Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

Queen's University

$13,000.00

The design and analysis of large-scale computer experiments/variance estimations in complex surveys

EF 2010-11

Résumé
With the rapid growth in computer power, scientists are increasingly performing their experiments on computers, with the underlying physical mechanisms represented and implemented by computer codes. In some applications, the number of inputs is very large. In others, the computer codes are fast and call for a large number of trials. In addition, these large-scale experiments can have different types of inputs and outputs. For example, the inputs can be qualitative and quantitative while the outputs can be multivariate, functional or space-time. Such applications have brought new and important challenges to the design and analysis of computer experiments. The objective of the proposed research is to develop new statistical methodology for helping scientists conduct and analyze large-scale computer experiments. This proposal will address the following problems: 1) to develop methodology for building an emulator with a large set of trials; 2) to construct designs for screening experiments; 3) to construct mixed-level orthogonal arrays of high strength and space-filling designs with high dimensional projection property; 4) to generate large robust designs for computer experiments with both control and noise variables. It is expected that the proposed research will provide theoretical development and practical recommendations for large-scale computer experiments. In addition, new collections of designs will be made available online for quick and broad dissemination. The second part of the proposal is motivated by the National Assessment of Educational Progress Survey. There is concern that the first-stage sampling fractions may not be negligible, while at the second stage it should be treated as negligible. As the variance estimation is done presently, the primary sampling units are treated as if they were sampled with replacement. This generally leads to overestimation of the variance. The proposed research will develop methods to handle this issue. New results will have significant impacts on the variance estimation methodology for survey agencies such as Statistics Canada.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
371401-2009
Name nom
Lin, ChunfangDevon
Department département
Mathematics and Statistics
Organization id
29
Institution établissement
Queen's University
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2009
Award amount
13000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
14
Committee name en
Statistical Sciences
Committee name fr
Statistique
Area of application code
1200
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Advancement of knowledge
Area of application fr
Avancement des connaissances
Research subject code
3004
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Applied statistics
Research subject fr
Statistique appliquée
Partie
2
Nb partie
5
Application title
The design and analysis of large-scale computer experiments/variance estimations in complex surveys
Keyword
NSERC
Application summary
With the rapid growth in computer power, scientists are increasingly performing their experiments on computers, with the underlying physical mechanisms represented and implemented by computer codes. In some applications, the number of inputs is very large. In others, the computer codes are fast and call for a large number of trials. In addition, these large-scale experiments can have different types of inputs and outputs. For example, the inputs can be qualitative and quantitative while the outputs can be multivariate, functional or space-time. Such applications have brought new and important challenges to the design and analysis of computer experiments. The objective of the proposed research is to develop new statistical methodology for helping scientists conduct and analyze large-scale computer experiments. This proposal will address the following problems: 1) to develop methodology for building an emulator with a large set of trials; 2) to construct designs for screening experiments; 3) to construct mixed-level orthogonal arrays of high strength and space-filling designs with high dimensional projection property; 4) to generate large robust designs for computer experiments with both control and noise variables. It is expected that the proposed research will provide theoretical development and practical recommendations for large-scale computer experiments. In addition, new collections of designs will be made available online for quick and broad dissemination. The second part of the proposal is motivated by the National Assessment of Educational Progress Survey. There is concern that the first-stage sampling fractions may not be negligible, while at the second stage it should be treated as negligible. As the variance estimation is done presently, the primary sampling units are treated as if they were sampled with replacement. This generally leads to overestimation of the variance. The proposed research will develop methods to handle this issue. New results will have significant impacts on the variance estimation methodology for survey agencies such as Statistics Canada.