Acquisition of Beowulf Cluster for Economic Research

在美国国家科学基金会重大研究仪器奖的支持下，德克萨斯大学奥斯汀分校经济系的一组研究人员将获得一个计算节点集群。这个集群将使研究小组能够在对一些经济问题的定量研究中使用并行处理技术。提案中概述的应用侧重于国际贸易、国际金融、计量经济学和宏观经济学领域的一些有趣的经济问题。这些应用是通过使用通用的计算技术联系在一起的，许多应用依赖于相对复杂的动态优化问题的解决方案。这些优化问题的结构是为了提供对经济问题的见解，既有积极的（理解我们所观察到的），也有政策相关性。这项资助使我们能够克服计算限制：问题的复杂性使得使用传统的计算工具来解决它们的成本高得令人望而却步。这项研究将产生三大影响。首先，这项工作将直接提高我们评估各种政策对经济的影响的能力，无论是在微观层面还是在总量层面。第二，我们将把我们在编程方面的进步提供给其他用户。这将通过为UT经济学集群开发一个网页，通过电子工作文件系列传播方案和研究成果。因此，从这一经验中获得的知识不仅将促进我们的个人研究项目，而且将通过这种互动使博士生、访问学者和其他研究人员受益。第三，我们将积极培训研究人员使用集群进行计算。正如提案中所概述的那样，在UT，我们有一个活跃的计算领域，这些课程的博士生将可以访问集群进行适当的项目。此外，我们有一个积极的访问研究员计划，访问集群，沿着培训，将成为访问学者的一部分。

With support from a National Science Foundation Major Research Instrumentation Award, a group of researchers in the Economics Department at the University of Texas at Austin will acquire a cluster of compute nodes. This cluster will allow the research group to use parallel processing techniques in their quantitative study of a number of economic problems. The applications summarized in the proposal focus on a number of interesting economic questions in the areas of international trade, international finance, econometrics and macroeconomics. These applications are linked by the use of common computational techniques.Many of the applications rely on the solution of relatively complex dynamic optimization problems. These optimization problems are structured to provide insights into economic questions with both positive (understanding what we observe) and policy relevance. This grant allows us to overcome the computational constraint: the complexity of the problems makes them prohibitively expensive to solve using conventional computational tools. The proposed research will have three broad impacts. First, the work will directly advance our ability to evaluate the impact of various policies on the economy, both at the micro-level and atthe aggregate level. Second, we will make our advances in programming available to other users. This will be through the development of a web page for the UT Economics cluster to disseminate programs as well as research results through an electronic working paper series. Thus knowledge gained from this experience will not only further our individual research projects but will benefits PhD students, visiting scholars and other researchers through this interaction.Third, we will actively train researchers in use of the cluster for computational purposes. As outlined in the proposal, at UT we have an active computational field and PhD students in theseclasses will have access to the cluster for appropriate projects. Further, we have an active visiting researchers program and access to the cluster, along with training, will be made part of apackage for visiting scholars.