Course details

# High Performance Computations

VNV Acad. year 2019/2020 Summer semester 5 credits

The course is aimed at practical methods of solving sophisticated problems encountered in science and engineering. Serial and parallel computations are compared with respect to a stability of a numerical computation. A special methodology of parallel computations based on differential equations is presented. A new original method based on direct use of Taylor series is used for numerical solution of differential equations. There is the TKSL simulation language with an equation input of the analysed problem at disposal. A close relationship between equation and block representation is presented. The course also includes design of special architectures for the numerical solution of differential equations.

Guarantor

Deputy Guarantor

Language of instruction

Czech

Completion

Examination (written)

Time span

26 hrs lectures, 26 hrs pc labs

Assessment points

60 exam, 20 half-term test, 20 labs

Department

Lecturer

Instructor

Šátek Václav, Ing., Ph.D. (DITS FIT BUT)
Veigend Petr, Ing. (DITS FIT BUT)

Course Web Pages

Subject specific learning outcomes and competences

Ability to transform a sophisticated technical problem to a system of differential equations. Ability to solve sophisticated systems of differential equations using simulation language TKSL.

Generic learning outcomes and competences

Ability to create parallel and quasiparallel computations of large tasks.

Learning objectives

To provide overview and basics of practical use of parallel and quasiparallel methods for numerical solutions of sophisticated problems encountered in science and engineering.

Why is the course taught

Supercomputers are often used to solve large technical and scientific problems. Before writing the first line of code, the user should perfectly understand the problem, that is being solved.

The goal of this course is to familiarize the students with the physics behind the problems, that are often solved in practice. To be able to see connection between the equations that govern the problem (and then solve it using differential calculus) and the real system. The students should also understand the numerical methods that are being used in the often used software packages as "black boxes". To be able to choose a proper numerical method for a specific problem and not just pick one at random.

Study literature

• Hairer, E., Norsett, S. P., Wanner, G.: Solving Ordinary Differential Equations I, vol. Nonstiff Problems. Springer-Verlag Berlin Heidelberg, 1987.
• Hairer, E., Wanner, G.: Solving Ordinary Differential Equations II, vol. Stiff And Differential-Algebraic Problems. Springer-Verlag Berlin Heidelberg, 1996.
• Butcher, J. C.: Numerical Methods for Ordinary Differential
• Lecture notes written in PDF format,
• Source codes (TKSL) of all computer laboratories

Fundamental literature

• Kunovský, J.: Modern Taylor Series Method, habilitation thesis, VUT Brno, 1995
• Hairer, E., Norsett, S. P., Wanner, G.: Solving Ordinary Differential Equations I, vol. Nonstiff Problems. Springer-Verlag Berlin Heidelberg, 1987.
• Hairer, E., Wanner, G.: Solving Ordinary Differential Equations II, vol. Stiff And Differential-Algebraic Problems. Springer-Verlag Berlin Heidelberg, 1996.
• Shampine, L. F.: Numerical Solution of ordinary differential equations, Chapman and Hall/CRC, 1994
• Strang, G.: Introduction to applied mathematics, Wellesley-Cambridge Press, 1986
• Meurant, G.: Computer Solution of Large Linear System, North Holland, 1999
• Saad, Y.: Iterative methods for sparse linear systems, Society for Industrial and Applied Mathematics, 2003
• Burden, R. L.: Numerical analysis,  Cengage Learning, 2015
• LeVeque, R. J.: Finite Difference Methods for Ordinary and Partial Differential Equations: Steady-State and Time-dependent Problems (Classics in Applied Mathematics), 2007
• Strikwerda, J. C.: Finite Difference Schemes and Partial Differential Equations,  Society for Industrial and Applied Mathematics, 2004
• Golub, G. H.: Matrix computations, Hopkins Uni. Press, 2013
• Duff, I. S.: Direct Methods for Sparse Matrices (Numerical Mathematics and Scientific Computation), Oxford University Press, 2017
• Corliss, G. F.: Automatic differentiation of algorithms, Springer-Verlag New York Inc., 2002
• Griewank, A.: Evaluating Derivatives: Principles and Techniques of Algorithmic Differentiation, Society for Industrial and Applied Mathematics, 2008
• Press, W. H.: Numerical recipes : the art of scientific computing, Cambridge University Press, 2007

Syllabus of lectures

1. Methodology of sequential and parallel computation (feedback stability of parallel computations)
2. Extremely precise solutions of differential equations by the Taylor series method
3. Parallel properties of the Taylor series method
4. Basic programming of specialised parallel problems by methods using the calculus (close relationship of equation and block description)
5. Parallel solutions of ordinary differential equations with constant coefficients, library subroutines for precise computations
6. Adjunct differential operators and parallel solutions of differential equations with variable coefficients
7. Methods of solution of large systems of algebraic equations by transforming them into ordinary differential equations
8. The Bairstow method for finding the roots of high-order algebraic equations
9. Fourier series and finite integrals
10. Simulation of electric circuits
11. Solution of practical problems described by partial differential equations
12. Control circuits
13. Conception of the elementary processor of a specialised parallel computation system.

Syllabus - others, projects and individual work of students

Elaborating of all computer laboratories results.

Progress assessment

Half Term Exam and Term Exam. The minimal number of points which can be obtained from the final exam is 29. Otherwise, no points will be assigned to a student.

Controlled instruction

During the semester, there will be evaluated computer laboratories. Any laboratory should be replaced in the final weeks of the semester.

Schedule

DayTypeWeeksRoomStartEndLect.grpGroupsInfo
Monlecturelectures A113 15:0016:50 1MIT 2MIT MMM
Monlecturelectures A113 17:0017:50 1MIT 2MIT MMM
Moncomp.lablectures N105 17:0018:50 1MIT 2MIT xx
Moncomp.lablectures N105 19:0019:50 1MIT 2MIT xx
Tuecomp.lablectures N203 12:0013:50 1MIT 2MIT xx

Course inclusion in study plans