SC09 Education Program Curriculum Fair Poster Titles


Jump to:
Blue Waters | Course/Curriculum | Research/Applications

Blue Waters Project Related

B01 - Blue Waters Undergraduate Petascale Education Program
Jeff Krause, Sharon Glotzer, Bob Panoff, Scott Lathrop
Shodor, Blue Waters

The Blue Waters project aims to revolutionize computational science and engineering education by promoting new educational resources, models and methods that transcend traditional boundaries of discipline and institution. Over the next four years, through a systematic effort of materials development, faculty workshops, and student internships, we will greatly increase the nation’s capabilities to address the educational opportunities offered by petascale computing across all undergraduate fields and will broaden participation in high performance computing. Faculty from 2 year and 4 year colleges and universities are invited to contribute to the development of inter- and multi-disciplinary modules demonstrating modeling approaches and applications at the petascale. Five week-long faculty workshops a year will provide training for faculty in taking their scientific modeling to the petascale. Undergraduate internships will provide up to a year of funding for rising juniors and seniors who are interested in working with a petascale research group.

B02 - Scaling Software Using Different Parallel Paradigms
M. Edlefsen, A. F. Gibbon, B. Johnson-Stalhut, S. Leeman-Munk, G. Schuerger, A. Weeden, C. Peck
Earlham College

Blue Waters is a petascale computational resource coming on-line in 2012 at the National Center for Supercomputing Applications. This machine will be capable of sustained petaflop performance, that is over 1015 floating point operations per second, significantly more than any other known computational resource in our galaxy. Our group is participating in an on-going project to explore the scalability of common N-Body software to petascale computational levels. This involves both the core algorithms and their implementation on hardware platforms with multiple levels of shared and distributed memory hierarchies.

Two applications were used for this phase of the project: a simple N-Body simulation called GalaxSee, and an integration application using Riemann sums. By running each of these on a wide range of computational resources we explore how the existing algorithms map to the computing resources and what can be done to improve those algorithms and mappings as the available core-count increases to over 100,000 and the computing power reaches petascale levels.

B03 - Science Visualization, Collaboration, and Education through Virtual Worlds
Andrew Fitz Gibbon, Tom Murphy, Charlie Peck
Earlham College, Contra Costa College

The goal of this project is to construct the mechanisms necessary to tie together the real-time visualization of a simulation of galaxy formation into a collaborative virtual environment, ScienceSim. The underlying galaxy formation simulation software is GalaxSee, a parallelized N-Body program running on a cluster separate from the visualization server running ScienceSim.

The ability to control various aspects of a simulation and re-examine the model from different viewpoints (often simultaneously) while all participants are potentially spread throughout the world is vital to developing new forms of teaching in the digital era. Our project not only allows remote participants to observe the real-time simulation, but allows that extra-world simulation to be run on any cluster resource that has an open connection to the Internet.

B04 - Stochastic modeling of dynamical systems: The excitable membrane of nerve
Santiago Santana and Eric Jakobsson
Univ. of Illinois

The project discusses the differences between continuum models and stochastic models in calculus. Continuum models offer very good representations of natural occurring phenomenon but they do not represent everything. Stochastic representation takes into account random probability that is inherit in all real world events and thus represents more then its counter-part. I explain this by using the action potential of neurons as a specific example and show that the continuum model that Alan Hodgkin and Andrew Huxley does not properly represent the firing of action potential in neurons at threshold voltages. Creating a stochastic model of the action potential takes into account that probability of the gates opening and closing and at threshold levels mimics the true reaction of an action potential at threshold voltages.

Course and Curriculum Development

C01 - Back to the Future or Forward to the Past?
B. Bennett, R. Eimers, and M. Houseman
North Polk Jr./Sr. High School

North Polk has had a team of teachers involved in computational science since the SC00 conference. Their efforts have included integrating STELLA, Vensim, Excel, SketchUp, and Jeroo into various curricular areas.

North Polk’s introductory computer programming course changed to an introductory computational science class, which incorporates activities from Shodor’s Interactivate site, Excel spreadsheets and computer programming.

Team members have attended summer NCIS workshops on Introduction to Computational Science and curriculum writing as a follow up to attendance at several SC Education Program conferences.

C02 - Associate of Science in Computational Science (Biology)
Jean Zorko
Stark State College of Technology

Stark State College of Technology is part of a coalition of three two-year colleges working under the auspices of the Ralph Regula School of Computational Science (RRSCS) which is an initiative of the Ohio Supercomputer Center (OSC). The coalition was established as part of an NSF-ATE grant titled Computational Science Program for Ohio Community and Technical Colleges (Grant number 0703087). As a result of the efforts of this coalition, Stark State College of Technology plans to offer an Associate of Science Degree in Computational Science (Biology) upon OBR approval. The curriculum was established through a state-wide collaboration of academic and industrial advisors. Program specific courses are: Computational Science Methods, Modeling and Simulation, and Computational Biology. This program uses teaching techniques such as problem-driven learning that will better engage the student and potentially increase retention in science, technology, engineering, and mathematics (STEM areas). The competencies for each course will be met through the use of modules that contain instructor materials, instructor-led classroom activities and assessment tools. After completing the proposed program, the student will be better prepared to perform in a four-year program in terms of research, independent study, and scientific focus.

C03 - Developing Computational Reasoning in High School Science and Mathematics
S. Ragan, C. Trout, S. A. Sinex, and C. Begandy
Maryland Virtual High School, PSC

Applying computational science to problem-solving in the classroom setting gives teachers the opportunity to demonstrate to their students how exciting and dynamic science can be. However, to use computational tools effectively, teachers must become computational thinkers, meaning they need to understand how to analyze and visualize data, develop mathematical models, and use computer models. The Maryland Virtual High School and the Pittsburgh Supercomputing Center have collaborated to create a community of computationally literate middle and high school teachers of science and mathematics in western Pennsylvania. Approximately fifty teachers in the Computation and Science for Teachers program have developed their own lessons showing the linkages among hands-on experiments, scientific theory, and computational models. Our next step is to use the program as a staff development model for the Pittsburgh Science and Technology Academy for whom we are developing a series of guided-inquiry modules to help teachers infuse computational reasoning into their curriculum.

C04 - CR Inventive Thinking with Innovative Technologies
Karen North and Bonnie Bracey
CR Inventive Thinking

Computer Science is a core technology that can bridge the digital divide to engage full participation in STEM fields by under-represented groups. CS Education needs to begin in Kindergarten to improve mathematical thinking. The benefits of using CS outweigh traditional teaching methods because it brings interest back into learning through technology, and gives teachers tools to reach their students beyond traditional methods. Innovative technologies include Bee-Bot World programming and inventive thinking clubs with a focus on environmental issues.

Specialists who integrate CS into the classroom are needed. This will expand the expertise of teachers by having master CS teachers as mentors. Mentors need opportunities to dialog and receive feedback. We need a formal way to have our practice assessed. Until there are specialists to coordinate after-school programs and classroom activities, and a formalized network is established, the problem of under-represented minorities in Engineering and CS will continue.

C05 - High Performance Computing in a Small Institution
Hong Lin
Univ. of Houston - Downtown

We present a cluster computing environment established at a minority serving institution that aims to provide services for research and education in an undergraduate student body. We will address the cost efficiency of the cluster computing environment; student participation in the configuration of the lab; and research activities performed on the cluster that involves faculty/students from various departments and academic fields. Our research experiences showed that cluster computing offered a cost effective way to support undergraduate education in a small institution. The cluster centered grid at UHD greatly improved the research and education infrastructure to allow for experiences in advanced technologies. Students play active roles in multi-disciplinary research projects and are better trained for post-baccalaureate degree programs and career paths in industry. The establishment of the grid computing environment also enables research in various areas that promote inter-disciplinary and inter-institutional collaborations.

C06 - New Beowulf cluster at Calvin
Joel Adams
Calvin College

Multicore CPUs permit the construction of computationally dense Beowulf clusters, making it possible for undergraduate students at small colleges to experience tera-scale computing. Dahl.calvin.edu is a 45-node Beowulf cluster at Calvin College with 368 cores (92 Intel Xeon quad-core CPUs), 768 GB of RAM, 23.5 TB of disk space, Infiniband and Gigabit Ethernet networks, and three KVM units, all within a single 47U rack. Each of its 44 compute nodes has eight cores and 16 GB of RAM, while its head node has sixteen cores and 64 GB of RAM for memory-intensive applications. Providing 3.68 TFLOPS of performance (RPEAK) for about $150,000, Dahl is an excellent platform for giving undergraduates hands-on experience with: shared-memory parallelism by writing OpenMP programs that run on the head node, and distributed-memory parallelism by writing MPI programs that run across the full cluster. This project is supported by NSF-OCI-0722819

C07 - Learning Chemistry through Computational Activities
Roxanna Delgado, Karen Ricardo and Carlos M. Torres
Univ. of Puerto Rico

Attaining deep learning of chemistry concepts through computational chemistry activities is a project that has been developed during the last two years at UPR-RP. Through this, students have the opportunity to learn and/or do research activities, in the computational chemistry field. In the General Chemistry laboratory students have done calculations for the stabilization energy for inorganic compounds using ArgusLab. This is an open source program and they installed it in their own computer. Computational, Physical and Research Chemistry students used WebMO and the programs already installed in it, especially Gaussian. Students of the Computational Chemistry course worked with an individual project at the time they are advancing in the classes. In the Physical Chemistry laboratory a previous experiment of calculations of the excited states energies was improved considering advanced theories and basis sets. The students doing computational research are focused in simple aromatic molecules and their dimers, doing calculations of their ground state and excited state energies using advanced theories and basis sets.

C08 - Preparing Computer Science Students for Ubiquitous Parallelism
Dan Ernst
Univ. Of Wisconsin - Eau Claire

The emergence of ubiquitous parallelism raises several questions for computer science educators, mostly related to how we can best prepare students to take advantage of cheaply available parallel resources.

Recently, I initiated changes in the curriculum at the University of Wisconsin – Eau Claire in an attempt to address its parallelism deficiency. Since then, I have been sharing these changes with the ACM SIGCSE community in an attempt to encourage other institutions to follow suit.

Our approach is not to add parallel programming as a separate class, but to integrate these concepts into traditional material throughout a student’s coursework, beginning as early as “CS1”. Our goal is for students to gain both familiarity and confidence in using parallelism to their advantage. In this poster, we describe the principles we used in developing our curricular materials and provide examples of assignments that we used to integrate this material into typical existing coursework.

C09 - Institute for Chemistry Literacy through Computational Science
R. Anderson, R. Baldwin, . Block, G. Hermann, J. Sparks, S. Stephens, and J. Stewart
NCSA - Univ. of Illinois

The Institute for Chemistry Literacy Through Computational Science (ICLCS) at the University of Illinois at Urbana-Champaign is a Math Science Partnership grant from National Science Foundation to increase the chemistry literacy and chemistry-related pedagogical skills of rural Illinois high school teachers. The ICLCS immerses Fellows in new models of instruction focusing on computational tools and resources that will improve student achievement and prepare students for 21st Century careers. We do this through a intensive, multi-year summer Institutes for teachers built upon existing, successful curricula and methods, enhanced with state-of-the-art science research data and applications.

The Institute is designed to build teachers' competence and confidence in teaching chemistry, to use computational tools and methods in their curriculum, and to create a community of practice among research faculty and high school teachers working together as colleagues to improve student achievement.

C10 - Incorporating genetic algorithm optimization of radiative transfer models of embedded objects in the Small Magellanic Cloud into student coursework
David Joiner
Kean University

We are incorporating the use of a 1040-core Dell cluster into student coursework at Kean University. Students are enrolled in a course on High Performance Computing covering topics in OpenMP an MPI programming as well as independent research in Physics and Computational Mathematics. We present the problem of function optimization by genetic algorithms as a template for a student project that involves parallel computing, numerical simulation, and can easily be applied to problems in a variety of disciplines. We investigate the scalability of GA as a function of problem size, and present our results as related to fitting spectra from embedded objects.

C11 - Using Swift and OOPS on the Open Science Grid
Erin M. Hodgess
Univ. of Houston - Downtown

We have used the statistical package R, a scripting language Swift, and a protein folding program, OOPS, on the Open Science Grid. We have seen that using these packages together can yield meaningful results for moderate size experiments. We produce regression analysis of our initial results. We will demonstrate how these techniques can be used effectively in an undergraduate statistical computing course.

C12 - CMIST: Integrating Educational Technology to Teach Science, Technology, Engineering, and Math
Domaracki, J.A., Ishwad, P., Czech, J., Dittrich, M., Stiles, J.R.
Pittsburgh Supercomputer Center

Evidence has shown that static images and printed explanations of traditional textbooks are not well-suited to convey foundational scientific concepts. Science education aims to teach these concepts meaningfully and make students aware of how these concepts can be used in their daily lives. In this process, learning the basic scientific concepts during the primary and secondary education is crucial for subsequent mastery of advanced concepts. Interdisciplinary teaching approaches are needed to teach complex concepts in science education and relate them to concepts across the curriculum. To this end, we have developed the CMIST (Computational Modules in Science Teaching) program modules which incorporate modeling and simulations to build understanding of foundational and complex scientific concepts in science and math classrooms.

C13 - Computation Exploration
Robin Flaus
Pittsburgh Supercomputer Center

Computation Exploration is a high performance computing teaching module that can be incorporated into existing curricula in high school computer science classrooms. This program teaches high school students the basics of parallel programming and guides them through the development of their own parallel program. Once the students have completed their parallel programs they are introduced to various high performance computing platforms, on which they run their application. In addition to traditional HPC architectures, students are introduced to cloud computing and run their MPI program on Amazon’s Elastic Compute Cloud. Finally, students are introduced to large scale data analysis using Hadoop.

C14 - The Computational Science Education Reference Desk
Jeff Krause
Shodor

CSERD represents a national partnership to evaluate, modify, collect, tag, and (where rights can be obtained) archive the best computational model based materials; to provide mechanisms for NSDL users, content creators, and collection maintainers to add to and annotate the collection; to provide an outlet for peer reviewed publication for content creators; to provide training and technical assistance to collection maintainers; and to provide an infrastructure for educators to bind disparate learning objects into functional units tied to state and national standards. The hallmark of our effort is the rigorous Verification, Validation, and Accreditation review and the Journal of Computational Science Education (JOCSE). JOCSE promotes the use of computation in education through disseminating unique uses of computation in the classroom as well as research findings in computational science education, with submissions from both professionals and students. JOCSE utilizes internet technology and a web-based format to allow for enhanced interactivity in all publications.

C15 - Curriculum development and learning outcome assessment in a new dual-degree computational science program
J. R.Manson, R. J. Olsen and M. Sharobeam
The Richard Stockton College of New Jersey

The Richard Stockton College of New Jersey has established a new B.S./M.S. dual degree in computational science over the last few years. This poster will describe the program and curriculum and explain how it has evolved during these early years. We will discuss adaptations in response to student demand, changing technology and institutional support. We will also describe approaches to assessment which we are piloting. The learning outcome assessment tools we are piloting include the use of project portfolios and assessment rubrics and carefully constructed short assessment tests. Preliminary results are presented and discussed.

Research and Applications

R01 - Google Android
Jon Roshoko and Keith Vander Linden
Calvin College

Google Android is an exciting new platform for software development on mobile devices. Backed by Google's powerful Maps API, location-based applications can be built for scientific, educational, and a variety of other purposes.

R02 - The MapReduce Programming Model - Is it so hard to understand and use it?
Vojislav Stojkovic
Morgan State University

This work presents and discusses the MapReduce programming model and its implementations in functional programming languages such as Haskell and Cat. The MapReduce programming model is one of the most popular programming models designed to support large data sets generating and processing using concurrent, parallel, and distributed computation. The MapReduce programming model is based on the MapReduce algorithm. In addition to the input data, the MapReduce programming model requires from the user to pass two functions, map and reduce, to describe the computation task. The map function generates intermediate data sets. The reduce function combines intermediate data sets into smaller data sets. Programs written in this way may be automatically parallelized and executed on a large cluster of commodity machines. This allows programmers without experience with distributed computation to easily do distribute programming. The given solutions are very sophisticated. Each of them may be given as the student’s research project.

R03 - Evaluation of oligonucleotide rice microarrays as a high-throughput, low-cost genotyping platform
Cheryl Dunham, Steve Geislinger and Megan Sweeney
Arcadia High School, Univ. of Arizona

Rice is the most consumed food on the planet, providing the primary caloric source for millions of people. Microarrays containing probes complementary to unique, polymorphic regions of the rice genome were used to genotype large numbers of markers on a large number of individuals in a cost effective manner. O.sativa genomic DNA was directly labeled with Alexafluor 3 and ALexafluor 5 dyes and hybridized to a DNA microarray. Microarray analysis was performed using R with LIMMA and Marray. A statistic was developed and characterized to ascertain the presence or absence of given loci based on the hybridization signal intensities. The microarray assay was optimized to reduce cost to about $40/ sample including replication. Eventually, data from all 880 loci will be analyzed to infer ancestry of different chromosome segments.

R04 - An Overview of Current Cellular Modeling Systems
Michelle Medema Eisen, Stephen Matheson, and Joel Adams
Calvin College

With the recent explosion in biological modeling software, many biologists interested in modeling are sure to find themselves overwhelmed with options. This project explores current biological modeling programs and languages in an effort to consolidate the information on available programs while helping biologists interested in modeling choose a program that fits their needs. Only programs which are free for academic research, build and simulate on the cellular scale (not atomic, and not larger than one or two cells in scale), and are active or have a user base (if a program directs one to download off of www.sourceforge.net, a minimum of 100 downloads must have been reached and websites should have been updated within the last year or two) are evaluated. The focus of this project is standalone programs which do not depend on proprietary software (ie. Matlab, Mathematica, etc.) to run simulations.

R05 - Implementing a Chemical Reaction Based Programming Model
Jeremy Kemp and Hong Lin
Univ. of Houston - Downtown

Gamma Calculus is an inherently parallel, high-level programming model, which allows simple programming 'molecules' to interact creating a complex system with minimum of coding. Gamma calculus modeled programs were written on top of IBM's TSpaces middleware, which is Java based and uses a 'Tuple Space' based model for communication, similar to Gamma. A minimal parser was written in C++ to translate the Gamma syntax. This was implemented on UHD's grid cluster (grid.uhd.edu), and in an effort to increase performance and scalability, existing Gamma programs are being transferred to Nvidia's CUDA architecture. General Purpose GPU computing is well suited to run Gamma programs, as GPU's excel at running the same operation on a large data set, and potentially offer a large speedup.

R06 - Finding Optimal Solutions and Bounds for Complex Problems
A. D. Sarma, J. Hurwitz, A. Oza, R. Puttagunta, D. Tolnay, C. Zhao
Blair Magnet Team

Computationally intensive problems require computational power and elegant algorithms. Hurwitz found a lower bound for the decycling density of any regular tessellation and exact decycling densities for regular and semiregular tilings. Oza classified polynomials based on whether they guaranteed that any c-coloring of the naturals would have a pair of like-colored numbers with difference p(d). Puttagunta investigated the conditions that ensure that every point in a grid can be colored with one of c colors so that fewer than s rectangles have vertices of the same color. Zhao’s project involved distributing data usage over several servers in order to maximize the amount of utilized bandwidth. Tolnay’s software, which allows non-programmers to create their own educational cell phone games, is being tested for its efficacy in transmitting information to players. Das Sarma used Monte Carlo simulations of 1D surfaces to investigate dynamic scaling properties arising from various sets of rules for discrete particle deposition.

R07 - Visual Perception Enhancement using Animation
Ahmed Emam
Western Kentucky University

This project will develop and use a model of visual attention combined with eye tracking to drive content-based retrieval of image data to facilitate understanding and create knowledge. This project draws upon approaches from several fields of science and engineering, computer science, engineering design, and cognitive science to provide a new paradigm of data retrieval, visualization, and organization. In particular the methodology, algorithms, and tools developed in this research will stimulate new products and services in multiple disciples. The specific objectives of this project are to develop eye tracking guided model of visual attention and to extract visual features via eye tracking.

R08 - Improving the Precision of Activity Analysis on MPI Programs
Barbara Kreaseck
La Sierra University

Message passing via MPI is widely used in single-program, multiple-data (SPMD) parallel programs. Existing data-flow frameworks do not model the semantics of message-passing SPMD programs, which can result in less precise and even incorrect analysis results. We present a data-flow analysis framework for performing interprocedural analysis of message-passing SPMD programs. The framework is based on the MPI-ICFG representation, which is an interprocedural control-flow graph augmented with communication edges between possible send and receive pairs and context-sensitivity.

We demonstrate our techniques on the nonseparable analysis, activity analysis. Activity analysis is a domain-specific analysis used to reduce the computation and storage requirements for automatically differentiated MPI programs. Our experimental results show that using the MPI-ICFG data-flow analysis framework improves the precision of activity analysis and as a result significantly reduces memory requirements for the automatically differentiated versions of a set of parallel benchmarks.

R9 - Computational studies of temporal and spatial-temporal dynamic systems by undergraduate students
M. J. Laielli, J. R. Manson, R. J. Olsen and R. T. Page
The Richard Stockton College of New Jersey

This poster presents two undergraduate research studies. In the first we investigate the nonlinear dynamics of a two-variable model of star formation. The variables are the mass fractions of atomic and hydrogen molecular hydrogen in the galaxy. The primary parameters are efficiency of production of molecular hydrogen and efficiency of triggered star formation. Steady and oscillatory states are tracked as the parameters are varied using MATCONT, an integrated suite of MATLAB functions combining the capabilities of AUTO and CONTENT, both established numerical bifurcation analysis codes. The second study concerns a spatial-temporal dynamic problem. Application areas include pollution transport, weather forecasting and spatially explicit ecological models. An intelligent domain decomposition, based on the vector field of the system equations, is explained and advocated. Resulting solutions to classic transport problems are shown to be conservative, monotonic and highly accurate. In addition we demonstrate scalability of the algorithm on some high performance computer architectures.