Institute for Mathematical Sciences Programs & Activities
High Performance and Parallel Computing for Materials Defects and Multiphase Flows
(1 January - 31 March 2015)
Jointly organized with Department of Mathematics, NUS
Organizing Committee · Visitors and Participants · Overview · Activities · Venue
Co-chairs
- Weizhu Bao (National University of Singapore)
- Weiqing Ren (National University of Singapore and Institute of High Performance Computing, A*STAR)
- Ulrich RÃ¼de (University Erlangen-Nuremberg)
Members
- Gautam Biswas (Indian Institute of Technology Kanpur)
- Hans-Joachim Bungartz (Technische UniversitÃ¤t MÃ¼nchen)
- Pierre Degond (Imperial College London)
- Song Jiang (Institute of Applied Physics and Computational Mathematics)
- David E. Keyes (King Abdullah University of Science and Technology)
- Hans Petter Langtangen (Simula Research Laboratory and University of Oslo)
- Lois Curfman McInnes (Argonne National Laboratory)
Supercomputers continue to evolve very rapidly, driven by advances in semiconductor technology. However, unlike prior generations of architectures, the increased computational power is delivered predominantly by a further dramatic increase in parallelism. Future supercomputers will be organized in complex, hierarchically structured, heterogeneous arrangements of memory and interconnection networks. Reaching exascale with at most a moderate increase in power consumption will present an enormous technical challenge and will impose dramatic architectural constraints that will also imply constraints at the algorithmic level. With this significant advance in computer hardware in terms of speed and complexity, researchers can employ new architectures to solve extremely large-scale scientific and engineering problems so as to understand nature better and to improve human life. In fact, scientific discovery and innovative engineering design depend increasingly on computational tools. High performance and parallel computing have been applied to solve numerous scientific and engineering problems such as weather prediction, airplane design, molecular dynamics, chemical reaction, and many more.
The current trends in high performance and parallel computing require increased interdisciplinary collaboration. This situation calls in particular for a greater participation of mathematicians and computational scientists to work together with scientists in the application domains to address fundamental questions related to emerging applications with multiscale and multiphysics phenomena. In fact, high performance computing systems can be used efficiently for solving application problems only when combined with proper mathematical models and suitable ultra-parallel algorithms. Thus, physically meaningful simulation software must also rest on a sound mathematical foundation for the numerical modeling of complex phenomena. Success in devising novel mathematical models and numerical methods will provide a foundation for developing more realistic high-resolution computational models that could represent the true physics with ever-increasing accuracy. It will become feasible to couple multiple scales and to represent the interaction between different physical phenomena correctly, provided that we succeed in developing and adapting the computational models and algorithms to these new heterogeneous ultra-parallel computer architectures.
The three-month program will bring computational and applied mathematicians, computer scientists, computational materials scientists and other computational scientists together to review, develop, and promote interdisciplinary research on high performance and parallel computing with applications in the simulation of materials defects, material-related processes, multiphase flows, and complex fluids. The program will provide a forum to highlight progress in a broad range of application areas, especially in the simulation of materials defects and multiphase/complex fluids, within a coherent theme and with strong emphasis on high performance and parallel computing.
The program participants will:
- present recent developments in high performance and parallel computing and their applications in simulating material defects and multiphase/complex fluids;
- provide a forum for interaction and discussion among experts in mathematical modeling, algorithm development, and supercomputing, with the goal of initiating the collaborative co-design of models, algorithms and exascale systems for progress in materials sciences and the understanding of complex flows;
- accelerate the interaction ofÂ mathematicians and computer and/or computational scientists by stimulating lively debate on important research issues related to high performance and parallel computing with applications, and promote highly interdisciplinary research with emerging applications and cross fertilization of ideas;
- develop and foster international and local collaborations of scientific researchers in high performance and parallel computing;
- help to train junior researchers and graduate students by exposing them to a broad area of mathematical knowledge as well as parallel and high performance computing environments and facilities through tutorial lectures, parallel computing practice, public lectures, research seminars, and collaborations.
The program activities will consist of three workshops, tutorial lectures, public lectures, working seminars and collaborative research.
1. Collaborative Research: 1 Jan - 31 Mar 2015
During the period of the program, the Institute for Mathematical Sciences will provide an excellent research environment for local and international visitors from computational and applied mathematics, computer science, computational materials science, computational fluid dynamics, and computational science to discuss and to conduct collaborating research.
2. Embedded meeting: 7 - 11 Jan 2015
The 9th International Conference on Computational Physics, National University of Singapore, Singapore
3. Workshop I: 12 - 16 Jan 2015
Title: Recent Advances in Parallel and High Performance Computing Techniques and Applications
Rationale: This workshop will focus on recent advances in parallel and high performance computing environments, exascale computer structures, parallel programming platform such as MPI and OpenMP, and techniques for parallelization of existing algorithms. Some other issues include techniques in parallel programming, parallel computing platforms, parallel computing libraries, and parallel computing software. Exascale enabled algorithms will be discussed for several essential applications through multiscale modeling and simulations.
4. Tutorial on Parallel Domain Decomposition Methods and Parallel Hierarchical Grid (PHG): 19 - 20 Jan 2015
5. Special Talks: 30 Jan, 6, 11, 17 Feb 2015
6. Workshop II: 9 - 13 Feb 2015
Title: High Performance and Parallel Computing Methods and Algorithms for Materials Defects
Rationale: This workshop will focus on mathematical models and exascale enabled algorithms for simulating materials defects, including dislocation dynamics, quantized vortices, grain boundary growth, solid-state dewetting problems, etc. We will focus on how to choose and develop multiscale mathematical models for these problems, to design efficient and accurate numerical methods that are highly parallelizable, and to implement these algorithms in parallel programming. Comparisons will be discussed among different algorithms, and essential applications will be reported for simulating materials defects.
7. Workshop III: 2 - 6 Mar 2015
Title: High Performance and Parallel Computing Methods and Algorithms for Multiphase/Complex Fluids
Rationale: This workshop will focus on mathematical models and exascale-enabled algorithms for simulating multiphase/complex fluids, including contact line problems, particulate flows, bubble and foam coarsening, fluid structure interaction, quantum fluids, etc. We will focus on how to choose and develop multiscale mathematical models for these problems, to design efficient and accurate numerical methods that are highly parallelizable, such as lattice Boltzmann methods, and to implement these algorithms in parallel programming. Comparisons will be discussed among different algorithms, and essential applications will be reported for simulating multiphase/complex fluids.
8. Winter School: 23 Feb - 12 Mar 2015
Title: Computational and Mathematical Methods for Materials Defects and Multiphase Flows
The Winter School brings together senior undergraduate students and graduate students to the study of modeling and simulation of materials defects and multiphase flows and applications. It consists of two weeks of tutorials and student oriented seminars. Distinguished researchers will give tutorial lectures on topics in computational and mathematical methods, physical modeling, mathematical theory, and applications related to materials defects and multiphase flows in this winter school. In addition, all participants in this winter school can attend Workshop III during the program.
9. Public Lecture
Date: |
Monday, 9 February 2015 |
Venue: |
Possibility Room, Level 5 |
6:30pm - 7:30pm |
Bubbles-Foams, Grains-Metals: Curvature Flow in Cellular Materials |
Please note that our office will be closed on the following public holidays.
- 1 Jan 2015, New Year's Day
- 19 - 20 Feb 2015, Chinese New Year
Students and researchers who are interested in attending these activities are requested to complete the online registration form.
The following do not need to register:
- Those invited to participate.
Organizing Committee · Visitors and Participants · Overview · Activities · Venue