題目：Investigating the Point-based Surrogate Model for Solving the Spatial Facility Location Problem
報告簡介：The spatial facility location problem (SFLP) considers covering continuous demand of a region with facilities that can be placed anywhere. One of the approaches to solving the SFLP is to first aggregate spatial demand into discrete points and then solve the point-based model for the conventional facility location problem (FLP) as a surrogate model to the original SFLP. The solution quality of the surrogate model, however, is measured by the percentage of the continuous space actually covered in the original SFLP. We explore the behavior of this solution approach and examine the factors contributing to the quality of the solution to the SFLP. Three sources of errors are examined: 1) the spacing of the point representation, 2) the existence of multiple solutions to the surrogate point-based model, and 3) the round-off error induced by the computer representation of numbers. A surrogate error measure is proposed to characterize the relationship between the spacing, the coverage area of a facility, and the error of the spatial coverage using the point-based surrogate model. The point-based surrogate model is enhanced to extract from the multiple solution set the solutions that reduce the coverage error. The effect of the round-off error on the quality of the solution to the SFLP is also discussed.
報告人簡介：林偉華是美國亞利桑那大學系統與工業工程系副教授，他分別在1985年、1989年、1995年于美國三所不同的名校獲得了計算機科學專業學士學位、數學專業碩士學位、和土木工程專業博士學位，并在加州伯克利進行了為期兩年的博士后研究工作，之后任教至今。他在物流系統優化領域中的諸多頂級期刊都有第一作者的論文發表，如Transportation Science、Transportation Research Parts E、以及IEEE Transaction on Intelligent Transportation Systems等；他目前共發表論文84篇，總引用達1352次，h因子為17（Google Scholar）。