desk


News

New REFM Research Compute Cluster

The Technology Solutions Research team recently built and deployed a new dedicated compute cluster for the Real Estate and Financial Markets (REFM) Lab. This gives REFM a fresh, isolated cluster for their computing needs. The new cluster fulfills REFM's storage and performance requirements, and ensures greater stability within the Haas Data Center. The dedicated REFM compute cluster is a part of Research Computing's effort to expand the Data Center to meet Haas' growing data needs.

Technology Solutions Partners with Berkeley Campus Departments to Support and Expand Data Science Research and Teaching

Haas Technology Solutions (TS) recently established a collaborative partnership with the Berkeley Data Science Educational Program (DSEP) and Berkeley Research Computing in Information Systems & Technology (IST) to host at Haas, the Berkeley Summer Sessions Undergraduate course: Foundations of Data Science (Data 8).

The partnership was led by the TS Director for Research Computing, Zane Cooper, and works with the TS Enterprise Services & Research Team to provide and operate the Data Science compute cluster and enterprise storage. DSEP will provide IT development and operations knowledge for supporting Data Science courses.

DSEP is a key component of Berkeley's Data Science ecosystem, and offers an interdisciplinary curriculum that provides a foundation for Berkeley Undergraduates of all majors to engage capably and critically with data. The program starts at the introductory level, with the path-breaking foundational Data 8 course, which combines instruction in core computational and statistics concepts while enabling students to work hands-on with real data in a range of fields.

The course does not require prior data science experience, and satisfies prerequisites in many departments for the Letters and Science Quantitative Reasoning requirement. As Data 8 operates at scales of up to 500+ students enrolled at one time, it has helped DSEP broaden systems and expertise in running IT infrastructure for Campus and other programs.

DSEP and TS are also partnering to share resources and knowledge. TS resources help expand DSEP's capabilities to evaluate and test new Data Science tools and to integrate high-performance research computing capabilities. In turn, these tools will help TS support the more rigorous requirements of its Faculty and Student Research and instructional support in the Masters of Financial Engineering degree program.

The Data 8 course content (development and student content support) for Summer Sessions will be supported directly by the DSEP Staff, course Teaching Assistants and Instructors. DSEP operates with the cooperation of the Department of Statistics, Economics, Computer Science, and the College of Letters and Science.

AEoD Helps MFE Students Win Finance Competition

A group of Haas Masters of Financial Engineering (MFE) Students recently won a Morgan Stanley Finance Competition with the help of a new locally-developed research platform called, Analytics Environment on Demand (AEoD). This platform, developed from a Campus working group project, features virtualization technologies and software that clients can access from anywhere and almost any device.

The winning MFE Students wielded AEoD to build and test a framework and methods that determine if financial news headlines have meaningful impact on stock prices. This resulted in their winning research paper, Using Natural Language Processing Techniques for Stock Return Predictions.

AEoD was developed by a Campus-wide partnership featuring MFEs Aaron Nebres and Haas Technology Solutions (TS) Usha Manandhar and Research Computing Director Zane Cooper. It supports multiple use cases, and works well for single-task functions, such as optical character recognition applications or geo-coding. But it also works well for ad-hoc and planned research projects that use multiple applications and have a finite lifecycle. Each client receives his/her own virtual desktop or server through Citrix with a custom software build. If more resources are needed they can be added on-the-fly, with up to 384GB of RAM, 24 CPU cores, and virtually infinite storage (though resource fees may apply). This makes AEoD both scalable and efficient.

Because the TS Infrastructure group is a founding AEoD partner, and since Cooper collaborates with the Berkeley Research Computing Consultants team, Haas receives an allocation of AEoD free of charge. TS then provided this allocation to MFE student teams who participated in the Morgan Stanley Competition.

The AEoD working group continues to expand the service to new Berkeley campus clients, developing and enhancing capabilities with plans to add virtualized applications, and Linux support in the near future. For AEoD, Haas currently partners with Berkeley Research Computing (IST), the Goldman School of Public Policy, Digital Humanities, and the Data Lab.

Technology Solutions Enterprise & Research Services group Expands Data Science Support at Haas

Haas Technology Solutions (TS) has expanded its support for Faculty, Research Centers and Institutes at the School with new services for data science infrastructure and analytics resources that are optimized specifically for each client. These expanded services augment TS' high-performance compute and remote desktop cluster with statistics applications, programming languages and other tools for data analytics and research.

As the demand for research computing expands, more projects delivering these services to clients have been launched, including:

  • additional research storage
  • data acquisition and transformation
  • relational and non-relational databases to optimize data mining and retrieval
  • data science languages and tools (e.g., Python, R, Jupyter Notebook, etc.)
  • distributed computing and big data infrastructure
  • virtual and containerized computing on demand
  • cloud computing and storage
  • geocoding application deployment
  • graphics processing unit computing

About Research Services

Our Goal: Provide the best performance and service quality at the lowest cost, developed through consulting engagements, evaluation, analysis and partnerships.

Research Computing provides high-touch, in-depth services that connect Faculty and Students to the best-available infrastructure technologies for their particular needs and use cases. This includes:

  • high-performance compute servers
  • economical storage and backup
  • analytics and visualization software
  • Cloud, on-premise and hybrid computing services
  • IT architecture design for research data acquisition, transformation, and optimization; research analytics and visualization
  • research workflow improvments

These services may be provided at the Haas datacenter, at UCB Central Computing, at a national Supercomputing facility (e.g., SDSC or XSEDE), or in the Cloud by a commercial provider.  Recharge fees may apply to cover the cost of compute, storage, or backup services.