2007-2008 Abstracts
CPSC 681 Graduate Seminar
(only required for new graduate students):
Graduate Orientation I: Overview of Department Resources & Contacts,
Honor Code, and Student Organizations
MANDATORY FOR NEW GRAD STUDENTS (but not other CSPC 681 students)
4:10-6:00 p.m., Monday August 27, 2007
Room 124, Bright Building
Abstract
This meeting will concentrate on the essentials the students will need to
settle in. It will include an introduction to departmental administration
(staff who's who, payroll, mailboxes, phones), the computing resources
(computer use/accounts, printer quotas, lab access/tours),
the academic advising staff and resources,
the TAMU honor system, TAMU Libraries, Graduate Teaching Academy, Student Engineers' Council
and relevant student organizations (CSGSA,
AWICS,
TACS
(TAMU ACM and IEEE student chapter), UPE and
TAGD).
MANDATORY FOR NEW GRAD STUDENTS (but not other CSPC 681 students)
CPSC 681 Graduate Seminar
(required for all new graduate students and all CPSC 681 students):
Graduate Orientation II:
Current Grad & Faculty Panels, Poster Session, & PIZZA!
MANDATORY FOR NEW GRAD STUDENTS. Counts as CPSC 681 seminar for CPSC 681 students.
4:10-6:00 p.m., Wednesday August 29, 2007
Room 124, Bright Building
Abstract
- 4:10-4:40 p.m. - Student Panel:
Current grad students share tips about how to succeed in graduate school.
- 4:40-5:10 p.m. - Faculty Panel:
Faculty share their ideas about what they are looking for in a graduate
student.
- 5:10-6:00 p.m. - Pizza & Current Student Poster Session - new students
can meet current grads and learn about ongoing research projects.
MANDATORY FOR NEW GRAD STUDENTS and CPSC 681 STUDENTS
CPSC 681 Graduate Seminar:
Large-Scale Analysis of Collections of Evolutionary Trees
Dr. Tiffani Williams,
Assistant Professor in Computer Science, Texas A&M University
4:10 p.m., Wednesday September 5, 2007
Room 124, Bright Building
Abstract
Phylogenetics is concerned with inferring the genealogical relationships between a group of organisms (or taxa) and this relationship is usually expressed as an evolutionary tree. However, obtaining such trees is very difficult (most approaches use NP-hard optimization criteria). As a result, most phylogenetic analyses rely on heuristics to obtain accurate (best-scoring) trees. It is not uncommon for heuristics to return hundreds to thousands of best-scoring trees.
Hence, fast post-processing techniques are needed in order to summarize effectively the relationships depicted among the evolutionary trees. In this talk, I will present new techniques to compute the topological differences of large phylogenetic tree collections efficiently. In particular, we develop a new family of randomized algorithms to compute the Robinson-Foulds distance matrix for a collection of evolutionary histories. Finally, I will discuss additional applications of our randomized approach to facilitate the reconstruction of accurate phylogenies.
Biography
Tiffani L. Williams is an Assistant Professor in the Department of Computer Science at Texas A&M University. She earned her B.S. in computer science from Marquette University and Ph.D. in computer science from the University of Central Florida. Afterward, she was a postdoctoral fellow at the University of New Mexico. Her honors include a Radcliffe Institute Fellowship, an Alfred P. Sloan Foundation Postdoctoral Fellowship, and a McKnight Doctoral Fellowship. Her research interests are in the areas of bioinformatics and high-performance computing.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Wireless Sensor Networks: A Systems Approach
Dr. Radu Stoleru,
Assistant Professor in Computer Science, Texas A&M University
4:10 p.m., Wednesday September 12, 2007
Room 124, Bright Building
Abstract
Wireless Sensor Network (WSN) systems have been used in many promising applications including military surveillance, habitat monitoring and wildlife tracking.
In this talk I will describe VigilNet, one of the major efforts in the WSN community to build an integrated system for energy efficient surveillance. This talk will also give key insights into why the node localization problem, i.e., how a node find its position, remains one of the most difficult research challenges to be solved practically. I will present the answer to this challenge, an asymmetric architecture, in which the complexity associated with node localization is removed from sensor nodes. First, I will present an event-based localization scheme, the first to achieve tens of centimeter accuracy, with virtually no additional hardware on sensor nodes, in a realistic, outdoor deployment. Elements of coding theory and the asymmetry in node localization architecture are used to develop new techniques for efficiently distributing events in the sensor network. Second, I will describe the first localization scheme that, simultaneously, executes extremely fast (milliseconds to seconds) and achieves meter level accuracy. For this image-based, passive localization system we propose a constraint based label relaxation algorithm and four primitive and hybrid constraints.
Recommended Paper:
R. Stoleru et al.: A High-Accuracy, Low-Cost Localization System for Wireless Sensor Networks
Biography
Dr. Stoleru is an assistant professor in the Department of Computer Science at Texas A&M University. He received his PhD degree from the University of Virginia in 2007, under Prof. John A. Stankovic. Dr. Stoleru's research interests include deeply embedded wireless sensor network systems, distributed computing and computer networking. He is an author and co-author of over twenty publications and was awarded the 2007 Outstanding Graduate Student Research Award from the Department of Computer Science, University of Virginia. Dr Stoleru is a member of ACM and IEEE.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Economy-based Grid Computing and the Gridbus Middleware
Dr. Rajkumar Buyya,
Associate Professor and Reader, Department of Computer Science and Software Engineering, The University of Melbourne in Australia
4:10 p.m., Monday September 17, 2007
Room 124, Bright Building
Abstract
Grid computing, one of the latest buzzwords in the ICT industry, is emerging as a new paradigm for Internet-based parallel and distributing computing. It enables the sharing, selection, and aggregation of geographically distributed autonomous resources, such as computers (PCs, servers, clusters, supercomputers), databases, and scientific instruments, for solving large-scale problems in science, engineering, and commerce. It leverages existing IT infrastructure to optimize compute resources and manage data and computing workloads. The developers of Grids and Grid applications need to address numerous challenges: security, heterogeneity, dynamicity, scalability, reliability, service creation and pricing, resource discovery, resource management, application decomposition and service composition, and quality of services. A number of projects around the world are developing technologies that help address one or more of these challenges. To address some these challenges, the Gridbus Project at the University of Melbourne has developed grid middleware technologies that support rapid creation and deployment of eScience and eBusiness applications on enterprise and global Grids.
In this seminar, we present technological evolution and key challenges in building and managing Utility Grids. We place emphasis on fundamental challenges of Grid economy, how to design and develop Grid technologies and applications capable of dynamically leasing services of distributed resources at runtime depending on their availability, capability, performance, cost, and users' quality of service requirements. We then introduce Gridbus Project R&D efforts with focus on distributed computational economy for effective management of resources. We briefly present various components of the Gridbus Toolkit and then discuss, in detail, the Gridbus service broker that supports composition and deployment of applications on utility Grids. Case studies on the use of Gridbus middleware in creation of various Grid applications (such as distributed molecular docking, high energy physics, and natural language processing) and their deployment on national and international Grids will also be highlighted.
Biography
Dr. Rajkumar Buyya is an Associate Professor and Reader of Computer Science and Software Engineering; and Director of the Grid Computing and Distributed Systems (GRIDS) Laboratory at the University of Melbourne, Australia. He was awarded the Dharma Ratnakara Memorial Trust Gold Medal in 1992 for his academic excellence at the University of Mysore, India. He received the "Research Excellence Award" from the University of Melbourne for productive and quality research in computer science and software engineering in 2005. Based on an analysis of ISI citations, the Journal of Information and Software Technology (Jan 2007 issue) ranked Dr. Buyya's work (published in Software: Practice and Experience Journal in 2002) as one among the "Top 20 cited Software Engineering Articles in 1986-2005."
Dr. Buyya has authored/co-authored over 200 publications. He has co-authored three books: Microprocessor x86 Programming, BPB Press, New Delhi, 1995, Mastering C++, Tata McGraw Hill Press, New Delhi, 1997, and Design of PARAS Microkernel. He is serving as an Associate Editor of the Future Generation Computer Systems Journal, Elsevier Press, The Netherlands. He also serves as the Chair of the IEEE Technical Committee on Scalable Computing (TCSC). Dr. Buyya has co-founded and chaired four IEEE/ACM international conferences: CCGrid, Cluster, Grid, and E-Science. He has presented over 140 invited talks (keynotes, tutorials, and seminars) on his vision on IT Futures and advanced computing technologies in several international conferences and institutions in Asia, Australia, Europe, North America, and South America. For further information on Dr. Buyya, please visit http://www.buyya.com.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Biologically-Inspired Robotics: Learning from Nature
Dr. Alfredo Weitzenfeld,
Professor in the Computer Engineering Department, Instituto Tecnológico Autónomo de México (ITAM)
4:10 p.m., Monday September 24, 2007
Room 124, Bright Building
Abstract
Biology has been an important source of inspiration in creating new technology. In the robotics realm, an extensive number of architectures have taken inspiration from animal behavior. These robotic architectures have sought to mimic animal ethology, i.e. behavior described by higher-level brain processes, and to a lesser extent animal neuroethology, i.e. behavior mapped to underlying neural structures. From a neuroscientific perspective, advances in brain theory have provided an ever increasing understanding of the underlying mechanisms involved in aspects such as memory, adaptation and learning. To model such systems we have a developed a multi-level schema and neural networks approach that we now apply to the development of biologically inspired robotic architectures. We expect these advances to help develop new generations of adaptive robotics systems having an increasing impact on real world applications.
The work to be presented in this talk overviews a number of biologically-inspired robotic architectures developed by our group. A number of models and corresponding robotic architectures are presented describing animal behaviors such as prey acquisition, predator avoidance and exploration, based on studies from praying mantis, frogs and toads, rats and monkeys. This work has been funded by a number of national and international research collaborations.
Biography
Alfredo Weitzenfeld is a Professor at the Computer Engineering Department in Mexico's Autonomous Institute of Technology (ITAM) where he directs the BioRobotics (CANNES -
http://cannes.itam.mx) and Robotics Laboratories (http://robotica.itam.mx/ingles/robotica/directores.phtml). He is currently on sabbatical leave as a Visiting Professor at the Department of Computer Science and Engineering at the University of South Florida (USF -
http://www.cse.usf.edu/USL). He obtained his BS in Electrical Engineering from Technion, Israel Institute of Technology, an MS in Computer Engineering and PhD in Computer Science both from the University of Southern California (USC) where he later stayed as a Research Assistant Professor. He has participated in various research collaborations funded in Mexico by CONACYT (Mexico main research funding agency) and by national and international funding agencies including NSF and UC MEXUS in the US, and LAFMI in France. He is a member of Mexico National Research System (SNI). He is an ACM and IEEE Senior Member. He is a charter member of the IEEE Latin American and Mexican Robotics Councils. He currently chairs the IEEE-RAS Mexico Robotics Chapter (http://ieeerasmexico.itam.mx) and co-chairs the IEEE-RAS Standing Committee for Chapters and International Activities. He has been recently appointed as an IEEE-RAS Distinguished Lecturer for 2007. He is the founder and current chair for the 2007 Latin American Robotics Symposium (LARS) technically co-sponsored by IEEE-RAS. He is the director of the Eagle Knights RoboCup soccer teams at ITAM (http://robotica.itam.mx/ingles/index.phtml), the current Latin American champion for both the small-size and four-legged leagues. He recently founded and currently directs the USF RoboBulls RoboCup small-size team. He is the main author of The Neural Simulation Language NSL: A System for Brain Modeling (coauthors M. Arbib and A. Alexander) published in 2002 by MIT Press and the author of Object Oriented Software Engineering with UML, Java and Internet published in 2004 by Thomson Learning. He is the main designer of the Neural Simulation Language (NSL -
http://www.neuralsimulationlanguage.org &
http://nsl.usc.edu), the Abstract Schema Language (ASL), and the Mobile Internet Robotics system (MIRO).
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Geometric Algorithms in Automated Manufacturing
Dr. A. Frank van der Stappen,
Associate Professor in the Department of Information and Computing Sciences, Utrecht University in The Netherlands
4:10 p.m., Wednesday September 26, 2007
Room 124, Bright Building
Abstract
Automated manufacturing systems commonly operate in highly-structured environments. They involve modest sensing capabilities and feature relatively simple physical actions performed by hardware components that are often simpler than those encountered in traditional robotic solutions. These characteristics bear a promise of reliability, easy reconfigurability, low cost, and suitability for automated system design. It turns out that system design poses algorithmic rather than mechanical challenges.
Geometry plays a major role in the development of effective solutions for automated manufacturing tasks. We will focus on two fundamental tasks in manufacturing, part holding and part feeding, and see how geometric concepts and techniques aid in all stages of the design process, from the analysis of part behavior to the actual algorithmic design of all solutions for the manufacturing task at hand.
Biography
Frank van der Stappen received the PhD degree from Utrecht University in 1994 and MSc and PDEng degrees from Eindhoven University of Technology in 1988 and 1990. Currently, he is an associate professor at the Department of Information and Computing Sciences at Utrecht University in the Netherlands. His current research focuses on the use of geometric techniques in the solution of problems involving manipulation and motion. He has served on various PCs in robotics and automation, and on the editorial board of IEEE T-ASE. He is an associate vice-president for technical activities for the IEEE Robotics and Automation Society, and the program leader of Utrecht University's MSc program on Game and Media Technology.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
An Automated Method for Estimating Missing Well Log Zones Using Neural Networks
Dr. Yingwei Yu,
Specialized Software Developer at Seismic Micro-Technology, Inc.
4:10 p.m., Wednesday October 10, 2007
Room 124, Bright Building
Abstract
A neural network-based algorithm is presented that estimates missing intervals of log curves using log curves from the same borehole and other nearby wells. A unique feature of this algorithm is the extensive analysis and preprocessing of the candidate log curves to determine the set of curves, as well as log data samples within each curve, that will yield the best estimate of the missing log interval. This refined set of curves and data samples are used to train a general regression neural network (GRNN) which then estimates the missing log interval. An additional output of this analysis is a confidence value for each estimated log sample which provides a qualitative measure of the accuracy of the neural network prediction. In experiments where we assumed certain intervals of an existing log were "missing", the new technique provides log data estimates that are within 2% to 6% of the correct sample values.
Biography
Yingwei Yu received his B.Eng. degree in computer science from Beihang University (BUAA), Beijing, China, in 1997, and the Ph.D. degree in computer science from Texas A&M University, in 2006. From 2002 to 2006, he was advised by Dr. Yoonsuck Choe. His research included cognitive modeling and computational neuroscience.
Currently, he is a specialized software developer at Seismic Micro-Technology, Inc. His current research interest focuses on applying neural network and pattern recognition technologies in geoscience applications.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Breaking the Deadlock!
Dr. Jianer Chen, Yang Liu, and Songjian Lu,
Department of Computer Science, Texas A&M University
4:10 p.m., Wednesday October 17, 2007
Room 124, Bright Building
Abstract
Deadlock problem is among the most well-known problems studied in computer science and information management. A system cannot proceed when a deadlock occurs. Thus, a set of processes in the system needs to be rolled back in order to recover from the deadlock. It is desired that only a small set of processes is rolled back because process rolling back is expensive. Theoretically, the problem is formulated as the Feedback Vertex Set problem, which is a well-known NP-complete problem. It has been an outstanding open problem (explicitly listed as an important open problem in more than 20 published papers in the literature) in algorithmic research whether there is a polynomial time algorithm for the problem if the set of rolled back processes is small.
In this talk, we resolve this open problem and present an O(n4) time algorithm for the problem when the set of rolled back processes is bounded. The talk will be split into three sub-sessions: (1) Jianer Chen will give a historical overview of the problem; (2) Songjian Lu will present an algorithm solving a related multi-way cut problem; and (3) Yang Liu will give the algorithm for the Feedback Vertex Set problem by showing how the problem is reduced to the problem discussed in (2).
This is a joint work with B. O'Sullivan and I. Razgon at University College Cork, Ireland.
Biography
Jianer Chen is a professor, and Yang Liu and Songjian Lu are Ph.D. students, all with the Department of Computer Science, Texas A&M University. They have co-authored and published more than a half dozen papers in parameterized algorithms. The current talk is part of their research project.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Creative and Expressive Systems: Frameworks, Methods, Results
Dr. Andruid Kerne, Assistant Professor in Computer Science, Texas A&M University
4:10 p.m., Wednesday October 24, 2007
Room 124, Bright Building
Abstract
As we move deeper into the digital era, computers can be used for more than the automation of rote tasks. The goal of research into creative and expressive systems is to augment some of the most exciting parts of human life, namely how we form and express ideas and intentions, how we learn and innovate. We develop media semantic models, interactive interfaces, open source development frameworks, creative and expressive systems, and evaluation modalities. Information discovery addresses how people create new ideas while finding information. combinFormation is an information discovery tool that integrates browsing, searching, and collecting in a visual composition space, with the assistance of software agents. The grounded and objective information discovery evaluation methodology for creative systems involves quantitative measurement of the emergence of new ideas, as well as qualitative investigation of how new ideas are formed. Location-aware non-mimetic simulation games, based on fire emergency response practice, teach team coordination in fun, operational environments. They leverage the unique affordances of embodiment in human-to-computer and human-to-human interaction. Motion-tracking interfaces recognize gestures, while providing visual affordances that represent the present state of the system, and opportunities for interaction.
Biography
Andruid Kerne is a researcher working at the intersection of arts and sciences. He is an assistant professor of Computer Science at Texas A&M University, where he directs the Interface Ecology Lab [http://ecologylab.cs.tamu.edu] and the Perceptive Sensory Systems Lab, and serves in the Center for the Study of Digital Libraries. The Interface Ecology Lab investigates human-centered computing support for expression, creativity, and social engagement. Andruid holds a B.A. in applied mathematics / electronic media from Harvard, an M.A. in music composition from Wesleyan, and a Ph.D. in computer science from NYU.
Kerne's output has been presented by the Guggenheim Museum (New York), ACM SIGGCHI, SIGGRAPH, Multimedia and Document Engineering, New York Digital Salon (New York, Spain, London, Beijing), ISEA (Paris, San Jose), Computational Semiotics in Games and New Media (COSIGN), the Milia New Talent Competition (Cannes), the Ars Electronica Center (Linz), the Boston Cyber Arts Festival, the Pan-African Theater Festival (Ghana), and the town square of the village of Anyako (Ghana). His work has been supported by the National Science Foundation, the Rockefeller Foundation, Dance Theater Workshop, the Spaulding-Potter Fund for Innovative Education, and the Texas A&M Department of Computer Science, Arts Foundation, and Humanities Informatics Initiative.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Spam 2.0: Challenges and (Some) Solutions for Web-Based Open Systems
Dr. James Caverlee, Assistant Professor in Computer Science, Texas A&M University
4:10 p.m., Monday October 29, 2007
Room 124, Bright Building
Abstract
As the Web has grown and increasingly become the primary portal for sharing information and supporting online commerce, there has been a rise in efforts to pollute the Web with low-quality and dangerous Web content and to manipulate how users view and interact with the Web.
With the increasing reliance on Web-based marketplaces, social networks, and large-scale information sharing communities, individuals and their computer systems are at risk for abuse and exploitation at the hands of malicious actors. These vulnerabilities degrade the quality of information on the Web and place the user at risk for exploitation.
In this talk, I will highlight some of the unique challenges facing the Web and online communities, discuss some of the recent successes we've had in developing spam-resilient algorithms, and provide a roadmap of future research opportunities. In particular, I will discuss our efforts in building spam-resilient algorithms that have applications to Web search engines. Since ranking systems (like those offered by search engines) play a central role in organizing the Web for the vast majority of Web users, Web spammers spend a considerable effort on manipulating the underlying algorithms that drive these ranking systems. In this talk, I will describe a source-centric model for Web ranking that promotes a hierarchical abstraction of the Web graph based on the strong Web link structure. Complementary to the traditional page-based view of the Web, the source approach naturally limits spam opportunities and incorporates a novel notion of influence throttling for countering the influence of spammers.
Relevant paper: J. Caverlee, S. Webb, and L. Liu.
Spam-Resilient Web Rankings via Influence Throttling. In the 21st IEEE International Parallel and Distributed Processing Symposium (IPDPS) 2007.
Biography
Dr. Caverlee is an assistant professor in the Department of Computer Science at Texas A&M University. His research interests are generally in the areas of Web and Distributed Information Management, with an emphasis on: (1) Spam-Resilient Web-Scale Computing; (2) Web Information Retrieval and Management; and (3) Enterprise Computing and Workflow Management. Dr. Caverlee graduated magna cum laude from Duke University in 1996 with a B.A. in Economics. He received the M.S. degree in Engineering-Economic Systems & Operations Research in 2000, and the M.S. degree in Computer Science in 2001, both from Stanford University. Dr. Caverlee earned his Ph.D. from Georgia Tech in 2007 under Prof. Ling Liu.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Mobile Collaboration Applications
Dr. Surendar Chandra, Assistant Professor in Computer Science and Engineering, University of Notre Dame
4:10 p.m., Monday November 19, 2007
Room 124, Bright Building
Abstract
Wireless laptops are gradually replacing desktops as the primary computing
platform for many users. Traditionally laptops were dependent on services
provided by a wired infrastructure. Mobile laptops may be offline for long
durations making dependence on infrastructure problematic. Our goal was to
understand the viability of collaboration among wireless users without depending
on the wired infrastructure. Earlier research efforts did not have the benefit
of the critical mass of available wireless devices for their analysis.
We used the observed mobile user behavior in two medium sized universities,
Notre Dame and Dartmouth, as the basis for our study. We observed that wireless
devices tended to exhibit short durations during which they were available
followed by extended durations when they were not available. The churn frequency
itself was not high. The node availability exhibited diurnal distribution with
far fewer nodes available early in the morning. However, the temporal
consistency values were high: both for analyzing the same users availability
behavior or for any two pairs of users. Users who were part of the high
consistency set can provide better collaborative services.
Next, we analyzed the epidemic propagation rates for varying group sizes. We
showed that the propagation can exhibit large delays. On average, a single
update can reach about 60% of the collaborators in about 24 hours while
reaching 90% of the members in over ten days. We improved these propagation
durations by carefully selecting the forwarding peers using local information.
Unlike Vahdat et al. who used a random node mobility model, our realistic
analysis could not achieve 100% propagation rates even after ten days.
Biography
Surendar Chandra is an assistant professor in the Computer Science and
Engineering department at the University of Notre Dame. His research interests
are directed towards operating system topics in multimedia, storage, security,
networks and sensor systems. He received his Ph.D. in Computer Science from Duke
University under the supervision of Carla Schlatter Ellis. He spent 2000-2002 as
a faculty member at the University of Georgia. His work was supported by the
Defense Intelligence Agency, HP, the National Science Foundation and Yamacraw. He
is the recipient of an NSF CAREER award.
CPSC 681 Faculty Contact:
Dr. Bjarne Stroustrup (bs [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Experiments in the Automatic Generation of Subject Indices for Books
Dr. Robert A. Metzger,
Convey Computer Corporation
4:10 p.m., Wednesday January 30, 2008
Room 124, Bright Building
Abstract
Generating a list of important words and concepts for the subject index of a
book is a challenging task.
As a result, people who earn their living creating subject indices have no fear
of being made obsolete
by any existing software product.
This presentation describes experiments in developing a software system that may
become competitive with
professional indexers. The approach uses statistical parsing and semantic
databases to generate subject
indices from book texts. The control for the experiment includes indices
prepared both by a professional
indexer and the author of the book. The limitations of the approach are
explained, as well as improvements
that will be used to overcome them.
Biography
Robert Metzger developed and managed the development of compilers, programming
tools, and system administration
software for high performance computers at Convex Computer Corporation and the
Hewlett-Packard Company from 1986 to 2007.
He currently designs optimizing compilers at Convey Computer Corporation, a startup
company in Richardson, TX,
which is developing innovative computer systems for the high performance
computing marketplace.
He is the author of Debugging by Thinking: A Multidisciplinary Approach
(Elsevier Digital Press, 2004)
and co-author of Automatic Algorithm Recognition and Replacement: A New Approach
to Program Optimization,
(MIT Press, 2000).
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
tba
Dr. Vivek Sarkar,
E.D. Butcher Professor of Computer Science, Rice University
4:10 p.m., Monday February 4, 2008
Room 124, Bright Building
Abstract
This decade marks a resurgence for parallel computing with high-end systems moving to petascale and mainstream systems moving to multi-core processors. Unlike previous generations of hardware evolution, this shift will have a major impact on existing software. For petascale, it is widely recognized by application experts that past approaches based on domain decomposition will not scale to exploit the parallelism available in future high-end systems. For multicore, it is acknowledged by hardware vendors that enablement of mainstream software for execution on multiple cores is the major open problem that needs to be solved in support of this hardware trend. These software challenges are further compounded by an increased adoption of high performance computing in new application domains that may not fit the patterns of parallelism that have been studied by the community thus far.
In this talk, we compare and contrast the software stacks that are being developed for petascale and multicore parallel systems, and the challenges that they pose to the programmer. We discuss ongoing work on high productivity languages and tools that can help address these challenges for petascale applications on high-end systems. We also discuss ongoing work on concurrency in virtual machines (managed runtimes) to support lightweight concurrency for mainstream applications on multicore systems. Examples will be give from research projects under way in these areas including UPC, CAF, Java Concurrency Utilities, and X10. Finally, we outline the new Habanero research project at Rice University that aims to unify elements of the petascale and multicore software stacks so as to produce portable software that can run unchanged on a range of homogeneous and heterogeneous multicore systems.
Biography
Professor Vivek Sarkar conducts research in programming languages, program analysis, compiler optimizations and virtual machines for parallel and high performance computer systems. His past projects include the X10 programming language, the Jikes Research Virtual Machine for the Java language, the ASTI optimizer used in IBM's XL Fortran product compilers, the PTRAN automatic parallelization system, and profile-directed partitioning and scheduling of Sisal programs. He is in the process of starting up the Habanero Multicore Software project at Rice University which spans the areas of programming languages, optimizing and parallelizing compilers, virtual machines, and concurrency libraries for homogeneous and heterogeneous multicore processors.
Vivek became a member of the IBM Academy of Technology in 1995, an ACM Distinguished Scientist in 2006, and the E. D. Butcher Professor of Computer Science at Rice University in 2007. Prior to joining Rice University in July 2007, Professor Sarkar was Senior Manager of Programming Technologies at IBM Research. His responsibilities at IBM included leading IBM's research efforts in Programming Model, Tools, and Productivity in the PERCS project during 2002 - 2007 as part of the DARPA High Productivity Computing System program. Vivek holds a B.Tech. degree from the Indian Institute of Technology, Kanpur, an M.S. degree from University of Wisconsin-Madison, and a Ph.D. from Stanford University. In 1997, he was on sabbatical as a visiting associate professor at MIT, where he was a founding member of the MIT RAW multicore project.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
CGAL: The Open Source Computational Geometry Algorithms Library
Dr. Sylvain Pion,
INRIA Sophia Antipolis, France
4:10 p.m., Wednesday March 5, 2008
Room 124, Bright Building
Abstract
In this talk, I will introduce the CGAL project. CGAL is a 10 year old Open Source project aiming at building a Computational Geometry Algorithms Library in C++.
We will begin with a general presentation of the project: its mission statement, history, internal organization, partners, and some illustrative numbers; followed by a survey of CGAL's content: the available data structures and algorithms, as well as examples of how and by whom they are used. Topics include triangulations, Voronoi diagrams, Boolean operations on polygons and polyhedra, arrangements of curves and their applications, mesh generation, geometry processing, alpha shapes, convex hull algorithms, operations on polygons, search structures, interpolation, shape analysis, fitting, distances, and kinetic data structures. I will then show how we use the generic programming paradigm; CGAL data structures are C++ template classes and functions, usually taking several template parameters with default values for ease of use. This gives developers an incredible flexibility to adapt the data structures to their needs, which is important internally for code reuse and important for end users, as they typically integrate CGAL in already existing applications. Parts of CGAL are also interfaced with languages and software like Python, Java, Scilab, Qt and the Ipe drawing editor. I will also mention the key use that CGAL makes of the exact geometric computing paradigm in order to achieve robustness. I will then present how to make geometric algorithms correct, robust, and fast by combining floating point arithmetic with exact arithmetic and clever filtering mechanisms to switch between these two modes. These mechanisms can be used for geometric predicates, as well as for geometric constructions, which instead of a discrete return value generate new geometric entities.
Biography
Dr. Pion became involved in the CGAL project while working towards his PhD, which he received in 1999 at INRIA. He then worked on providing generic solutions to numerical robustness issues arising in geometric algorithms. Later he worked on the efficiency of some fundamental geometric algorithms such as 3D Delaunay triangulations. He is now also involved in C++ standardization, and is working on parallel geometric algorithms. Dr. Pion is a researcher at INRIA, and is the current chair of the CGAL Editorial Board.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
They REALLY Are Coming For You: How Rescue Robots Can Save Lives, Boost the Economy, and Change the Way We Think About Autonomy
Dr. Robin R. Murphy,
The University of South Florida
4:10 p.m., Monday March 24, 2008
Room 124, Bright Building
Abstract
Since the watershed year of 1995 with both the Kobe Earthquake and the Oklahoma City Bombing, our research has focused on rescue robotics. This talk will review our experiences at seven actual disasters, including the World Trade Center, Hurricane Katrina, and the Crandall Canyon Utah mine collapse, where we have deployed the first unmanned ground, aerial, and water surface vehicles plus our current directions in victim management. Our research suggests that poor human-robot interaction is the fundamental limiting factor in rescue robotics, not the devices' mechanical or computational capabilities, and that the taskable agent view of autonomy is inappropriate for emergency response applications. We believe shared autonomy, where the responder projects his/her remote presence into the incident via robots and sensor networks, is the key to the successful mixed teams for emergency response. As military robots, especially small unmanned aerial systems, become increasingly commonplace, the need for civilian homeland security robots are expected to create a multi-billion dollar market derived primarily from start up companies. The success of these systems will depend on university research and industry partnerships, such as the NSF Safety Security Rescue industry/university cooperative research center.
Biography
Robin Roberson Murphy is a Professor in the Computer Science and Engineering Department at the University of South Florida with a joint appointment in Cognitive and Neural Sciences in the Department of Psychology. She is a founder and international leader in both rescue robotics and human-robot interaction with over 100 publications in those areas and was recognized by TIME Magazine in 2004 as an innovator in artificial intelligence. She introduced ground, air, and sea robots to disaster response, and has been honored with the NIUSR Eagle award and the US Air Force Exemplary Civilian Service medal. Dr. Murphy serves on the NSF CISE Advisory Council and recently completed a term with the DARPA Information Science and Technology study group.
CPSC 681 Faculty Contact:
Dr. Nancy Amato (amato [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Tools and Techniques for Performance Analysis on Intel Processors
Dr. Ramesh Peri,
Intel
4:10 p.m., Wednesday March 26, 2008
Room 124, Bright Building
Abstract
Intel Core 2 processors are based on a new micro-architecture with a short pipeline and many enhancements to the hardware performance monitoring unit (PMU). In the first part of the presentation I will talk about some features of core 2 architecture from the point of view of a performance analysis and show the methodology and tools (Intel performance tuning utility available at http://whatif.intel.com) for effectively utilizing the PMU to analyze the performance of applications. In the second part of the presentation I will talk about Pin - a dynamic instrumentation system (available at http://rogue.colorado.edu/pin) from Intel and a variety of tools built using this system for understanding the behavior of applications.
Biography
Dr. Ramesh Peri is a Principal Engineer at Intel in the Developer Products Division and is managing a group of engineers producing various performance and program analysis tools based on sampling and instrumentation techniques. Dr. Peri received his Ph.D from University of Virginia in 1995 and worked at Hewlett Packard, Lucent Technologies and Panasonic AVC Labs in the area of software development tools before joining Intel in 1999.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Concepts and Generic Programming in C++0x
Dr. Bjarne Stroustrup,
Texas A&M University
4:10 p.m., Monday March 31, 2008
Room 124, Bright Building
Abstract
C++ templates are immensely flexible and the basis of most modern C++ high-performance programming techniques and of many elegant library designs. They are the key language feature behind the standard library's algorithms and containers: the STL. However, they can also be tricky to use, cause spectacularly bad error messages when misused, and sometimes require unreasonable amounts of code to express apparently simple ideas. C++0x will address these issues directly, and the key to resolving the problems with templates without loss of flexibility or loss of performance is "concepts." Concepts provide a type system for C++ types and for combinations of C++ types and values. Thus, we are able to provide what feels a lot like conventional type checking for template arguments (including simple and elegant overloading based on template arguments). This presentation explains the notion of concepts and shows how to use concepts to write clearer and more robust generic code using templates. People who can't wait for C++0x before trying out concepts (and other new C++0x features related to generic programming) can try the proof-of-concept implementation, ConceptGCC.
Biography
Bjarne Stroustrup
is the designer and original implementer of C++
and the author of "The C++ Programming Language"
(1st edition 1985, 2nd edition 1991, 3rd edition 1997, "special" edition 2000) and The Design and Evolution of C++.
His research interests include distributed systems, design, programming techniques,
software development tools, and programming languages.
Dr. Stroustrup is the College of Engineering Chair Professor in Computer Science at
Texas A&M University.
He retains a link with AT&T Labs - Research
as an AT&T Fellow. He was elected member of The National Academy of Engineering in 2004.
He was given the IEEE Computer Society's 2004
Computer Entrepreneur Award
and was awarded the 2005
William Procter Prize for Scientific Achievement from Sigma Xi (the scientific research society).
He is an AT&T Bell Laboratories Fellow and an AT&T Fellow.
He is actively involved in the ANSI/ISO standardization of C++.
Recipient of the 1993 ACM Grace Murray Hopper award.
Dr Stroustrup is an ACM fellow, an IEEE Fellow, as well as being a member of the Texas Academy of Medicine, Engineering, and Science.
Born in Aarhus Denmark 1950.
Cand.Scient. (Mathematics and Computer Science), 1975,
University of Aarhus
Denmark.
Ph.D. (Computer Science) 1979,
Cambridge University, England.
His non-research interests include
general history, light literature, photography, hiking and running,
travel,
and music.
He lives in College Station (Texas, USA) with his wife;
their daughter is a medical doctor and their son is a graduate student studying systems biology.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
Automatic Grainsize Determination and Cache Affinity in TBB
Dr. Arch Robison,
Intel Corporation
4:10 p.m., Wednesday April 2, 2008
Room 124, Bright Building
Abstract
Intel Threading Building Blocks (Intel TBB) is a C++ library for shared-memory parallel programming. This talk covers the design of two recent features: automatic grain size determination and cache affinity. Their implementation is interesting because TBB's loop templates are generic and work on any recursively divisible iteration space, not only flat 1D iteration spaces. Furthermore, the underlying execution model is Cilk-style nested parallelism with work stealing, not OpenMP-style flat parallelism.
The automatic grainsize feature is a heuristic that automatically determines when recursive subdivision should stop. The capability is similar to OpenMP's "guided scheduling" heuristic, but done in the context of recursive subdivision.
The cache affinity feature accelerates algorithms that repeat parallel sweeps over a collection. At a high level, the programmer specifies that one or more loops should reuse mappings of loop iterations to threads. At a low level, the loop implementation hints that a task should run on the same thread as an earlier similar task. When choosing where to run such a task, a fundamental tension arises between using an available idle thread versus using the hinted thread. To resolve the tension, the loop templates deliberately delay spawning of some tasks. The performance results show that such procrastination improves performance.
Biography
Arch is the architect of Threading Building Blocks. He was the lead developer for KAI C++. At Shell he worked on seismic imaging on a 256 node nCUBE. He has a Ph.D. in computer science from the University of Illinois.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
The Art of Parallel Programming: from Algorithms to Systems
Dr. Calin Cascaval,
IBM T.J. Watson Research Center
4:10 p.m., Wednesday April 9, 2008
Room 124, Bright Building
Abstract
The development of scalable and efficient parallel applications
requires the ability to express parallel algorithms, design parallel
data structures, and map the code to parallel architectures.
Programmer productivity is impacted by program development and tuning
environment, such as the programming language and tools used. The
efficiency of execution is determined by the support across the
execution stack, from compiler, to runtime system and libraries.
In this talk we present our work using Partition Global Address Space
(PGAS) languages, and we shall focus on UPC. We discuss the advantages
and limitations of the PGAS programming paradigm, and demonstrate how
it can be tuned to exploit large scale architectures, such as Blue
Gene/L.
Biography
Dr. Calin Cascaval is a Research Staff Member and Manager of the
Programming Models and Tools for Scalable Systems at the IBM TJ Watson
Research Center. After obtaining his PhD from University of
Illinois at Urbana-Champaign in 2000, Dr. Cascaval joined the BlueGene
project at IBM Research. He developed the chip simulator for the
Cyclops processor and worked on system software for the BlueGene/C
machine. Since 2003, Dr. Cascaval has been leading the Compilers team
in the PERCS project (IBM's entry in the DARPA HPCS program). As
manager of the Programming Models and Tools Group, he is leading the
development of the IBM xlUPC compiler and the Continuous Program
Optimization project. The work on the xlUPC compiler demonstrated that
shared memory style programs can scale to hundreds of thousands of
nodes on the Blue Gene/L machine. These results were recognized
internationally by awards at the HPC Challenge Class 2 Competition for
two consecutive years. He is the project lead for evaluating
concurrency constructs (such as Transactional Memory) in IBM Research.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
CPSC 681 Graduate Seminar:
OpenMP: Beyond 3.0
Dr. Barbara Chapman,
Department of Computer Science
The University of Houston
4:10 p.m., Monday April 21, 2008
Room 124, Bright Building
Abstract
Dual-core machines are widely marketed for home computing. Systems with a larger number of cores are deployed in the server market. Some cores are capable of executing multiple threads. Clearly, the future is multi- and many-core, as well as many-threaded. It is also heterogeneous, as a variety of devices such as accelerator boards and GPUs may be configured on a single board. In future, high-end applications will need to execute on complex parallel platforms whose nodes will be heterogeneous and multicore.
OpenMP is a shared memory API that was designed to be a portable parallel programming interface for SMPs. It is increasingly expected to be a programming model of choice for exploiting the capabilities of multicore machines. In this presentation, we review the status of OpenMP, including the enhancements to its functionality in version 3.0, and consider how it may need to evolve to meet the programming challenges posed by emerging computer platforms.
Biography
Dr. Chapman is a native of New Zealand who studied Mathematics and Computer Science in her home country, Germany and Northern Ireland, where she completed her Ph.D. on software support for distributed memory programming. She has been engaged in research on parallel programming languages and compiler technology for more than 15 years. In 2001, Dr. Chapman founded cOMPunity, a not-for-profit organization that enables research participation in the development and maintenance of the OpenMP industry standard for parallel programming. Since that time, she has been involved in the evolution of OpenMP. Her research group at the University of Houston has developed OpenUH, a reference compiler for OpenMP that is also used to explore language and compiler techniques for multithreaded programming.
CPSC 681 Faculty Contact:
Dr. Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
