REDMOND, Wash., July 28, 2003 — Cornell University researcher Gerd Heber used to believe that databases could largely be ignored in his work studying structural engineering failures. If he thought about databases at all, he thought of them as appropriate for a Kmart or an Amazon.com — a handy tool for retailers to keep track of customer records. Why bother with a database, he reasoned, when you can store data in binary and text files?
But Heber, a senior research associate at the Cornell Theory Center in Ithaca, N.Y., has come around to a different way of thinking. He recognizes that databases and data-mining technology offer him a unique view into data he collects for his projects — a vision he will share at the Microsoft Research Faculty Summit 2003 this week (July 27-29) at the Microsoft Conference Center.
Heber researches computational fracture mechanics. His studies of structural failures investigate everything from the hairline crack in the compression disk of a jet engine to the fissure in a concrete dam. The discipline has enormous implications for understanding, predicting — and, ideally, preventing — civil and industrial engineering catastrophes. But it relies on multi-scale computer simulations that traditional engineering tools can’t handle.
With the help of colleagues and graduate students, and collaboration from Microsoft Research, Heber has developed a mechanism to fill that gap — a real-time visualization tool that lets researchers navigate a simulation and
weaknesses in a virtual structure. The result, which Heber will demonstrate at this week’s Faculty Summit: unique insight into how and why fractures occur and propagate, and the potential to avert failures in real structures.
Summit Showcases Collaboration and the Future of Technology
According to Douglas Leland, worldwide director of the Microsoft Research University Relations group, which organized the Faculty Summit, the goal of the conference is to stimulate discussion on key computing problems and find areas for mutually beneficial collaboration.
“Our success as an industry depends on the tremendously talented people who graduate every year from university,”
“The relationship between industry and academia fuels a cycle of innovation, a ‘virtuous cycle,’ and we see it as a corporate imperative to engage in and nurture this cycle. As one of the world’s largest computer science research centers, we’re committed to working with academia to advance research and scholarship through an open sharing of ideas, resources and technology.”
This year’s fourth such summit brings together more than 340 academic researchers representing 115 institutions in 11 countries. It offers them opportunities to network with colleagues, learn about the latest Microsoft technologies, and showcase the intellectual value of collaborating with a world-class research lab. Attendees come not only to get information, but also to provide it. University researchers comprise about 40 percent of the speaker count across all sessions, and they will present half of the technology demos, reflecting the conference goal of fueling innovation through partnership.
Heber is one of 21 faculty members slated to present his work at this week’s event. Leland notes that collaborations tend to span about two years, and they typically address technology areas that blend interesting research with the potential for commercial application.
The Cornell project, along with the two other projects profiled below, will be part of an opening presentation to faculty attendees, highlighting several topics important to Microsoft and the academic community: Systems, Mobile and Wireless Computing, and Web Services. Microsoft Research hopes that research in these key areas will eventually translate into an enhanced customer computing experience.
Visualization Tool Mines Data for
What Went Wrong
The real-time visualization tool that gives researchers a new lens on fracture mechanics is the result of a year and a half of research and collaboration at the Cornell Theory Center. This innovative tool puts a graphical front end on a complex new environment in which simulation input data is read from a cluster of servers running Microsoft SQL Server. The database, which also stores huge volumes of simulation output data, runs on various multiprocessor configurations — high-performance, Windows-based server clusters that crunch hefty numbers to solve the underlying differential equations.
The tool integrates with the SQL Server cluster via Microsoft .NET technology and lets scientists interactively explore large-scale, 3-D models and underlying engineering data. This solution lets users do the computational steering. Users extract data dynamically by manipulating visual probes and selectors. Their choices are automatically translated into queries in SQL (Structured Query Language, the standard programming language for defining and accessing relational databases), and almost instantly, new features are displayed on-screen.
Heber predicts that the collaboration will make a crucial and far-reaching contribution to the future of research.
“People in the Cornell Theory Center are very interested in early-adopting new technologies to benefit their work in various ways and to do new types of research that would be impossible without the ability to successfully mine large data sets,”
“We are excited about collaborating with Microsoft Research, because it would be almost impossible to do this work without a resource, both intellectually and financially. We want to produce something not just for publication in journals or presenting at conferences, but a product we can share with the community of researchers doing computational material sciences.”
For more information on the Cornell Theory Center project, see Related Links (at right).
Research Takes Mobility to New Levels
How will you control the home entertainment systems of the future?
Even today, it is increasingly difficult to configure and control your TVs, DVD players and personal video recorders, not to mention room lighting or the digital music stored on your PC. Imagine if, rather than spending hours manually programming functions into a universal remote control, you could have your mobile phone or Pocket PC use your home’s wireless network to automatically
all of the nearby devices and their capabilities. And then, using artificial intelligence, it could design an intuitive user interface to control them all.
Brad Myers doesn’t just imagine scenes like this; he’s working to make them real. Myers, a senior research scientist in the Human Computer Interaction Institute of Carnegie Mellon University’s School of Computer Science in Pittsburgh, admits that the idea presents tough research challenges. But he’s convinced that a handheld device can and will do the hard work, allowing people to relax and enjoy the music.
The concept, called the Personal Universal Controller (PUC), uses enabling technologies from Microsoft, such as Universal Plug and Play (UPnP), to carry on a two-way conversation with a wide variety of devices for entertainment and home and office control. Because this project was designed to support a wide variety of PDAs and mobile phones, Myers and his team use Microsoft’s C# programming language and the .NET Compact Framework. With these technologies, the development team can reuse much of the software on very different types of devices.
The team from Carnegie Mellon will show their work at this week’s Microsoft Research Faculty Summit. The work is part of an umbrella project at the Pennsylvania university called PEBBLES, or
“PDAs for the Entry of Both Bytes and Locations from External Sources.”
Born six years ago as a small undergraduate project, PEBBLES has grown to encompass about 30 students and millions of dollars in support from various government and industry sources, as well as Microsoft Research.
“We couldn’t have achieved what we have without Microsoft’s support, so it’s certainly been critical to our success,”
“And it’s not just the funding, but also the collaboration — the help we’ve gotten from Microsoft researchers and the interaction with the research and product groups. Having access to some of the leading researchers and developers of desktop solutions and PDAs has enabled us to create systems that are relevant to the real needs of consumers while still presenting interesting challenges and doing state-of-the-art research.”
The PEBBLES research focuses on the many ways that handheld devices, such as Pocket PCs and Smartphones, can be used in conjunction with other technology, such as desktop PCs and home appliances, to perform tasks more efficiently. Myers says researchers have shown that handhelds can be useful in nearly any environment where other technology is around, including offices, classrooms, homes, factories, cars and military settings, as well as for certain groups, such as people with disabilities.
“Our philosophy is, we can think of neat new things for these devices to do beyond their basic function as organizers and communication devices, and since millions of people have them anyway, let’s take advantage of them,”
He considers the scope of possibilities broad. In an earlier PEBBLES project, for example, researchers built an application called Slide Show Commander that targets the business meeting setting. Also supported by Microsoft Research and slated for a demo at Faculty Summit, the program lets a Pocket PC user remotely control a PowerPoint presentation running on a laptop or other computer. The application supports the full capabilities of PowerPoint, including the use of animation and rich media, and adds the ability to draw pictures using the stylus and view notes on the handheld device.
Projects like Slide Show Commander and the PUC offer only a glimpse into the many potential applications for this field of research. And Myers, who plans to pursue this line of work as long as it remains interesting, says it’s appropriate that a successful company like Microsoft helps pay for the research and the education of students who will be members of the next generation of researchers and developers in that area.
For more information on the PEBBLES research project at Carnegie Mellon, see Related Links (at right).
Web Services and .NET Facilitate Scientific Collaboration
At the University of Virginia in Charlottesville, Va., Microsoft Research is collaborating on a project that focuses on what Marty Humphrey calls
“high-end computing for big science.”
Humphrey, an assistant professor in Virginia’s Computer Science department, is developing a set of dynamic resource-sharing policies and mechanisms whereby a short list of scientific entities can create an ad-hoc
to solve a particular problem, then dissolve.
For example, assume that three scientists — one at a national lab, the second at a private company, and the third at a foundation — want to pursue a common drug-design experiment that will take a matter of weeks. All intend to contribute resources, especially computer processing cycles to run simulations, and all want to share data from the experiments with each other, but with no one else. Humphrey’s work aims to resolve the overriding problems that this sort of heterogeneous environment typically creates, namely scheduling, resource-sharing, ease of use and security.
The work grew out of Humphrey’s background in grid computing — the middleware that makes it easier for scientists to collaborate using linked supercomputers. Grid computing originates in the national labs and supercomputing centers.
Humphrey and his team are developing grid-computing middleware that builds on the strengths of Web services and the robust foundation of Microsoft .NET, including Web Services Enhancements (WSE), especially for run-time support of next-generation security standards. WSE is an add-on to the Microsoft Visual Studio .NET development system and the Microsoft .NET Framework.
“Specifically we’re leveraging the Web services standards WS-Security, WS-Trust and WS-Security Policy,”
“These are real, core Web services standards that support the autonomy and heterogeneity that we have to address.”
The research project has kept Humphrey, five Virginia graduate students and two professional staffers occupied for about nine months, with funding supplied from the National Science Foundation and the National Partnership for Advanced Computational Infrastructure (NPACI), as well as Microsoft Research.
This week at the Faculty Summit, Humphrey will demonstrate how the collaborative software developed by his team helps a transient virtual organization run a scientific application called NAMD, a highly scalable molecular simulation, for protein-folding experimentation. The demo will show how multiple resource owners register their computers with the virtual organization, then simulate a scientist securely executing her application on the registered computer that will do the work most quickly, regardless of physical location — all facilitated by .NET. And what’s implicit in all this, Humphrey points out, is the ability to dismantle the virtual organization quickly, simply by ordering the registered hosts to, in effect, disband.
Humphrey says Microsoft Research has been instrumental in the project’s success, bringing him in touch with architectural feedback, supportive researchers and developers, quick answers, responsive tools and .NET software.
“I come from the Grid Computing background, and I recognized that .NET is a robust implementation that we could leverage in our research,”
“At UVa, my research agenda is to facilitate the easier and more secure use of resources in terms of information and cycles. The true value of working with Microsoft Research is you’re able to bring Microsoft technologies into solving that problem. And I think our value to Microsoft is we’re evaluating Microsoft’s approach and providing feedback on that approach, in particular, how it works for computational grids.”
For more information on the University of Virginia project, see Related Links (at right).
The leading-edge university research projects on display at the Faculty Summit are the latest milestones along MSR’s technology roadmap in the academic arena. They demonstrate Microsoft’s vision and support for open collaboration through tools, resources and intellectual partnership.