題目：Kriging and Alternatives in Computer Experiments

報(bào)告時(shí)間：2012年5月9日（星期三）下午4：00——5：00

地點(diǎn)：研究生樓207

摘要：It has been increasingly practical and popular to simulate a complex system or product on a computer. This new modus operandi has seen many applications in engineering and physical sciences. One key technical problem for implementing the simulations of a large/complex system is the incredible amount of computing time required. To tackle this challenge, meta-models are used as a supplement to expensive computations like finite element simulations. The most commonly used technique for building a meta model is kriging, which originated in geo-statistics. There are known limitations with kriging, especially its numerical instability for large samples and/or dimensions. Alternative techniques to circumvent this problem have been proposed in recent years. They range from tweaking of the original kriging technique to the use of easy-to-compute functions as approximations to the unknown functions. Four such methods will be described and general comments be made.

演講人簡(jiǎn)介：吳建福(C.F. Jeff Wu)教授

國(guó)際著名統(tǒng)計(jì)學(xué)家，美國(guó)工程院院士，中央研究院院士。1971年畢業(yè)于臺(tái)灣大學(xué)數(shù)學(xué)系，1976年在加州大學(xué)伯克利分校獲統(tǒng)計(jì)學(xué)博士學(xué)位。先后執(zhí)教于威斯康星大學(xué)、滑鐵盧大學(xué)、密歇根大學(xué)和佐治亞理工大學(xué)，曾任密歇根大學(xué)統(tǒng)計(jì)系主任，現(xiàn)任喬治亞理工學(xué)院工程統(tǒng)計(jì)講座教授。吳建福教授在數(shù)理統(tǒng)計(jì)、工業(yè)統(tǒng)計(jì)、參數(shù)設(shè)計(jì)和實(shí)驗(yàn)設(shè)計(jì)等諸多領(lǐng)域，取得了令人矚目的杰出成就，發(fā)表論文140多篇，出版專著兩部，指導(dǎo)博士生37名。1987年獲COPSS獎(jiǎng)，2011年應(yīng)邀做COPSS Fisher演講，并將在2012年JSM上做Deming演講，2011年獲中國(guó)科院“愛(ài)因斯坦講席教授”稱號(hào)。