CUFE, M. Sc.

, 2015-2016

Computers & Numerical

Analysis (STR 681)

Introduction
Dr. Maha Moddather

Structural Engineering Department

Faculty of Engineering – Cairo University

mahamoddather@eng.cu.edu.eg

Spring 2016
Why Do We Need Numerical
Analysis Methods?
Introduction

Numerical methods are techniques by which

mathematical problems are formulated so that
they can be solved with arithmetic operations.

All numerical methods involves large numbers

of arithmetic calculations.
Introduction

There are three approaches for problem solving

using non-computer methods.

 Analytical (Exact Approach).

 Graphical Solution.

 Calculators and Slide Rules.

Introduction
1. Analytical (exact approach):

Excellent insight into behavior.

Derived for only a limited class of problems.

Approximated for linear models, simple

geometry, and low dimensionality.

Limited practical value

Real problems are nonlinear, complex, in shape

and processes.
Introduction
2. Graphical solutions:

Characterize the behavior of systems.

Used to solve complex problems.

Not very precise.

Extremely tedious and awkward to

implement.
Introduction

3. Calculators and Slide Rules:

Implement numerical methods manually.

Adequate for solving complex solutions.

But slow and tedious.

Consistency results are elusive: blunders.

Introduction

Computers and numerical methods provide an

alternative method for such calculations.

Using computer power, problems can be

approached without large simplifications or time-
intense techniques.
 Acknowledgement

THIS IS TO THANK DR. HESHAM SOBHY, DR. ASMAA

HASSAN AND DR. AHMED AMIR BAYOUMY FOR

GIVING THEIR VOLUNTARILY HELP IN PREPARING THIS

PRESENTATION
Course Outline
Regulations
Outline
 Systems of Linear Algebraic Equations
 Nonlinear Equations
 Polynomial Approximation & Interpolation
 Numerical Differentiation & Difference
Formulas
 Numerical Integration
 Discretization & Finite Difference Methods
Outline
 Weighted Residual Approach
 Piecewise Functions
 Finite Element Methods
 Optimization
 Curve Fitting
 Ordinary Differential Equations
 Partial Differential Equations
Outline
 Perturbation methods
 Fourier analysis
 Approximations & Round-off Errors
Outline

Programming
Language

MATLAB Program
Grading System
Grading System

Grading
100%

Term Work Final Exam

30% 70%

Assignments Project Midterm Exam

5% 10% 15%
Systems of Linear Algebraic
Equations
Systems of Linear Algebraic Equations

 Systems of linear algebraic equations:

These problems are concerned with the value of a

set of variables that satisfies a set of linear
equations.
Systems of Linear Algebraic Equations
Matrix Notations

Column 2 Column m

Row 1

Row n
Matrix Notations

Row vector, n = 1

Column vector, m = 1

Square matrix, n = m
Matrix Notations

Symmetric matrix

Diagonal matrix

Identity matrix
Matrix Notations

Upper triangular
matrix

Lower triangular
matrix

Banded matrix
Ex., Band width = 3
(tridiagonal matrix)
Matrix Operating Rules
Matrix Operating Rules
Matrix Operating Rules
Matrix Operating Rules
Matrix Operating Rules

column row
Matrix Operating Rules

3 by 2
2 by 2
3 by 2
Matrix Operating Rules
Matrix Operating Rules
Matrix Operating Rules

If A is a square matrix:

A-1 is the inverse of A

Matrix Operating Rules

If A is a square matrix:

AT is the transpose of A
Matrix Operating Rules

tr [A] is the trace of matrix [A]

Matrix Operating Rules

For a 2 x 2 determinant

Ex.,
Matrix Operating Rules

For a 3 x 3 determinant

minor

Ex.,

D = 0.3 x - 0.52 x +1x

D = 0.3 x (1x0.5 – 0.3 x 1.9) – 0.52 x 0.06 + 1 x 0.05 = -0.0022

Matrix Operating Rules

“Augmentation”

Such an expression has utility when we must perform a set of

identical operations on two matrices. Thus, we can perform the
operations on the single augmented matrix rather than on the
two individual matrices.
Matrix Operating Rules

No division in Matrices
Matrix Operating Rules
Solving Linear Algebraic Equations

Solving small number of equations:

 The Graphical Method.

 Cramer’s Rule.

 The Elimination of Unknowns.

Gauss Elimination

Gauss-Jordan

LU Decomposition

Gauss Seidel
Solving Linear Algebraic Equations

Direct Methods:

Cramer’s Rule.

Gauss Elimination.

Gauss Jordan.

Banded Matrix.

Skyline

Iterative Methods

 Jacobi Iteration.

 Gauss-Seidel
The Graphical Method

Suitable for small number of equations (≤ 3)

slope intercept
The Graphical Method

Ex., Solve

intercept

slope
The Graphical Method

Ex., Solve

intercept
slope
The Graphical Method

Ex., Solve
The Graphical Method

Is there cases where there will be no solution?

Parallel Lines : No solution

(Singular system)
The Graphical Method

Is there cases where there will be no solution?

Coincident Lines : Infinite solutions

(Singular system)
The Graphical Method

Is there cases where there will be no solution?

Ill conditioned system

Cramer’s Rule
This rule states that each unknown in a system of linear
algebraic equations may be expressed as a fraction of
two determinants with denominator D and with the
numerator obtained from D by replacing the column of
coefficients of the unknown in question by the constants
b1, b2, . . . , bn. For example, x1 would be computed as:
Cramer’s Rule

Ex.,
Cramer’s Rule

Ex.,
Cramer’s Rule

Ex.,
Cramer’s Rule

Ex.,
The Elimination of Unknowns

The basic strategy is to multiply the equations by

constants so that one of the unknowns will be eliminated
when the two equations are combined.

The result is a single equation that can be solved for the

remaining unknown.

This value can then be substituted into either of the

original equations to compute the other variable.
The Elimination of Unknowns

multiply by a21

multiply by a11

Cramer’s rule
The Elimination of Unknowns

Ex.,
Gauss Elimination
The procedure consisted of two steps:

1. The equations were manipulated to eliminate one

of the unknowns from the equations. The result of
this elimination step was that we had one equation
with one unknown.

2. Consequently, this equation could be solved

directly and the result back-substituted into one of
the original equations to solve for the remaining
unknown.
Gauss Elimination
Gauss Elimination

Ex.,
 Gauss Elimination
Pivot
x 0.1/3

x 0.3/3

Pivot x -0.19/7.003333
Gauss Elimination
Gauss Elimination
Gauss Elimination
Drawbacks of Gauss Elimination

Dividing by zero (or close to zero)

Solution: “Partial pivoting” switch rows and use

the largest value as a pivot
Pivot=0
Drawbacks of Gauss Elimination

Round-off errors

Important for more than 100 equations

You should always substitute your answers back

into the original equations to check whether a
substantial error has occurred.

Solution: scaling & use more significant fraction

figures
Drawbacks of Gauss Elimination

Correct answer
Ex.,
Drawbacks of Gauss Elimination

Use more significant fraction figures

Drawbacks of Gauss Elimination

Use scaling that maximum coefficient in each row is 1

Partial pivoting
Drawbacks of Gauss Elimination

3. Singular systems (D = 0)

4. Ill conditioned systems

Ill-conditioning is that a wide range of answers can

approximately satisfy the equations.

Because round-off errors can induce small changes in the

coefficients, these artificial changes can lead to large
solution errors for ill-conditioned systems.

D ≈ 0 (scaling numbers in the matrix to be less than one)

Solution: use more significant figures

Drawbacks of Gauss Elimination

Ex.,
Drawbacks of Gauss Elimination

Replace 1.1 by 1.05

Drawbacks of Gauss Elimination

Replace 1.1 by 1.05

6
x1+2x2=10
5 1.1x1+2x2=10.4
4
x2

3 Ill conditioned
2 condition
1
0
0 1 2 3 4 5 6 7 8 9 10
x1
Gauss Elimination