Abstract #453

# 453
High-throughput computing in support of dairy science.
M. Livny*1, 1University of Wisconsin, Madison, WI.

For over 3 decades, we have developed technologies to offer capabilities that address the needs of researchers with high-throughput applications. Our technologies have been adopted worldwide and the capabilities we support on the UW-Madison campus as part of the Center for High Throughput Computing (CHTC) have delivered to researchers and collaborators almost 400M core hours in the past 12 mo. These cycles have been used across the entire spectrum of science domains ranging from single PI efforts to the largest science endeavors yet undertaken in areas such as high-energy physics and astrophysics. Many of these high throughput applications involve the extraction, processing, and analysis of information from very large ensembles of observational data. A cornerstone of our technologies are tools that automate the management of very large (hundreds of thousands) interdependent computational tasks. Annually, the CHTC serves more than 200 projects across UW-Madison. Some of these high throughput projects harness more than 100K core hours per day from more than a dozen campus based computer resources. Autonomous resources are managed by the HTCondor resource and job management system and are federated by the CHTC to facilitate sharing. The CHTC also is an active member of the Open Science Grid national consortium that offers High Throughput services across more than 60 sites in the US. The CHTC has been closely involved in several UW-based phenotyping studies ranging from human health to corn yield. These studies involved machine learning, statistical inference, and image processing. In a recent partnership with UW dairy and optimization researchers, we initiated an effort to build a repository of dairy farm data and decision engine. The effort leverages the distributed computing capabilities of our technologies to automate the collection, cleaning, processing, and analyses of the collected data. Our talk will provide an overview of the open source technologies we offer, the principals that guided us in developing and deploying these technologies and how they have been leveraged by researchers in diverse fields.

Key Words: high-throughput applications

Speaker Bio
Miron Livny is a professor of computer science at the University of Wisconsin-Madison, principal scientist at Core Computational Technology of the Wisconsin Institutes for Discovery, chief technology officer of the Wisconsin Institutes for Discovery, director of the UW Center for High Throughput Computing (CHTC), director of the Software Assurance Marketplace, and technical director of the Open Science Grid (OSG). His research interests include distributed processing systems, high throughput computing, software assurance, cyberinfrastructure