1

Scientific Computing

SoICT 2023

2
 Lecturer

DR. VŨ VĂN THIỆU

DEPARTMENT OF COMPUTER SCIENCE
SCHOOL OF ICT, HUST
Office: P.702, B1
Tel: 0982928307
Email: thieuvv@soict.hust.edu.vn

3
 Content

1. What is Scientific Computing

2. Error
3. Stability of numerical method
4. List of topics
5. References

4
 What is Scientific Computing?

• Scientific computing involves the efficient use of

computers and numerical methods, taking advantage of
advanced computers to solve large-scale problems in
science and engineering.

• Computer science is the collection of all the tools,

techniques, and theories needed to develop and solve
mathematical models in science and engineering on
computers.

5
 What is Scientific Computing?

Scientific Real Problem

Computing
Applied
Science
Model

Math Algorithm

ICT Result

Deployment
6
 Role of Scientific Computing

• Together with theory and experimentation,

computational science now constitutes the third pillar
of scientific inquiry, enabling researchers to build
and test models of complex phenomena
(Report to the President, Computational Science: Ensuring
America's Competitiveness, USA, June 2005).

7
 Errors in Scientific Computing

• Error of Model
• Error in Calculation
• Truncation error
• Rounding error
• Error of Data
• Example:
• Formula to find area of the earth = 4πR2
• What are the errors?

8
 Two ways to calculate error

• Assuming x* approximates the real number x:

• Absolute error: | x* - x|

• Relative error: ,x≠0

|x*− x|
|x|

9
 Cumulative rounding error

• When an algorithm has to perform many operations

(loops), the error of each operation is small, but the error
of the algorithm can be large. We must consider how the
error of each operation accumulates in the algorithm.

10
 Overflow

• Large number overflow occurs when the obtained

number is larger than the largest number that can be
represented by the floating-point number: e > emax
• Small number overflow occurs when the resulting
number is less than the smallest number that can be
represented by the floating-point number: e < emin

11
 Stability of the numerical method

• The concept of numerical stability is related to the

accuracy of the algorithm because of rounding errors.
• An algorithm is said to be unstable if rounding errors
can lead to large errors in the results.

12
 List of topics

1) Matlab (8T)
2) System of Linear Equations (4T)
3) Fitting Curve (4T)
4) Solving Nonlinear Equation (4T)
5) Approximation of Derivative and Integral (4T)
6) Ordinary Differential Equations (4)
7) Unconstraint Minimization Methods (4)
8) Linear Programming (8T)

13
 References

1. Tạ Văn Đĩnh. Phương pháp tính. NXB Giáo dục,

1995.
2. Phạm Kỳ Anh. Giải tích số. NXB Đại học Quốc gia
Hà nội, 1996.
3. Lê Trọng Vinh. Giải tích số. NXB Khoa học Kỹ
thuật, Hà nội, 2000.
4. Nguyễn Đức Nghĩa. Tối ưu hoá (Quy hoạch tuyến
tính và rời rạc). NXB Giáo dục, 1996.

14
 References

5. Michael Heath. Scientific Computing: An

introductory survey. McGraw-Hill Inc. 2001.
6. Charles F. Van Loan, Introduction to Scientific
Computing. A Matrix-Vector Approach Using
Matlab. 2nd Edition. Prentice Hall, 2000.

15
 References

7. Duane Hanselman, Bruce Littlefield.

Mastering MATLAB 7. Pearson/Prentice Hall,
2005, 852 Pages.
8. K. Sigmon. MATLAB Primer. Univ. Florida,
1996.
9. Gene Golub and Charles Van Loan. Matrix
computations. John Hopkins University Press
1996.
10. Dammed J.W. Applied Numerical Linear
Algebra, 1996.

16
 References

11. Iserles. A First Course in Numerical Analysis of

Differential Equations, Cambridge University
Press, 1996.
12. C. Evans, Partial Differential Equations, AMS,
1998.
13. G. W. Stewart, Introduction to Matrix
Computations, Academic Press, 1998
14. Gill, P.E., Murray, W. and Wright, M. H.,
Numerical Linear Algebra and Optimization,
Volume 1, Addison-Wesley, Redwood City,
California, 1991.
15. Stephen G. Nash, Ariela Sofer, Linear and
Nonlinear Programming. McGraw-Hill, 1998.

17