题目： Generalized good lattice point sets
报告人：周永道 博士  四川大学
时间：2014年11月21日(周五)下午3:00-4:00
地点：维格堂113


 
摘要： Good lattice point sets are widely used in quasi-monte carlo method, uniform design and computer experiments, while the space-filling property of good lattice point sets needs to be improved especially when the number of factors is large.  We show that  linear level permutation of a regular design does not decrease the Lp-distance, as well as good lattice point sets. Then, using the technique of linear level permutation for good lattice point sets, we can obtain designs  with better space-filling property in the terms of maximin distance criterion and uniformity criterion, and denoted as generalized good lattice point  sets. Generalized good lattice point sets are shown to have better space-filling property than good lattice point set, orthogonal Latin hypercube design and orthogonal maximin Latin hypercube design in many cases. Generalized good lattice point sets are recommended to use due to the simple construction method and good properties even for large number of runs and number of factors.