Summary  

Education   

Visiting Professor Program

Relevance of Center's Mission to NC State

Goals and Objectives

Organization

Partnerships with Other Organizations

 

Summary

 

The advent of powerful supercomputers has led to the creation of a new scientific and engineering methodology, namely that of computational simulation of physical phenomena, in addition to the traditional methods of experimentation and analytical theory. This new methodology has opened new, very fruitful areas to quantitative investigation, where complex physical phenomena can be simulated at both macroscopic and microscopic scales, where different scenarios can be simulated, and where the influence of various parameters can be independently determined by altering and monitoring conditions in a manner that is often unreachable experimentally. New understanding and theories can thus be obtained solely or mainly through numerical simulation and their analysis. The latter often involves sophisticated visualization techniques, which have in recent years evolved as an integral part of high performance simulations.

 

For example, methods for simulating the properties and structure of materials and fluids can be conveniently divided into four types, depending on the length and time scales of primary interest:  

1.      Subatomic scale, based on application of ab initio quantum mechanical methods (density functional theory, Hartree-Fock, etc.) to systems of fundamental particles, such as electrons and nuclei.

2.      Atomistic simulation, based on Monte Carlo methods or on solutions of molecular dynamical equations for systems of molecules, including proteins, nanoparticles, etc.

3.      Mesoscale methods, e.g. self-consistent field, dissipative molecular dynamics, etc., applied to systems ranging in size from tens of nanometers to tens of microns.

4.       Macroscopic simulations, based on classical macro-scale equations of fluid mechanics, thermodynamics, etc. The equations of fluid dynamics are also used to simulate aerodynamic properties, blood flow, atmospheric and weather phenomena, the mechanical response of biomaterials and even the details of star formation.

According to recent Federal reports, such as that of the President's Information Technology Advisory Committee (PITAC), high performance computing (HPC) is one of the key high-visibility information technology (IT) areas expected to undergo rapid expansion in the next few decades. This expectation derives from continued advances in theory and the increasing reliability of theoretical descriptions and simulations in describing complex physical, chemical, biological and industrial phenomena and processes through sophisticated high-performance simulation. PITAC cites weather and climate forecasting and the design of new pharmaceuticals as important application areas as are those amenable to data mining of massive data sets. It is thus very timely to have a Center for High Performance Simulation in order to: (i) draw on the expertise already present at the various departments at NCSU, including Biochemistry, Bioengineering, Chemistry, Chemical Engineering, Civil Engineering, Computer Science, Genetics, Marine and Earth and Atmospheric Sciences, Mathematics, Mechanical and Aerospace Engineering, Materials Science and Engineering, Nuclear Engineering, and Physics; (ii) nurture interdisciplinary interactions between the basic sciences, the applied sciences, and engineering; and (iii) develop and support advanced training and education at the undergraduate and graduate levels. The Center will also facilitate competitive proposals for large Federal HPC grants. For example, while competitive pressures have led to a gradual shift of research funding to near term applied projects in many areas of Engineering, an interdisciplinary team of researchers addressing both fundamental and applied issues has a much greater chance of attracting long-term funding.

Education

 

Students in computational sciences and engineering will be the likely major beneficiaries of the Center. Because today's high performance technologies become tomorrow's mass market capabilities, the students participating in the Center's activities will acquire advanced training in the emerging high technologies of the future, and thus be able to speed up the development and transfer of these technologies for the benefit of the State and the Nation.

 

In particular, there are substantial unsatisfied needs in interdisciplinary education, which will be addressed by the Center. For example, each simulation area (1)-(4) described above is well represented by researchers in the various departments at NCSU. However, there is insufficient interaction between them at present, and there is no easy way for graduate students working in one type of simulation to learn methods used in the others. While industry recognizes the importance of simulation methods, a common complaint is that the graduates they hire are either unfamiliar with these methods or are familiar with only one narrowly specialized methodology. There is a need to bring the simulation community together and to organize courses in which graduate students become familiar with simulation methodologies that are complementary to, but outside the range of, their own specialized disciplines. The Center will offer graduate/senior undergraduate level courses in simulation methods. Initially, we envisage a two-semester series of courses covering these topics in detail. The courses would not only be video-based, but also internet-based, and would be made available within NC and to participating institutions outside NC. A potential outcome would be a graduate level textbook once the courses have been fully developed. In addition to the semester-long courses, short courses would be offered in special topics for more advanced researchers, and a research seminar series will be offered with weekly seminars during each semester.

 

Likewise, there is substantial potential for interdisciplinary courses in the various areas of computational fluid mechanics. For example, the models for weather prediction, blood flow, ground water modeling, aerodynamics and star formation use similar computational algorithms, which can be captured in a single, crossdisciplinary course. Furthermore, some of the algorithms developed for fluid flow modeling are also very effective in quantum-mechanical simulations of materials, indicating that this course could have a very diverse student audience.

 

The student members of the Center are likely to take several courses outside of their core discipline that are already taught in participating departments. For example, many of the students will take courses in computational mathematics at various levels, especially courses in numerical methods and in numerical solutions of differential equations. Equally likely, the students will want to take computer science courses that emphasize simulation methods, data structures, and parallel computing. A number of those students will want to formalize their interdisciplinary training by participating in the nationally-recognized Computational Engineering and Sciences (CES) graduate program at NCSU and obtaining a minor in this area. In this program students take a prescribed sequence of computational courses in addition to their major courses. The program has attracted Fellowship funds from the DoEd, DoE, and NSF, and it offers exciting opportunities to students interested in the computational sciences. Furthermore, the CES program has developed a very useful course in fundamentals of computer science. A successful completion of this course enables non-computer science majors to satisfy the prerequisites for taking graduate-level computer science classes. This course should be broadly useful to the students participating in the Center.

Visiting Professor Program

 

We anticipate offering Visiting Professorship positions to talented researchers from other institutions. These visits could be for one month or longer, and would be offered to researchers who might collaborate with Center researchers, or who might participate in teaching of general or special topics courses. Distinguished Visiting Professorships are also planned, in order to attract world-renowned scientists and engineers from academia, government laboratories and industry.

Relevance of Center's Mission to NC State

 

Much of NC State's reputation is due to the strength of its science and technology programs. While most of these programs are experimental by their very nature, the advent of powerful simulation tools will accelerate progress, lead to new scientific, technological and operational paradigms, enhance the learning environment and, consistent with NCSU's land-grant status, ultimately improve the quality of life of citizens of North Carolina and beyond.

Goals and Objectives

 

The goal of the Center will be to foster, pull together and further develop the formidable expertise in high performance computing and simulation that is already present at NCSU, and thus provide a synergistic multidisciplinary environment that will accelerate research, provide advanced training and education at both undergraduate and graduate levels, and serve as a resource to NCSU and North Carolina.

 

The participating faculty will be encouraged to interact, exchange ideas and ultimately team up in pursuit of team- and center-level grants from large-scale government initiatives, such as the multi-agency Information Technology for the Twenty-First Century (IT2) initiative, which will include substantial funds for grand-challenge research and simulation in science and technology, the National Nanotechnology Initiative, the upcoming initiatives in high-end computing revitalization, and from various DoD, DoE and NIH programs.

Organization

 

The administrative oversight of the Center for High Performance Simulation is provided jointly by the College of Engineering (COE) and the College of Sciences. The Center is organized along multidisciplinary thrust areas. The initial thrusts are in (i) nano- and bio-materials simulations, (ii) computational fluid dynamics (CFD), and (iii) computer science and computational mathematics. The nano- and bio-materials thrust includes faculty in Biochemistry, Chemistry, Chemical Engineering, Computer Science, Genetics, Mathematics, Materials Science and Engineering, and Physics. The CFD thrust includes members from Bioengineering, Marine, Earth and Atmospheric Sciences; Mechanical and Aerospace Engineering; Nuclear Engineering and Physics. The computer science and computational mathematics thrust will include faculty from Mathematics and Computer Science, and be crosscutting with the other thrusts. In each of the thrust areas there are already enough well-recognized faculty at NCSU to form nationally competitive teams. A common core curriculum in high performance simulation will be jointly developed by the teams. The students associated with the Center will receive high-quality interdisciplinary education in several technologically important areas. The Center will actively participate in recruiting efforts and work jointly with the relevant Departments. On a longer time-scale, expansion and consolidation of space will allow for sharing of major computational equipment and for locating of all faculty and students active in HPC in near proximity to each other.

 

 The first Director of the Center is Dr. J. Bernholc, Drexel Professor of Physics. Dr. Bernholc has authored about 200 articles in the subject area of the Center. He has served on a variety of advisory and technical committees dealing with high performance computing, nanotechnology, and computational mathematics. The Co-Director of the Center is Dr. K. E. Gubbins, W.H. Clark Distinguished University Professor of Chemical Engineering. Dr. Gubbins has published 400 articles and he is a member of the National Academy of Engineering. The Associate Director of the Center is Dr. C. Roland, Professor of Physics. Dr. Roland has published 85 papers in the areas of statistical mechanics, growth simulations and transport theory.

 

Partnerships with Other Organizations

 

When appropriate, the Center will partner with other organizations and Universities, for example in multi-center umbrella grants and activities. The Kenan Institute for Engineering, Technology and Science at NCSU has already agreed to provide important support for Center activities. Other institutions with which strong ties exist include the Oak Ridge National Laboratory, the North Carolina Supercomputing Center, the National Computational Science Alliance, and the Livermore National Laboratory.