Computational Sciences and Engineering (CSE): A New Paradigm in Scientific Research & Education. Abul K. M. Fahimuddin

Transcription

1 Computational Sciences and Engineering (CSE): A New Paradigm in Scientific Research & Education Abul K. M. Fahimuddin Scientific Research Staff Germany

2 Motivation: Chemical Dispersion in Urban Areas

3 Motivation: Natural Disaster Simulation

4 Motivation: Crash Test for Automobiles

5 Motivation: Research Related to Defense

6 Driving Forces behind these Applications? In depth understanding of the application problem, Advanced numerical techniques, and Tremendous growth in high performance computing (HPC) Computational Sciences and Engineering ( CSE )

7 Overview of the Presentation Evolution of Computational Sciences Definition & Scope of CSE Concrete Example of CSE Applications CSE Education Example CSE Education TU Braunschweig, Germany Research Challenges for Computational Scientists

8 First Part of the Presentation Scope & Application of CSE

9 Scientific Research: Methods of Investigation Many researchers claim that computing has become a third main method for doing investigations, besides theory and experiments

10 Computation: The Third Research Method Computer simulation makes it possible to investigate regimes: Beyond current experimental capabilities Phenomena that cannot be replicated in laboratories

11 What is Computational Science and Engineering? CSE is the practice of computer-based modeling and simulation for the study of scientific phenomena and engineering designs. CSE research is multidisciplinary, as it requires: knowledge and methodologies from the application fields, from computer science, and from mathematics; expertise in the use of the technologies that comprise today s computing environments; the integration of the information technologies into new combinations that provide new capabilities.

12 CSE in a Nutshell! Application science C S E CS hard/software Math techniques

13 CSE Application Fields Various engineering disciplines : Computational Mechanics, Large-scale civil engg. problems Computational fluid dynamics, Industrial optimization problems Computational electromagnetics, Hydro-informatics, Reservoir simulation Basic Sciences : Computational chemistry, Computational physics, Astronomy Computational biology, Economics, Finance...

14 Super Computers and CSE Knowledge of high-performance computer hardware and software is important for programming of computationally intensive applications

15 A Multi-disciplinary CSE Application Problem : Reservoir History Matching

16 Statement of History Matching Problem History matching is a difficult inverse problem arising in the petroleum industry. History matching involves adjusting model parameters, with the aim of obtaining a model output, which is as close to the history (dynamic) data as possible. Mathemtically, the aim is to determine a spatial distribution r(x) and a set of model parameters P, given the history data Θ h, such that Θ s Θ h 0, where Θ s = f(r, P) is the simulated/model output. If one assumes that the geological descriptions is true, then the history-matching problem reduces to determining the set of model parameters P.

17 Mathematical Model of History Matching The flow of oil in a fractured reservoir is described, in the simplest case, by a linear parabolic partial differential equation of the form: u t = ( κ u ) + f (1) where κ is the conductivity. The dependent variable u represents pressure and f accounts for the withdrawal or injection of fluid in the reservoir. Hence the histoty matching problem can be stated as: Knowing the source term f and the initial condition and given a measurement of u(x, t) at a discrete set of points x 1,, x µ, determine the spatial varying parameter κ (including the locations of discontinuities).

18 Multi-disciplinary Nature of History Mathching Required knowledge from the fields of: Geo-physics, seismic analysis Reservoir and chemical engineering Applied mathematics & numerical methods Large-scale software development, i.e. communication and programming Production, marketing, management & sales Researchers involved from : Geophysists & exploration experts Mathematicians & Statisticians Reservoir, civil, mechanical & chemical engineers Computer scientists & so on

19 Second Part of the Presentation CSE Education Contents

20 Nature of CSE Education The traditional teaching of science tends to focus on theory. In contrast, CSE offers an understanding of science through the computer applications of mathematical models. It also makes many science and mathematics concepts more easily accessible to students who may otherwise not be reached; for example, those students not interested in computer hardware, software, and algorithms for their own sake. In addition, by extending the examples used in education to include problems which may not have analytic solutions, CSE enriches the science curriculum by extending the range of problems open to study.

21 Nature of CSE Education... Computer Science Engineering & Science Computational Science Solved Problem Mathematics Numerics

22 Intellectual Content of CSE Education Learning high-level computer languages and high-performance computing; Obtaining knowledge of applied mathematics and computational methods; Learning the basics of simulation and modeling; Learning how to interpret and analyze data visually; Applying acquired computing skills to at least one application area; Learning to communicate solution methods and results

23 Required Knowledge for CSE Education Computational Tools High Performance Computing Applied Mathematics CSE Education Simulation & Modeling Scientific Application Visualization Tools

24 Third Part of the Presentation CSE Masters Germany

25 CSE Program at TU Braunschweig, Germany An international, interdisciplinary and bilingual Master s program in Computational Sciences in Engineering (CSE). CSE program is directed at students of engineering sciences, mathematics and computer science. The program combines 3 semesters of instructions and one semester for the Master s Thesis. Each student has the greatest possible f reedom in choosing individually from the wide range of courses and lectures.

26 CSE Course Structure at TU Braunschweig Basic Core Course Mathematics & computer science e.g. scientific computing, PDEs, algorithm & programming Basic engineering sciences e.g. solid mechanics, thermodynamics, electro-magnetic fields, systemics, fluid mechanics Elective Core Course Mathematics & computer sciences e.g. parallel computing, sparse linear systems, discrete mathematics, computer graphics, large nonlinear sysems etc. Basic engineering sciences e.g. finite element methods, CFD, continuum mechanics etc. In-depth Course Various fields Software management, computational bionics, model reduction etc.

27 Research TU Braunschweig As a Master s student of CSE: Possibilities to work as student-research worker Doing practical training in German industrial companies, e.g. VW, Siemens After finishing CSE studies: Working in the university as scientific research staff (Full time job) Start working as a PhD student in conjunction with industries Possibilities to begin career in various industries

28 An Example CSE Research Project: PLATON Software Framework GUI remote control steering Domain of the Thesis Manager Optimizer NM Simplex Surrogate Kriging Communication Middleware CTL Communication Template Library (MPI, PVM, TCP IP, etc...) Optimization Process ObjectiveFunction manages SimulationTasks Simulation user integrated simulation Simulation Surrogate obj func evaluation

29 Research Challenges for Computational Scientists Efficient portable parallel programming. Programming languages and libraries. Familiarity with partial differential equations describing physical phenomena Ability to carry out statistically meaningful analysis of data from both experiments and simulations. Reduction of problem size Patience & passion for solving inter-disciplinary research problems

30 The End More information on CSE Education worldwide : Further Questions or Inquiries? to: tu-bs.de