Friends of Gene Golub gathered at Chris Johnson's and Kate Coles' home on February 29, 2008 to remember Gene.
Those attending the event in Salt Lake City were:
Nelson Beebe, Department of Mathematics, University of Utah Adam Bargteil - Carnegie Mellon University Martin Berzins, School of Computing and SCI Institute, University of Utah Mary Anne Berzins, Human Resources, University of Utah Elaine Cohen, School of Computing, University of Utah Kate Coles, Department of English, University of Utah Steve Corbato, Office of Information Technology, University of Utah Chuck Hansen, School of Computing and SCI Institute, University of Utah Chris Johnson, School of Computing and SCI Institute, University of Utah Greg Jones, SCI Institute Tom Lyche, Department of Informatics, University of Oslo Rich Riesenfeld, School of Computing, University of Utah Kris Sikorski, School of Computing, University of Utah Claudio Silva, School of Computing and SCI Institute, University of Utah Barry Weller (Gene's Cousin), Department of English, University of Utah
The February 2008 Issue of Salt Lake magazine includes a profile of groundbreaking research being conducted at the University of Utah on the problem of Autism. Advancements in brain image analysis techniques developed by SCI researchers Guido Gerig, Ross Whitaker and P. Thomas Fletcher are specifically mentioned. (print version only)
Announcing VisTrails 1.0
Fig 1: The VisTrails history tree contains a node for each version of a workflow (or pipeline) as it evolves over time. This results in a complete audit trail of the steps that were taken in a computational task.
VisTrails is a new system that provides data and process management support for exploratory computational tasks. It combines features of both workflow and visualization systems. Like many workflow systems, it enables seamless integration of loosely-coupled resources such as specialized libraries, grid, and web services. Likewise, it parallels some visualization systems by providing a mechanism to perform parameter explorations and result comparisons. But unlike these systems, VisTrails was designed to manage exploratory activities, where computational tasks are iteratively refined as users formulate and test hypotheses. A key distinguishing feature of VisTrails is a comprehensive provenance infrastructure that maintains detailed history information about the steps followed and data derived in the course of an exploratory task. VisTrails leverages this information to provide novel operations and user interfaces that streamline this process.
The paper "Querying and Creating Visualizations by Analogy", by Carlos E. Scheidegger, Huy T. Vo, David Koop, Juliana Freire, and Cláudio Silva was selected as the "Best Paper" at IEEE Visualization 2007. In this paper, the authors introduce a new framework that allows users who are not necessarily programmers to query and refine pipelines (or workflows) by analogy. They describe a query-by-example interface which allows users to construct as complex, structure-based queries (e.g., find workflows that resample a data set before extracting an isurface) by example, using the same interface used to build pipelines. They also introduce analogy as a first-class operation to create and refine pipelines. The analogy operation allows casual users to modify pipelines without having to directly edit their definitions. These features have been implemented in their open-source workflow and provenance management system called VisTrails, which can be downloaded from the SCI Website.
Steve Parker Selected Member of DARPA Computer Science Study Group
Dr. Steve Parker has been selected by the Pentagon's Defense Advanced Research Projects Agency (DARPA) to be a member of the 2008 DARPA . The program supports university researchers with research grants for up to three years, while informing them of the Department of Defense's information technology needs and priorities. Dr. Parker will receive $100,000 the first year, and potentially as much as $750,000 during the second and third years.
CIBC Collaborator Mario Capecchi Wins Nobel Prize
The NIH Center for Integrative Biomedical Computing (CIBC) and the Scientific Computing and Imaging (SCI) Institute would like to join the international scientific community in extending our heartiest congratulations to our University of Utah colleague Dr. Mario Capecchi as one of the three recipients of the 2007 Nobel Prize in Physiology or Medicine. This award recognizes the ground-breaking work of Dr. Capecchi and his scientific contemporaries, Dr. Oliver Smithies of Cardiff University in the UK and Dr. Martin Evans of the University of North Carolina at Chapel Hill, in the development of genetic targeting of embryonic stem cells in mice. This novel technique selectively alters individual genes in the mouse DNA. Through studies employing genetic targeting, medical researchers now can better elucidate the roles that particular genes play in the animal development. The application of this method has revolutionized the study of mammalian biology and contributed to the development of new animal models for numerous human diseases in addition to cancers occurring in mice.
The Scientific Computing and Imaging Institute and the Musculoskeletal Research Laboratories are proud to announce the 1.0 release of the software FEBio, "Finite Elements for Biomechanics". FEBio is a nonlinear finite element software package that is specifically designed to address problems in computational biomechanics. Some of the features of note include capabilities for contact, rigid bodies and kinematic joints, nonlinear anisotropic constitutive models, simulation of active contraction, poroelasticity, element formulations for nearly-incompressible materials and parallel solution of the linear system of equations. After extensive testing in our lab and with our collaborators, we are happy to offer this free software to the research community. FEBio is currently available for WindowsXP, MacOS/X, Suse Linux (64 bit Opteron/Athlon64) and SGI Altix (64 bit Itanium2). We would be happy to port FEBio to other Unix/Linux platforms. The FEBio distribution includes the User's Manual, Theory Manual and several test problems to verify proper operation.
Claudio Silva Receives IBM Faculty Award for Third Year Running
Dr. Claudio Silva has been honored once again with a coveted IBM Faculty Award. This award is designed to promote innovative, collaborative research and honor outstanding faculty working in disciplines of interest to IBM. It includes $30,000 to support ongoing research. Dr. Silva is being recognized for his continuing success in developing efficient rendering techniques for large-scale scientific visualization. The increase in computational power has enabled the generation of large and complex simulations. Scientists and engineers are now faced with an incredible amount of data to analyze. Despite considerable effort, techniques for the visualization of large datasets, common in scientific simulations, still take too long. In addition, most current techniques suffer from strict data size limitations due to the reliance on having the complete datasets in main memory. In order to address the interactivity and scalability requirements, the Dr. Silva and his team are developing novel out-of-core, streaming, and parallel algorithms for optimizing the visualization of large datasets. The development of scalable rendering algorithms is of key importance to the advancement of visualization, graphics, and computational science.
SCI Graduate Student Andrew Kensler Wins Mozy Coding Deathmatch
Andrew Kensler took the $10,000 grand prize in Mozy.com's Coding Deathmatch. The contest consisted of three timed rounds of competition, two preliminary rounds were held online and the final round at Mozy.com's American Fork (Utah) headquarters. Only the top eight contestants were allowed to proceed to the final round. Contestants were asked to write programs in any of nine possible languages to to solve a series of problems. The winning code was graded on it's ability to solve the problem in the least amount of execution time. The problem for the final round was to write a program that would read through a 2GB file of 20 million records of 100 bytes each. The first 8 bytes represented the key for that record and the remainder was a payload. The program had to write out a file containing just the records that would be in the 0, 1M, 2M, 3M... positions as sorted by the key values. Ties between keys were to be broken by original order in the file. Wall-clock execution time of the correct programs was used to decide the final ranking for the contest. Timings were done on a virtual machine with 1GB of RAM and 4GB of scratch disk space available.