22 THIS ARTICLE HAS BEEN PEER-REVIEWED. COMPUTING IN SCIENCE & ENGINEERING P ETASCALE C OMPUTING Global competitiveness requires a 21st century approach to educating students and preparing them to exploit next-generation technologies. The Blue Waters project aims to revolutionize undergraduate and graduate education by promoting new educational resources, models, and methods that transcend traditional boundaries of discipline and institution. Challenges and Opportunities in Preparing Students for Petascale Computational Science and Engineering C omputer modeling, simulation, scien- tifc visualization, information analysis, and advancements in computing tech- nologies are among the most signif- cant developments for advancing scientifc inquiry in the 20th century. The integration of these ca- pabilities into comprehensive computing environ- ments are providing the means for understanding and predicting the behavior of real-world natural and engineered systems based on a knowledge of science’s fundamental laws. Advances in computa- tional methods, algorithms, and computer speeds are making possible quantitative predictions and problem-solving approaches in all felds of study. Such advances directly beneft today’s global so- ciety, offering improved weather and climate predictions, healthcare and disease prevention, structural and materials design, natural resources management, new energy sources, and food pro- duction and safety. Given these benefts, compu- tational science is now considered one of the three essential pillars of science, complementing theory and experimentation/observation. Powerful new computing technologies are coming to science and engineering in the form of many-core CPUs, graphics processors, and complex petascale architectures. To fully utilize these resources, the core competencies of high- performance computing must evolve—and high-performance computing education and train- ing must evolve with it. A critical mass of expertise in “bleeding edge” HPC is something few universi- ties have; most also lack formal and informal cur- ricula to educate and train students to exploit new computing technologies for scientifc discovery and engineering innovation. Indeed, national and inter- national studies report that students are unprepared to use HPC as a research tool. 1–5 Such studies point to several gaps in student education, including dif fculty in balancing domain topics and sci- • entifc computation in a way that provides both depth and breadth; 1521-9615/09/$26.00 © 2009 IEEE COPUBLISHED BY THE IEEE CS AND THE AIP Sharon C. Glotzer University of Michigan, Ann Arbor Bob Panoff and Scott Lathrop Shodor Education Foundation