Date/Time
Date(s) - 28/11/2023
3:30 pm - 5:00 pm

Location: University Hall (UH) 112 (Note: Speaker will be online, but audience meet in UH 112)

Speaker: Chunfang Devon Lin (Queen’s University)
Bio: Dr. Lin is a professor in the Department of Mathematics and Statistics at Queen’s University. She obtained her PhD from Simon Fraser University and joined Queen’s University in 2008. Her research interest lies in uncertainty quantification of computer experiments, design and analysis of experiments, and interface of data collection and statistical learning. She currently is the Chair-Elect for the American Statistical Association’s Section on Statistics in the Physical and Engineering Sciences (SPES) and serve the Provincial Advisory Committee (PAC) of CANSSI Ontario and Spring Research Conference (SRC) Management Committee. She is AE of Technometrics, Canadian Journal of Statistics, and the New England Journal of Statistics in Data Science.

Title: Modelling and Sequential Design of Computer Experiments with Quantitative and Qualitative Factors

 Abstract: Computer experiments with both quantitative and qualitative (QQ) inputs are commonly used in science and engineering applications. Constructing desirable emulators for such computer experiments remains a challenging problem. We propose an easy-to-interpret Gaussian process (EzGP) model for computer experiments to reflect the change of the computer model under the different level combinations of qualitative factors. The proposed modeling strategy, based on an additive Gaussian process, is flexible to address the heterogeneity of computer models involving multiple qualitative factors. Motivated by the need of finding optimal configuration in the high-performance computing (HPC) system, we propose an adaptive-region sequential design (ARSD) for optimization of computer experiments QQ inputs. Moreover, the adaptiveness of the proposed sequential procedure allows the selection of next design point from the adaptive design region. achieving a meaningful balance between exploitation and exploration for optimization. Theoretical justification of the adaptive design region is provided. The performance of the proposed method is evaluated by several numerical examples in simulations. The case study of HPC performance optimization further elaborates the merits of the proposed method.

Date/Time: Tuesday, November 28, 2023, 3.30-5 p.m. (refreshments – will bring it to UH 112)