Modeling and Estimation for Electronics and Semiconductor Products

该提案支持四名学生（两名由NSF资助，两名由工业赞助商资助）将研究生研究与对工业问题的实际贡献结合起来。主要的工业合作伙伴是北电，但我们在半导体研究公司、德州仪器和北卡罗莱纳州立大学先进电子材料加工中心的同事也将提供专业知识和实验室访问。因此，这个项目将在统计学和工程学之间建立一座桥梁，无论是在校内还是校外，都可以进行跨学科的研究。学生们将获得制定研究问题、确定统计和工程贡献、探索潜在的基础统计研究领域以及向不同学科的科学家（口头和书面）展示的经验。在本研究中，以下新研究将有利于工业应用和学术基础研究：(1)分析时间或空间相关的非正态离散或连续数据的灵活建模和估计方法；(2)双边时间序列模型的鲁棒估计方法；(3)非线性随机系数模型的鲁棒估计程序，用于时间相关的重复测量。提出的统计程序的性质将通过有限样本模拟和渐近理论应用于北电的实际例子进行评估。这个项目的经验将成为我们重组与三角研究园和罗利公司关系的典范。学生们每周将花10到20个小时在工业现场、工程实验室和统计部门进行研究。受训人员及其导师将在适当的会议上介绍他们的工作，并在统计和工程杂志上发表与项目有关的研究。该目标项目由MPS多学科活动办公室（OMA）和数学科学司（DMS）联合支持。

This proposal supports four students (two by NSF and two by industrial sponsors) to integrate graduate research with practical contributions to industrial problems. The main industrial partner is Nortel, but our colleagues in the Semiconductor Research Corporation, Texas Instruments, and the Center for Advanced Electronic Materials Processing at North Carolina State University will also provide expertise and laboratory access. Thus, this program will build a bridge between statistics and engineering, both on and off campus, for conducting interdisciplinary research. The students will gain experience in formulating research problems, in identifying statistical and engineering contributions, in exploring potential basic statistical research areas, and in presenting (oral and written) to scientists from different disciplines. In this proposal, the following new research will benefit both industrial applications and academic basic research: (1) a flexible modeling and estimation method for analyzing temporally or spatially correlated non-normal discrete or continuous data, (2) a robust estimation method in bilateral time series model, and (3) a robust estimation procedure in nonlinear random coefficient model for temporally correlated repeated measurements. The properties of the proposed statistical procedures will be evaluated through finite sample simulation and asymptotic theory applied to real-life examples from Nortel. The experience from this program will serve as a model to restructure our relationships with companies at Research Triangle Park and Raleigh. The students will spend 10 to 20 hours per week conducting research at industrial sites, engineering laboratories and the statistics department. The trainees and their mentors will present their work at appropriate conferences and publish research relevant to projects in statistical and engineering journals. This GOALI project is jointly supported by the MPS Office of Multidisciplinary Activitie s (OMA) and the Division of Mathematical Sciences (DMS).