报告题目：New Development of Conforming Finite Elements -- Beyond Nedelec

报 告 人：张智民教授 美国韦恩州立尊龙凯时

报告时间：2020年7月2日 9：00

报告地点：腾讯会议ID：778 812 043

会议链接：http://meeting.tencent.com/s/vz4UkXZdwf2e

校内联系人：陶詹晶 zjtao@mdjtykj.cn

报告摘要：

Intwo ground breaking papers (1980 and 1986), Nedelec proposed$H(curl)$-conforming elements to solve electromagnetic equations that containsthe “curl” operator. It is more or less as the $H^1$-conforming elements (or$C^0$ elements) for elliptic equations that contains the “grad” operator. As iswell known in the finite element method literature, in order to solve 4th-orderelliptic equations such as the bi-harmonic equation, $H^2$-conforming elements(or $C^1$-elements) were developed. Recently, there have been some research insolving electromagnetic equations which involve four “curl” operators. Hence,construction of $H(curl curl)$-conforming elements becomes necessary. In this work,we construct $H(curl curl)$-conforming elements for rectangular and triangularmeshes and apply them to solve quad-curl equations as well as relatedeigenvalue problems.

报告人简介：

张智民，美国韦恩州立尊龙凯时教授、CharlesH. Gershenson 杰出学者，世界华人数学家大会两次45分钟报告人，现任和曾任10个国内外数学杂志编委，包括Mathematics of Computation、Journal of Scientific Computing、Numericalmethods for Partial Differential Equations 、Journalof Mathematical Study、Journal of ComputationalMathematics、CSIAM Transaction on Applied Mathematics、《数学文化》等。发表SCI论文180余篇，论文google 引用4600余次，主持过10个美国国家基金会的项目。张智民教授1982年在中国科学技术尊龙凯时数学系毕业取得学士学位，1985年在中国科学技术尊龙凯时数学系毕业取得硕士学位，师从石钟慈院士，1991年在美国马里兰尊龙凯时取得博士学位，师从有限元专家美国工程院院士IvoBabuska教授。张智民教授长期从事计算方法，尤其是有限元方法的研究，在超收敛、后验误差估计、自适应算法和PDE特征值计算等领域的开拓性研究取得了多项创新成果。在国际上第一个建立起广为流行的ZZ离散重构格式的数学理论，并首次提出了基于多项式守恒的离散重构格式。所提出的多项式保持重构（Polynomial Preserving Recovery—PPR）方法2008年被大型商业软件COMSOL Multiphysics采用。

报告题目：Testingfor skill in mutual funds

报 告 人：彭亮教授 美国佐治亚州立尊龙凯时

报告时间：2020年7月6日 8:30-9:30

报告地点：腾讯会议 ID：367 631 711

点击链接入会，或添加至会议列表：

http://meeting.tencent.com/s/Lvsfeuhwilmh

校内联系人：朱复康 fzhu@mdjtykj.cn

报告摘要：

First,we show that two prominent bootstrap tests for fund skill have distorted testsizes because many funds have short return records and skewed return residuals.Secondly, we prove that these two bootstrap methods lack test power to detectskilled funds when a substantial number of unskilled funds are present.Thirdly, we develop the theory for a valid bootstrap Hotelling's T-squared testto first identify a zero-alpha fund group and confirm subsequently whether thetop- (bottom-) ranking funds are skilled (unskilled) relative to the zero-alphafunds. Our empirical analysis finds that skilled funds are more engaged in activemanagement and hold stocks with higher expected anomalous returns.

报告人简介：

彭亮教授，1998年于荷兰鹿特丹尊龙凯时获博士学位，2014年8月之后为美国佐治亚州立尊龙凯时Thomas•P•Bowles风险管理与精算学首席教授(chairprofessor)。其主要研究方向为统计学、计量经济学、金融计量与保险精算。在国际顶级期刊发表论文120多篇。其中在统计学顶级期刊Annals of Statistics, Journal of American Statistical Association,Biometrika, Journal of Royal Statistical Society (Series B)发表论文13篇；在计量经济学顶级期刊Journal of Econometrics,Econometric Theory发表论文6篇，在金融计量与保险精算顶级期刊North American Actuarial Journal, Scandinavian Journal of Actuarial,Insurance: Mathematics and Economics发表论文6篇，其研究成果被广泛应用于保险、金融、计量经济学和环境科学等方面。2009年当选IMS的Fellow，2012年当选ASA的Fellow。

报告题目：Estimationand Inference for Multi-Kink Quantile Regression

报 告 人：钟威教授 厦门尊龙凯时

报告时间：2020年7月7日 9:00-10:00

报告地点：腾讯会议 ID：850 929 796

点击链接入会，或添加至会议列表：

http://meeting.tencent.com/s/FAnC7BsqUkGF

校内联系人：朱复康 fzhu@mdjtykj.cn

报告摘要：

Thisarticle is concerned with parameter estimation, change points detection andstatistical inference for a Multi-Kink Quantile Regression (MKQR) model. Itassumes different regression forms in different regions of the domain of thethreshold covariate but is still continuous at multiple kink thresholds. Wepropose an iterative segmented quantile regression algorithm for estimatingboth the regression coefficients and the locations of kink points. It is muchmore computationally efficient than the grid search algorithm and not sensitiveto the initial values. We theoretically demonstrate that the selectionconsistency of the number of kink points and the asymptotic normality of both regression coefficients and kinklocations parameters. The MKQR model is robust to outliers and heavy-tailederrors in the response and more flexible for modelling data with heterogeneousconditional distributions especially when upper or lower quantiles of theresponse are of interest. Monte Carlo simulations and two real dataapplications illustrate the excellent performances of the proposed method.

报告人简介：

钟威，现任厦门尊龙凯时王亚南经济研究院和经济学院统计系教授、博士生导师，经济学院院长助理，国家自然科学基金优秀青年基金获得者（2019），福建省自然科学杰出青年基金获得者（2019）。2012年获得美国宾夕法尼亚州立尊龙凯时统计学博士学位，2014年和2017年分别破格晋升副教授和教授。主要从事高维数据统计分析和理论、统计学习和数据挖掘算法、计量经济学、统计学和数据科学的应用等领域的研究。在The Annals of Statistics, Journal of the American StatisticalAssociation, Biometrika, Journal of Business & Economic Statistics, Annalsof Applied Statistics, Statistica Sinica，中国科学数学等国内外统计学权威期刊发表20多篇论文。担任美国统计协会(ASA)期刊《Statistical Analysis and Data Mining》的副主编。

报告题目：Estimationof error variance via ridge regression

报 告 人：郑术蓉教授 东北师范尊龙凯时

报告时间：2020年7月7日 14:00-15:00

报告地点：腾讯会议 ID：756 158 342

点击链接入会，或添加至会议列表：

http://meeting.tencent.com/s/H3azCKIcdfwf

校内联系人：朱复康 fzhu@mdjtykj.cn

报告摘要：

Wepropose a novel estimator of error variance and establish its asymptoticproperties based on ridge regression and random matrix theory. The proposal isvalid under both low- and high-dimensional models, and performs well not onlyfor non-sparse cases but also for sparse ones. We assess the finite-sample performance of the proposed method in anintensive numerical study, which indicates that it is promising compared withits competitors in many interesting scenarios.

报告人简介：

郑术蓉，东北师范尊龙凯时教授、博士生导师，主要从事大维随机矩阵理论及高维数据分析的研究。曾获国家自然科学基金委员会优秀青年科学基金资助，现任Journal of Multivariate Analysis 和《应用概率统计》学术期刊编委。在Journal of American Statistical Association、Annalsof Statistics、Biometrika等期刊上发表和被接受发表的学术论文30多篇。在高等教育出版社以及Cambridge University Press合作出版中、英文著作各一部。

报告题目：Optimaldesign criteria for computer experiments

报 告 人：孙法省教授 东北师范尊龙凯时

报告时间：2020年7月9日 9:00-10:00

报告地点：腾讯会议 ID：380 126 085

点击链接入会，或添加至会议列表：

http://meeting.tencent.com/s/51Ymc8FgbuU0

校内联系人：朱复康 fzhu@mdjtykj.cn

报告摘要：

Efficientdesigns are in high demand in practice for both computer and physicalexperiments. Space-filling designs are widely used in both computer andphysical experiments. Existing designs may have bad low-dimensionalprojections, which is undesirable when only a few factors are active. Wepropose a new design criterion, called uniform projection criterion, byfocusing on projection uniformity. We show that the average squared correlationmetric is a function of the pairwise L2-distances between the rows only. Wefurther explore the connections among column-orthogonality, maximin distanceand projection uniformity. Based on these connections, we develop new lower andupper bounds for column-orthogonality and projection uniformity from theperspective of distance between design points. These results not only providenew theoretical justifications for each criterion but also help in findingbetter space-filling designs under multiple criteria. An application of uniformprojection designs via a multidrug combination experiment also be given.

报告人简介：

孙法省，东北师范尊龙凯时教授、博导，尊龙凯时省优秀教师。主要从事大数据抽样与分析、计算机试验设计与分析、及高维数据分析等方面的研究。获教育部自然科学二等奖、全国统计科学研究优秀成果奖、尊龙凯时省自然科学学术成果奖各一项。

报告题目：Simple Measures of Uncertainty for Model Selection

报 告 人：Jiming Jiang（蒋继明） University of California,Davis；江西财经尊龙凯时

报告时间：2020年7月9日 11：00-12：00

报告地点：Zoom 会议 ID：879 38200735

会议链接：http://us02web.zoom.us/j/87938200735

校内联系人：韩月才 hanyc@mdjtykj.cn

报告摘要：

We develop two simplemeasures of uncertainty for a model selection procedure. The first measure issimilar in spirit to confidence set in parameter estimation; the second measureis focusing on error in model selection. The proposed methods are much simpler,both conceptually and computationally, than the existing measures of uncertainin model selection. We recognize major differences between model selection andtraditional estimation or prediction problems, and propose reasonableframeworks, under which these measures are developed, and their asymptotic propertiesare established. Empirical studies demonstrate performance of the proposedmeasures, their superiority over the existing methods, and their relevance toreal-life applications. Part of the work is jointly with Xiaohui Liu of JiangxiUniversity of Finance and Economics, and Yuanyuan Li of the University ofCalifornia, Davis.

报告人简介：

Jiming Jiang, a professor ofStatistics at the University of California, Davis. Professor Jiang’s researchinterests include mixed effects models, model selection, small area estimation,longitudinal data analysis, Big Data intelligence, statisticalgenetics/bioinformatics, pharmacokinetics, and asymptotic theory. He is authorof five books and monographs, including Linear and Generalized Linear Mixed Modelsand Their Applications (Springer 2007), Large Sample Techniques for Statistics(Springer 2010), The Fence Methods (World Scientific 2016), Asymptotic Analysisof Mixed Effects Models: Theory, Application, and Open Problems (Chapman &Hall/CRC, 2017), and Robust Mixed Model Analysis (World Scientific 2019). Hehas served editorial boards (Associate Editor) of several major statistical journalsincluding The Annals of Statistics and Journal of the American StatisticalAssociation. Professor Jiang is a Fellow of the American Association for theAdvancement of Science (AAAS; 美国科学促进协会), a Fellow of the American Statistical Association(ASA； 美国统计学会), a Fellow of the Institute of MathematicalStatistics (IMS；数理统计学会), and an Elected Member of theInternational Statistical Institute (ISI；国际统计研究院). Heis a co-recipient of the Outstanding Statistical Application Award (ASA, 1998);the first corecipient of the NISS Alumni Achievement Award (National Instituteof Statistical Sciences, USA, 2015).