Stochastic Intrinsic Kriging for Simulation Metamodelling

Research output: Working paperDiscussion paperOther research output

516 Downloads (Pure)

Abstract

We derive intrinsic Kriging, using Matherons intrinsic random functions which eliminate the trend in classic Kriging. We formulate this intrinsic Kriging as a metamodel in deterministic and random simulation models. For random simulation we derive an experimental design that also specifies the number of replications that varies with the input combinations. We compare intrinsic Kriging and classic Kriging in several numerical experiments with deterministic and random simulations. These experiments suggest that intrinsic Kriging gives more accurate metamodel, in most experiments.
Original languageEnglish
Place of PublicationTilburg
PublisherOperations research
Pages1-19
Number of pages19
Volume2014-054
Publication statusPublished - 8 Sep 2014

Publication series

NameCentER Discussion Paper
Volume2014-054

Keywords

  • Gaussian process
  • Kriging
  • intrinsic Kriging
  • metamodel
  • computer experiment
  • simulation

Cite this

Mehdad, E., & Kleijnen, J. P. C. (2014). Stochastic Intrinsic Kriging for Simulation Metamodelling. (pp. 1-19). (CentER Discussion Paper; Vol. 2014-054). Tilburg: Operations research.
Mehdad, E. ; Kleijnen, Jack P.C. / Stochastic Intrinsic Kriging for Simulation Metamodelling. Tilburg : Operations research, 2014. pp. 1-19 (CentER Discussion Paper).
@techreport{9ab2e856d971475da842de073d7b13b3,
title = "Stochastic Intrinsic Kriging for Simulation Metamodelling",
abstract = "We derive intrinsic Kriging, using Matherons intrinsic random functions which eliminate the trend in classic Kriging. We formulate this intrinsic Kriging as a metamodel in deterministic and random simulation models. For random simulation we derive an experimental design that also specifies the number of replications that varies with the input combinations. We compare intrinsic Kriging and classic Kriging in several numerical experiments with deterministic and random simulations. These experiments suggest that intrinsic Kriging gives more accurate metamodel, in most experiments.",
keywords = "Gaussian process, Kriging, intrinsic Kriging, metamodel, computer experiment, simulation",
author = "E. Mehdad and Kleijnen, {Jack P.C.}",
year = "2014",
month = "9",
day = "8",
language = "English",
volume = "2014-054",
series = "CentER Discussion Paper",
publisher = "Operations research",
pages = "1--19",
type = "WorkingPaper",
institution = "Operations research",

}

Mehdad, E & Kleijnen, JPC 2014 'Stochastic Intrinsic Kriging for Simulation Metamodelling' CentER Discussion Paper, vol. 2014-054, Operations research, Tilburg, pp. 1-19.

Stochastic Intrinsic Kriging for Simulation Metamodelling. / Mehdad, E.; Kleijnen, Jack P.C.

Tilburg : Operations research, 2014. p. 1-19 (CentER Discussion Paper; Vol. 2014-054).

Research output: Working paperDiscussion paperOther research output

TY - UNPB

T1 - Stochastic Intrinsic Kriging for Simulation Metamodelling

AU - Mehdad, E.

AU - Kleijnen, Jack P.C.

PY - 2014/9/8

Y1 - 2014/9/8

N2 - We derive intrinsic Kriging, using Matherons intrinsic random functions which eliminate the trend in classic Kriging. We formulate this intrinsic Kriging as a metamodel in deterministic and random simulation models. For random simulation we derive an experimental design that also specifies the number of replications that varies with the input combinations. We compare intrinsic Kriging and classic Kriging in several numerical experiments with deterministic and random simulations. These experiments suggest that intrinsic Kriging gives more accurate metamodel, in most experiments.

AB - We derive intrinsic Kriging, using Matherons intrinsic random functions which eliminate the trend in classic Kriging. We formulate this intrinsic Kriging as a metamodel in deterministic and random simulation models. For random simulation we derive an experimental design that also specifies the number of replications that varies with the input combinations. We compare intrinsic Kriging and classic Kriging in several numerical experiments with deterministic and random simulations. These experiments suggest that intrinsic Kriging gives more accurate metamodel, in most experiments.

KW - Gaussian process

KW - Kriging

KW - intrinsic Kriging

KW - metamodel

KW - computer experiment

KW - simulation

M3 - Discussion paper

VL - 2014-054

T3 - CentER Discussion Paper

SP - 1

EP - 19

BT - Stochastic Intrinsic Kriging for Simulation Metamodelling

PB - Operations research

CY - Tilburg

ER -

Mehdad E, Kleijnen JPC. Stochastic Intrinsic Kriging for Simulation Metamodelling. Tilburg: Operations research. 2014 Sep 8, p. 1-19. (CentER Discussion Paper).