Batch Sequencing Methods for Computer Experiments

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Aaron Quan*+
• Keywords: computer experiments ; batch ; space filling ; batch sequencing ; expected improvement ; data mining

Abstract

Previously, computer experiments simulations have been performed with the use of a single computer. In order to take advantage of the presence of multiple computers, these simulations need to be run in batches with the batch size determined by the number of computers. Various methods to perform these simulations in batches have been proposed to supplant the approach of sampling one site at a time. The batch methods make use of expected improvement criteria and space filling designs in order to efficiently find all optima. These methods pick the site with the highest expected improvement with the rest of the batch sites selected through various methods that have the right balance of having high expected improvement and being space filling. These methods are tested on various problem types, and the results are compared and contrasted with each other to determine the ideal approach to sampling in batches.