fingerprint

Experiments

Keywords

  • Kriging
  • robustness
  • simulation-model error

Cite this

Siem, A. Y. D., & den Hertog, D. (2007). Kriging Models That Are Robust With Respect to Simulation Errors. (CentER Discussion Paper; Vol. 2007-68). Tilburg: Operations research.
Siem, A.Y.D. ; den Hertog, D. / Kriging Models That Are Robust With Respect to Simulation Errors. Tilburg : Operations research, 2007. (CentER Discussion Paper).
@techreport{fe73dc8b20d64f5095ebf96212a93c00,
title = "Kriging Models That Are Robust With Respect to Simulation Errors",
abstract = "In the field of the Design and Analysis of Computer Experiments (DACE) meta-models are used to approximate time-consuming simulations. These simulations often contain simulation-model errors in the output variables. In the construction of meta-models, these errors are often ignored. Simulation-model errors may be magnified by the meta-model. Therefore, in this paper, we study the construction of Kriging models that are robust with respect to simulation-model errors. We introduce a robustness criterion, to quantify the robustness of a Kriging model. Based on this robustness criterion, two new methods to find robust Kriging models are introduced. We illustrate these methods with the approximation of the Six-hump camel back function and a real life example. Furthermore, we validate the two methods by simulating artificial perturbations. Finally, we consider the influence of the Design of Computer Experiments (DoCE) on the robustness of Kriging models.",
keywords = "Kriging, robustness, simulation-model error",
author = "A.Y.D. Siem and {den Hertog}, D.",
note = "Pagination: 28",
year = "2007",
language = "English",
volume = "2007-68",
series = "CentER Discussion Paper",
publisher = "Operations research",
type = "WorkingPaper",
institution = "Operations research",

}

Siem, AYD & den Hertog, D 2007 'Kriging Models That Are Robust With Respect to Simulation Errors' CentER Discussion Paper, vol. 2007-68, Operations research, Tilburg.

Kriging Models That Are Robust With Respect to Simulation Errors. / Siem, A.Y.D.; den Hertog, D.

Tilburg : Operations research, 2007. (CentER Discussion Paper; Vol. 2007-68).

Research output: Working paperDiscussion paperOther research output

TY - UNPB

T1 - Kriging Models That Are Robust With Respect to Simulation Errors

AU - Siem, A.Y.D.

AU - den Hertog, D.

N1 - Pagination: 28

PY - 2007

Y1 - 2007

N2 - In the field of the Design and Analysis of Computer Experiments (DACE) meta-models are used to approximate time-consuming simulations. These simulations often contain simulation-model errors in the output variables. In the construction of meta-models, these errors are often ignored. Simulation-model errors may be magnified by the meta-model. Therefore, in this paper, we study the construction of Kriging models that are robust with respect to simulation-model errors. We introduce a robustness criterion, to quantify the robustness of a Kriging model. Based on this robustness criterion, two new methods to find robust Kriging models are introduced. We illustrate these methods with the approximation of the Six-hump camel back function and a real life example. Furthermore, we validate the two methods by simulating artificial perturbations. Finally, we consider the influence of the Design of Computer Experiments (DoCE) on the robustness of Kriging models.

AB - In the field of the Design and Analysis of Computer Experiments (DACE) meta-models are used to approximate time-consuming simulations. These simulations often contain simulation-model errors in the output variables. In the construction of meta-models, these errors are often ignored. Simulation-model errors may be magnified by the meta-model. Therefore, in this paper, we study the construction of Kriging models that are robust with respect to simulation-model errors. We introduce a robustness criterion, to quantify the robustness of a Kriging model. Based on this robustness criterion, two new methods to find robust Kriging models are introduced. We illustrate these methods with the approximation of the Six-hump camel back function and a real life example. Furthermore, we validate the two methods by simulating artificial perturbations. Finally, we consider the influence of the Design of Computer Experiments (DoCE) on the robustness of Kriging models.

KW - Kriging

KW - robustness

KW - simulation-model error

M3 - Discussion paper

VL - 2007-68

T3 - CentER Discussion Paper

BT - Kriging Models That Are Robust With Respect to Simulation Errors

PB - Operations research

CY - Tilburg

ER -

Siem AYD, den Hertog D. Kriging Models That Are Robust With Respect to Simulation Errors. Tilburg: Operations research. 2007. (CentER Discussion Paper).