UM-ATLAS PROJECT

The ATLAS Experiment - the UM-ATLAS Collaboratory Project’s First Collaboratory 7

This document provides an update on the activities and plans of the University of Michigan ATLAS Collaboratory Project, a novel interdisciplinary program initially authorized by the University in February 1997. The goal of the Project is to study and advance the technologies and practices required for the organization and execution of modern, large-scale collaborative research experiments.

The need for studies of “collaboratories,” and for optimizing their functioning, has emerged from the growing need in many disciplines for the simultaneous, active involvement of a large number of geographically separated investigators in the conduct of key research and development projects. These scientists are often unified by a shared desire to achieve a certain major research objective under the constrained circumstances of the need for access to a one-of-a-kind instrument or facility, or by the overwhelming size and complexity of the task being undertaken.

In fields requiring such approaches, the mushrooming of the number of participants and the increasing remoteness of specialized facilities might be regarded as a natural evolution in the way scientists must organize themselves to achieve common goals. But, while there is considerable reason for elation that we still have the technological ingenuity for pursuing large scale collaborative studies of this type, it should be noted that there are potential downsides for the system of higher education we have watched perform so well in the past. Specifically, we must ask what these new modalities for carrying out research do to the interactions between undergraduate students, their professors, graduate students, teaching assistants and postdocs.

At the very time the public is expressing increasing concern about the amount of time faculty spend off-campus, often at remote research facilities, many faculty in truly large collaborations are finding that they need to spend even more time away. This has an obvious impact on the time faculty can effectively devote to their undergraduate teaching responsibilities. Graduate education is also affected, with many graduate students finding themselves needing to be on campus (to take courses, act as teaching assistants, etc.) at the very time significant decisions and studies that will form the basis for their thesis work are being conducted at a distant site.

It would be disastrous if these trends resulted in the disengagement of university faculty and students from frontier research, or if they were allowed to rupture the traditional bonds between undergraduate education and graduate education.

We assert that the very expanding technology which makes possible new studies of challenging problems in the physical, biological and social sciences, should be harnessed to help preserve those key elements of higher education being threatened by the new reach those technologies are giving the disciplines in the first place. Restated, we ask: what are the collaborative tools made possible by new technologies that will reduce the demands on faculty to be away from campus, and that will facilitate the full remote participation of faculty and their students in all aspects of their research?

It is not suggested that the need for the physical presence of faculty and students at an experimental site will be eliminated. Rather, the goal is to preserve those instances of on-site presence for occasions truly requiring such presence, and to greatly expand the possibilities for meaningful contact and interaction with the experiment via remote means.

Were these issues to affect only one discipline, it might be argued that it is the responsibility of that discipline to seek whatever remedies are required. But the fact is that today there are more and more fields being presented with the same set of problems associated with collaborative research. Thus, although it was the changing paradigms of the field of high energy physics, and issues presented to the UM high energy physics group as it moves to play a major role in the 2000-person ATLAS Experiment that was the primary motivation for the formation of the current study, there are many examples from other fields that reflect the same challenges.

Before describing the ATLAS Experiment in some detail, it is instructive to consider a few other examples. One example of current large collaborations is the Human Genome Project, a 15-year effort formally begun in October 1990 to discover all the 60,000 to 80,000 human genes (the human genome) and make them accessible for further biological study. In the United States, there are more than 30 institutions receiving federal funding to work on the project. Internationally, about 1000 individuals from 50 countries are members of HUGO, which helps to coordinate international collaboration in the Genome Project. Even without describing any detail, it is readily understood that the task of managing the activities of so many organizations requires sophisticated tools for communication and organization.

The Upper Atmospheric Research Collaboratory (UARC) is another very important example of the iterative development of collaborative technology. Through UARC technology, atmospheric scientists around the world study a set of scientific and environmental issues in a collaborative setting. Early versions of the collaboratory technology permitted access for only a limited number of users (less than 12), but improvements have increased the capacity to over 50 participants, at sites ranging from Alaska to the former Soviet Union. During what are called campaigns, scientists located all over the planet work with data being collected at sites around the northern hemisphere — from the Sondestrom facility in Greenland to EISCAT in Norway, to Millstone in Massachusetts, and through an array of high-frequency radar sites across Canada — as well as from two satellites orbiting Earth.

We believe that UM is singularly well suited to carry out the studies proposed by the UM-ATLAS Collaboratory Project. UM is one of the largest and most comprehensive research universities in the U.S., and one of only a handful of universities with targeted programs in collaboratory studies. UM is also nationally recognized as one of the most innovative in linking undergraduate education and research (RAIRE Award), and one with faculty talent that has demonstrated its capability to develop nationwide internets (NSFNET) and Internet browsers (Mosaic). The UM is closely affiliated with consortia designed to develop the next generation Internet (UCAID-Internet2, ALLIANCE). Finally, the fact that we are engaging the challenges of collaboratory studies at CERN – where the World Wide Web was first developed – positions us to carry out the planned R/D in an almost ideal setting.

Presented below are descriptions of the initial proposed set of research topics, a review of the preparations underway, and a presentation of the organizational and governance structure for the project. In addition, the problems that will be ameliorated by the use of emerging technologies in computer networking and human factor studies will also be described, with particular reference to the areas where the University of Michigan is positioned to make extraordinary contributions.

THE P ROBLEMS TO BE A DDRESSED AND THE I NTELLECTUAL C HALLENGES

The research problems to be addressed by the UM-ATLAS Collaboratory Project fall into three categories: collaborative tools that enrich undergraduate and graduate education, technologies that enable globally dispersed scientists to work together seamlessly, and last, facilitation of the network enhancements necessary for the first two foci to become a reality.

There are emerging technologies and collaborative tools that can be applied to expand, develop and enrich interactions between students and their non co-located mentors. For instance, graduate students in a research group should have high quality video links with their faculty advisors on a daily basis. Undergraduate students should be able to follow the work of their summer faculty mentors all year long, perhaps even sitting in on weekly meetings via video-links. These same tools could be used to promote outreach to students who might otherwise never have an opportunity to interact with a large international research project. Through the development of better distance learning tools, students will not only maintain access to the faculty who are conducting research at a distant laboratory, but they will be given an opportunity to interact with the ongoing research project in a way not currently possible.

Some key R/D issues in this realm include the development of a set of hardware and applications toolkits and related protocols supportive of mentoring communications between students and faculty members. This will require input from experts working on conferencing packages, application sharing, whiteboard systems, human factor analyses, as well as networking quality and reliability. The new “quality of service” protocols will be evaluated for their impact on the improvement of conferencing activities.

Given the interest of the University of Michigan Media Union in the topic of how technology can be used in support of education, we expect it will play the lead role in carrying out this research.

The second area of research for the Project seeks to identify or develop technologies that permit collaborating scientists to more effectively work together on projects in a dispersed, global setting. Virtual offices should make it almost impossible for, say, three occupants in Geneva and three in Ann Arbor to determine who is where; each should feel like one of six people in one office. Work within the Project, to design and test collaboratory tools that make communication seamless, and to literally eliminate all significant differences between local and remote classifications, will benefit an enormous constituency throughout science and industry. Even within a single field, the impact will embrace not only the facilitation of the traditional meetings required to manage a large project, but the development of tools that enable discovery and discussion of scientific results between non co-located research scientists as the results from the experiments are streaming in.

An example of an R/D subject in this area is the further development of the shared application tool “WIRED” for which the School of Information has received a license from CERN to examine and extend. This package permits remotely located researchers to jointly view, discuss and manipulate three-dimensional images in a collaborative setting.

In the third focus, the Project will engage in the development, along with other partners, of the infrastructure necessary for the next generation of scientific investigations. By addressing the weaknesses in the current infrastructure (e.g., in global network communications), all areas, not just those devoted to scientific collaboration, will benefit. Clearly, one area that deserves special attention, as we move into increased international collaboration, is the improvement of data links to Europe, and of systems that will provide dedicated bandwidth for designated activities.

An example of a R/D effort in this category would be the tracking of improvements in videoconferencing achievable through the use of the new “ quality-of-service” protocols. This is currently being pursued in discussions with UCAID-INTERNET2 and MERIT.

The above represent important problems for which current technology has not provided adequate answers. A focussed effort to describe the need, to apply existing technology, and to extend technology where needed, can have a significant impact in addressing the remote collaboration issues we have identified.

Ideally, an institution that dares to pioneer in the use of technology to foster the collaborative research of faculty and students should have certain characteristics. It should have a large and diverse set of research programs, encompassing both those which are traditional in nature and located on-campus, as well as those in the physical, social, medical and other sciences that have a strong globally distributed component. It should have faculty who are experts on the operation and management of computing networks, with strong ties to every major network improvement effort in the world. It should have access to a set of its researchers who are willing to be used as “subjects” in the development of new protocols and new technologies. Finally, it should have a demonstrated commitment to improvements in undergraduate and graduate education. These characteristics almost uniquely identify the University of Michigan, and highlight why we believe our university must play a major role in this area.

The University of Michigan continues to be ranked as the nation's #1 research university, measured by the volume of its annual R/D expenditures.
UM was awarded one of the first NSF Recognition Awards for the Integration of Research and Education (RAIRE), granted in recognition of our contributions to the integration of research and undergraduate education.
We helped pioneer and operate NSFNET, the first network connecting U.S. research universities. Many of the researchers that were involved in this project are still active members of our campus.
We have faculty instrumental in the development of web browsers and the concept of digital libraries.
UCAID-INTERNET2 is headquartered in Ann Arbor, with many of its leaders having strong ties to the University of Michigan.

It is difficult to think of any other university in the nation with such a set of relevant linkages and expertise.

Though many examples could be given to further emphasize why we believe that the University of Michigan should be the locus of the proposed research and development, we cite below just four examples. We note the activities of our new School of Information, expertise and facilities in the College of Engineering, the close ties between UM and UCAID-Internet2, and the participation of the Physics Department in the ATLAS Experiment.

All of the components that have been described thus far in this paper as necessary for a modern, large-scale research program on “ collaboratories” are integrated and studied at the School of Information (SI), a modern, interdisciplinary school offering a holistic view of information systems. At SI, researchers in organizational and cognitive psychology, computer science, economics, sociology, library science and information studies come together to pursue the issues of modern knowledge environments. As mentioned above, a major collaboratory initiative of the School of Information is the UARC Project (its successor is SPARC ¹ ). UARC successfully demonstrates how modern collaboratory tools can transform the modus operandi within a science domain, providing for capabilities in global-scale experimental work and in rigorous data-theory comparisons that simply did not exist prior to the project. Over a six-year period, a team of space physicists, computer scientists, and behavioral scientists evolved a suite of collaboration capabilities providing rich, real-time access to a wide variety of data and modeling resources. Now, the SPARC project will significantly extend the power of technology-mediated, distributed knowledge-networking systems. It combines experimental data streams and their interpretation, theoretical models, real-time campaign support, capture mid-replay of collaborative sessions, post-hoc analysis workshops, access to archival data and digital libraries, and educational/outreach modules. An important outcome of SPARC for the science community will be a functional and ope rational space weather predictive capability. Equally important, SPARC will be a major test-bed to further understand and design collaborative knowledge work systems from a merger of social and technical principles.

At the University of Michigan, the College of Engineering is a member of the National Partnership for Advanced Computational Infrastructure (NPACI), which is building the nation's computation infrastructure. NPACI provides a distributed, national metacomputing infrastructure that enables data-intensive computing. This extraordinary resource permits us to explore issues associated with the management of large data sets in a regional setting. Questions such as how to acquire and maintain updated images of data produced at the core remote facility and how to share those data with other universities in the local region, will be of increasing relevance in the future.

Internet2, a project of the University Corporation for Advanced Internet Development (UCAID), is led by over 135 top U.S. universities (including the University of Michigan), over 30 non-profit affiliates, and a growing list of international organizations with similar missions. The UCAID participants are developing broadband applications, as well as engineering and network management tools, for research and education. They are working with industry and government to enable and facilitate the advanced network applications necessary to meet emerging needs in higher education, including international connectivity. The UM-ATLAS Collaboratory Project expects to be closely involved with UCAID in conducting research and development on several topics of shared interest, as will be explained below.

The ATLAS Experiment - the UM-ATLAS Collaboratory Project’s First Collaboratory

The European Organization for Nuclear Research (CERN) Laboratory’s primary goal for the next decade is the construction and operation of the Large Hadron Collider (LHC), the highest energy particle accelerator ever built. It will serve as a precise and powerful instrument for probing matter at the deepest levels, and will elucidate such concepts as the origin of mass, and the fundamental nature of the basic building blocks of matter and the forces by which they interact. A major LHC experiment, ATLAS, in which the UM Department of Physics is playing a key role, both demonstrates the need for high quality collaboratory tools, and serves as a test-bed for some of the ideas that will emerge from the Project. (This is why the term “ATLAS” has been used as a temporary placeholder in the title of the collaboratory project.) The second major LHC experiment is the CMS Experiment. Although competitors, ATLAS and CMS cooperate on various technical challenges, such as the development and effective use of collaborative tools.

The ATLAS Experiment is a worldwide, 2000 member, high energy physics program. The Michigan faculty group includes Robert Ball, Jay Chapman, David Gerdes, Homer Neal, Jianming Qian, Greg Tarle, Rudi Thun and Bing Zhou. UM is proud to have one of the largest institutional groups represented in the entire collaboration. These Michigan faculty are playing a major role in the experiment by designing, prototyping, testing, and constructing (in Ann Arbor), a significant part of a 90,000 component muon forward spectrometer. The UM group is also overseeing the design and construction of the trigger electronics for the muon system, and will work on various components of the computing system for the full experiment.

These responsibilities will require a massive amount of activity over the next seven years, both in Ann Arbor and at CERN. On any given occasion there will be UM ATLAS faculty, students and staff in Ann Arbor and UM ATLAS faculty, students and staff at CERN. Daily communication must take place between these groups and between these groups and their collaborators around the world.

As an aside, the LHC project in Geneva is an outgrowth of the failed U.S. Superconducting Super Collider Project proposed for Dallas, Texas, and cancelled in 1993. In commenting on the demise of that project, the United States Congress noted that all future basic research projects of this scale should be international in scope. That is, the design, siting and funding for such facilities must be managed by an international entity. A fairly straightforward interpretation of this statement is that a smaller and smaller fraction of large research facilities will be located in the U.S. This fact provides additional urgency to studies of how collaborative tools can be optimally utilized.

This past year of the UM-ATLAS Collaboratory Project has been devoted to building relationships and connecting the groups that will be necessary both for the collaborative research to be conducted effectively, and for locating the necessary international expertise required to carry out the Project goals.

Researchers from the School of Information visited CERN in the fall of 1997. During the visit, they met with ATLAS and other LHC physicists, support personnel for the videoconferencing and web-based tools used in ATLAS and CMS, as well as members of the CERN Web Office, who hosted the visit. Following that visit, a series of bi-weekly web and video conferences have been held between CERN and SI. Recently these meetings have been expanded to include scientists from the CMS Experiment, researchers from Argonne National Laboratory and colleagues from Internet2.

Dr. Douglas Van Houweling, President and CEO of UCAID, visited CERN in the summer of 1998 to hold a series of discussions with CERN scientists concerning ways in which UCAID and CERN might cooperate in meeting the networking needs for U.S. universities associated with the LHC experiments. Those discussions are ongoing and it is likely that CERN will become an official member (and the only international member) of UCAID. This would greatly benefit not only the collaboratory R/D activities described above, but the very accessibility of CERN physics data that will be needed by the UM physicists.

In August 1998, the University of Michigan hosted one of the ATLAS Software Workshops. Collaborative experts from the School of Information made presentations on the state of work being done at UM, and impressed the audience with a demonstration of potential tools that might be designed for physicists. Dr. Van Houweling also addressed the meeting, and his presence helped forge needed connections with ESNET and DOE 2000.

Professor Daniel Atkins (SI) presenting to the ATLAS Software Workshop. Ann Arbor, Michigan

Bridges have also been built to Washington, where it has been important to inform Congressional leaders of the impact that large-scale international science will have on higher education. Professor Neal was invited to deliver testimony to the House Science Committee on International Science and, in March 1998, used the opportunity to stress the necessity for deploying collaboratory tools in the execution of large-scale international science projects. Over the past year, Dr. Neal has also had meetings with officials from the Department of Energy and the National Science Foundation to discuss the importance of developing the collaborative tools that will make large scale international scientific projects feasible.

In adding to the list of Michigan’s leadership activities that involve linking research with undergraduate education, we note the successful National Science Foundation REU Program which was introduced by the NSF in 1987 following the recommendation of the Neal Commission of the National Science Board. The Commission noted the very important benefits to be derived by having undergraduates, during the summer, work on actual scientific projects under the mentorship of distinguished faculty.

Homer Neal of Michigan and Steve Reucroft of Northeastern University launched a new program in the summer of 1998. It creates the first international component for the NSF REU. Each summer, ten students travel to CERN to spend twelve weeks working on the LHC experiments, ATLAS and CMS. This REU provides an opportunity for the brightest undergraduates from around the United States to discover the excitement of international high energy physics. Although CERN has had a summer program for four decades, it was not open to U.S. students until now. This proved to be an outstanding opportunity for a select group of U.S. undergraduates last year, and student selection for summer 1999 is now underway. Two main ingredients of the program include joining the day-to-day activities of CERN research groups, and attending a lecture series on a wide range of topics in theoretical and experimental particle physics and associated techniques. Finally, the students have ample opportunity to interact socially with the large and eclectic international community found at CERN.

We note that it is our intention to have these students participate in the collaboratory R/D programs developed within the Project.

During the next twelve months, the ATLAS Project will initiate its first long-term studies of the key issues relevant to the success of global high energy physics collaborations, in the setting of the CERN Large Hadron Collider experiments. This work will encompass the scientific observation of several working groups within the ATLAS Experiment, the evaluation of both the relevant human factors issues, as well as the tracking and evaluation of relevant emerging hardware and software technologies. Also, the networking deficiencies that continue to be a barrier to effecting global communication will be addressed through partnerships with external organizations and industrial partners.

Researchers from the School of Information will begin the task of outfitting Video Conferencing Laboratories — one at CERN and one at the University of Michigan. These laboratories will be designed and equipped with state of the art technology. While they are being used for regular meetings of physics researchers going about their routine experimental tasks, behavioral researchers and their graduate students will study the scientific interactions of those physicists. There will be technical researchers to evaluate the facility design, with the goal of iteratively enhancing the facilities as needs require. These rooms will also be used to test faculty-student interactions via video conference technology.

As previously mentioned, not all interactions are full-scale meetings; cooperative efforts between scientists on- and off-site in planning and evaluation activities can be just as critical. Thus, another facet of the project will involve technologies for workgroup collaboration. Using Internet-based collaboration tools such as WIRED, as well as other tools developed by SI, the interdisciplinary team of collaborative tool developers found at the School of Information will again assess the needs and current practices of the physicists working on the muon front end electronics. The lessons learned with this group, and the tools developed, will be deployed more widely within ATLAS and CMS during the life of the Project.

Another potential workgroup test-bed is the MONARC Project at the CERN LHC. The approach taken in MONARC is to develop and execute discrete event simulations of the various candidate distributed computing systems. There are approximately 50 researchers on this project, broken into four working groups, and their interactions will be evaluated as part of the UM-ATLAS Collaboratory Project.

In order for collaborative technologies to be effective there must be improvements to networking connections between the U.S. and Europe. This is not only important for the project at CERN, but for all global communications. Faculty from the College of Engineering, the School of Information, and colleagues from UCAID-Internet2 and CalTech, will continue to work together to develop the best possible connections with Europe. They will also track the impact and opportunities provided by the new “quality of service” technology, which allows the assignment of priorities to individual network packets.

Although the initial efforts of the UM-ATLAS Collaboratory Project are to be focussed on experimental high energy physics, and the ATLAS Experiment in particular, lessons learned from work in this area will be applied to other developing large-scale collaborations. In particular, those with an international component, where the struggle to meet global communications challenges will be critical, will benefit greatly from efforts to improve connectivity across national boundaries. As the anticipated success in this first collaboratory is reported, the Project hopes to be a valuable resource to other programs on campus, providing the necessary structure for these research endeavors to be successful.

There are several items for possible longer term research that are under current consideration. They will be discussed and evaluated in the context of the work of the new advisory committee, which is described below.

In the interim, efforts are underway to secure planning funds from the NSF Knowledge and Distributed Intelligence (KDI) Office, and from the U.S. Department of Energy.

The UM-ATLAS Collaboratory Project reports to the Provost of the University of Michigan, through its current Director, Homer A. Neal. The Project has been given support and space by the School of Information. It will be advised by an advisory committee comprised of knowledgeable individuals from a variety of disciplines at the University of Michigan.

The structural components of an initial organizational chart are presented below.

National Partnership for Advanced Computational Infrastructure. http://www.npaci.edu/ .

Publications and Reports of the National Academy of Sciences, National Academy of Engineering, Institute of Medicine and the National Research Council. http://www.nas.edu/publications/ .

Space Physics and Aeronomy Research Collaboratory (SPARC). Collaboratory for Research on Electronic Work, University of Michigan. http://www.crew.umich.edu/UARC/ .