Math 311: Numerical Solutions to CE Problems

Numerical Analysis (NA) REPEAT

 A way to do higher mathematics problems on a computer xL -xR

SET xn+1 =
 Numerical answers can be obtained even when a problem 2
has no analytical solution
 Say: IF f(xn+1)* f(xL)<0
π SET xR= xn+1
∫ √1+ cos2 x dx
ENDIF
0

xL -xn+1
o Solution is always numerical UNTIL (|
𝑥𝑛+1
|) <tolerance value or f(xn+1) = e
o An approximation, but results can be made as
accurate as desired  The final value of x3 approximates the root, it is in error by
1
◥ Operations that NA can do: not more than |x -x |.
2 1 2
 Solve for the roots of a non-linear equation
NOTE: If f(x)=0⇨ discontinuous on [x1, x2] the method may give
 Solve large systems of linear equation
a false root.
 Get the solutions of a set of non-linear equation
 Interpolations (b-a)
 Find derivatives of any order Error after nth iterations < | |
2n
 Integrate any functions ☑ Sample Problem
 Solve ODE
 Solve boundary-value problems Bisection
 Obtain numerical solutions of partial Differential Equations
 Fit curves f(x)=x3+x2-3x-3=0, with of tolerance of 1E-2
 Approximations of functions

◥ 4 General steps in solving engineering or scientific problem:

II. Bracketing Method: Linear Interpolation method: The Secant
 State problem clearly Method

 Develop a mathematical statement of the problem Secant Line: a line that passes through 2 pts of a curve.

 Solve the equation f(xR)

 Interpret the numerical result to arrive at a decision

I. Bracketing Method: Interval Halving (Bisection)

 Its strategy is to begin with 2 values of x: a & b, that bracket

a root of f(x)=0. f(xR)
 Successively divides the interval in half and replaces one-
endpoint with the midpoint.

f(xL)
f(XR)
X2 X1 XR

f(X4)
◥ Similar Triangles
f(X5)
xR -xn+1 xR -xL
XL X3
=
X5
X4 f(𝐗 𝟎 ) f(xR )-f(xL )
XR

𝑿𝑹 − 𝑿𝑳 Solve for x2:

f(X3) 𝑿𝟑 =
𝟐
(xR -xL )
Xn+1= xR − f(xR ) ∗ f(x
f(XL) 𝑿𝑹 − 𝑿𝟑 R )-f(xL1 )
𝑿𝟒 =
𝟐
◥ Algorithm
◥ Algorithm
To determine a root f(x)=0, given two values, x R & xL, that are near
 To determine a root of f(x)=0 that is accurate within a the root.
specified tolerance value. Given value x L & xR such that
f(xL)*f(xR)<0 IF |f(xR )| < |f(xL )|

FRANCH MAVERICK LORILLA 1

Math 311: Numerical Solutions to CE Problems

Swap xR with xL

REPEAT
f(x0)
(xR -xL )
SET xn+1= xR − f(xR ) ∗ f(xR )-f(xL )

SET x0= x1 f(x0)

SET x1= x0 x1
x0
UNTIL |f(x1 )|<tolerance value
Note: If f(x) is not continuous the method may fail.
x1 x0- x1

☑ Sample Problem f(x1)

x0

Secant Method
f(x0 )
f(x)=x3+x2-3x-3=0, use same initial estimate of previous tan θ = f’(x0) =x
example and tolerance of 1E-3 0 -x1

f(x )
x1= x0-f'(x0 )
III. Bracketing Method: Linear Interpolation method: False 0
Position
f(xn )
 Regula falsi (Latin)
𝑥𝑛+1 = 𝑥𝑛 − ,
f'(xn)
 Similar to bisection except the next iterate is taken at the
intersection of a line between the pair of x-values and the x- n=0,1,2…
axis rather than the midpoint.
◥ Algorithm
◥ Algorithm
To determine a root f(x)=0, given a value x0 reasonably close to the
To determine a root of f(x)=0, given values of x 0 and x1 that bracket a root.
root, that is, f(x0) & f(x1) are opposite sign.
COMPUTE f(x0), f’(x0)
REPEAT
SET x1= x0
(x0 -x1 )
SET x2= x0 − f(x0 ) ∗
f(x0 )-f(x1 ) IF [f(x0) ≠0] and [f’(x0) ≠0]
IF f(x2)is opposite sign to f(x0)
SET x0= x1
SET x1= x2 f(x )
SET x1= x0-f'(x0 )
ELSE 0

SET x0= x2 UNTIL

ENDIF |x0= x1| <tolerance value or f(x1 )<tolerance value

UNTIL |f(x2)|< tolerance vale
☑ Sample Problem
Note: If f(x) is not continuous on the interval, the method may fail
Newton’s Method
☑ Sample Problem
f(x)=x3+x2-3x-3=0, start with a=1, b=2 and tolerance of 1E-2
Regula Falsi

f(x) = x3+x2-3x-3=0, start with a=1, b=2 and tolerance of 1E-2

Example: Locate 1st positive root of 𝑓(𝑥) = 𝑠𝑖𝑛𝑥 + 𝑐𝑜𝑠(1 + 𝑥 2 ) − 1
where x is in radian and tolerance of 1E-3

IV. Open Method: Newton’s Method

 Based on a linear approximation of the function, but does so

using a tangent to the curve.

FRANCH MAVERICK LORILLA 2

Math 311: Numerical Solutions to CE Problems

V. Open Method: Fixed Point Iteration

Y1=x

y0=g(x) System of Linear Equations

I. Determinants and Cramer’s Rule

◥ Determinants of order 2 & 3

(a) Order of 2

a b1
D= | 1 | = a1 b2 − a2 b1
X2 X1 X0 a2 b2
(b) Order of 3

a1 b1 c1 a1 b1
D= |a2 b2 c2| a2 b2
a3 b3 c3 a3 b3

D= +(a1b2c3+b1c2c3+c1a2b3) - (a3b2c1+b3c2a1+c3a2b1)
 X=g(x)
 Rearrange f(x) into an equivalent form x=g(x) ◥ Cramer’s Rule

◥ Iterative  Cramer’s rule is another solution technique that is best

suited to small numbers of equations.
Xn+1=g(xn), n=0,1,2,3…
Cramers rule states that if

Algorithm
a11x+a12y+a13z=b1

 To determine a root of f(x)=0 given a value x1 reasonably a21x+a22y+a23z=b2

close to the root
 Rearrange the equation to an equivalent form x=g(x) a31x+a32y+a33z=b3

SET x2=x1 then

Dx Dy Dz
x= D
, y= D
and z= D
REPEAT

SET x1= x2 where

SET x2= g(x1) a11 a12 a13

D= |a21 a22 a23|
UNTIL | x1- x2 |<tolerance value a31 a32 a33
Note: Converges if the magnitude of slope of g(x) is less than the b1 a12 a13
slope of the line x = f(x)
Dx = |b2 a22 a23 |
b3 a32 a33
☑ Sample Problem
i.e. the column has been replaced by R.H.S. b column,
Fixed-Point Iteration
a11 b1 a13
f(x)=x3+x2-3x-3=0, start with a=1, b=2 and tolerance of 1E-2 Dy = |a21 b2 a23 |
a31 b3 a33
i.e. the y-column has been replaced by the R.H.S. b column,
☑ Assignment
Given: a11 a13 b1
Dz = |a21 a23 b2 |
f(x)=3x+sin(x)-ex=0, for x0=1 & x1=0
a31 a33 b3
Requirement:
i.e. the z-column has been replaced by the R.H.S. b column.
☑ Compare the results using BISECTION, FALSE POSITION,
☑ Sample Problem
NEWTON’S & FIXED POINT ITERATION.
Solve the following simultaneous equations using Cramers rule

x+y+z=4
2x−3y+4z=33FRANCH MAVERICK LORILLA 3
3x−2y−2z=2
Math 311: Numerical Solutions to CE Problems

I. Gauss-Jordan Elimination Method

☑ Supplementary Problem
◥ Elementary row operations
Solve the following system by using the Gauss-Jordan
elimination method.
1. interchanging two rows 7x + 2y - 3z = -12
2. add one row to another row, or multiply one row first, then 2x + 5y - 3z = -20
adding it to another x – y - 6z = -26
3. Multiplying a row by any constant greater than zero

Form should be in row enchelon form: I. Iterative Method: Jacobi Method

1 0 0 x x  The Jacobi method is a method of solving a matrix equation
[0 1 0] [y] = [y] on a matrix that has no zeros along its main diagonal
0 0 1 z z (Bronshtein and Semendyayev 1997, p. 892). Each diagonal
element is solved for, and an approximate value plugged in.
The process is then iterated until it converges. This algorithm
☑ Sample Problem
is a stripped-down version of the Jacobi transformation
Solve the following system by using the Gauss-Jordan method of matrix diagonalization.
elimination method.
x+y+z=5 ☑ Sample
2x+3y+5z=8 Problem
Solve the following system by Jacobi Method
4x+5z=2 6x1-2x2+x3=11
-2x1+7x2+2x3=5
x1+2x2-5x3=-1
Solution: The augmented matrix of the system is the following.

1 1 1 5
[2 3 5 |8] 1st 2nd 3rd 4th 5th 6th .. 9th
4 0 5 2 x1 0 1.833 2.038 2.085 2.004 1.944 .. 2.000
We will now perform row operations until we obtain a matrix in x2 0 0.7143 1.181 1.053 1.001 0.990 .. 1.000
reduced row echelon form. x3 0 0.200 0.852 1.080 1.038 1.0001 .. 1.000

For x1:
11+2X2 -X3
X1=
6

For x2:
5+2X1 -2X3
X2=
7

For x3:
-1-X1 -2X2
X3=
-5

1st Iteration:
11+2(0)-(0)
X1= = 1.8333
6

5+2(0)-2(0)
X2= = 0.7143
7

-1-(0)-2(0)
X3= = 0.2
-5

2nd Iteration:
11+2(0.7143)-(0.2)
X1= = 2.0381
6

5+2(1.8333)-2(0.2)
X2= = 1.181
7

-1-1.8333 -2(0.7143 )
X3= = 0.852
-5
From this final matrix, we can read the solution of the system. It is

FRANCH MAVERICK LORILLA 4

Math 311: Numerical Solutions to CE Problems

3RD Iteration [𝐿][𝐿]−1 [𝐶 ] = [𝐿][𝑌]

11+2(1.181)-(0.852)
= 2.085
X1=
6 [𝐼 ][𝐶 ] = [𝐿][𝑌] — 
5+2(2.0381)-2(0.852)
X2= = 1.053 By Forward Substitution:
7

X3=
-1-2.0381 -2(0.852 )
-5
= 1.080 𝑥 0 0 𝑌1 𝐶1 ↔ 𝑌1
4th Iteration
[𝑥 0 0] [𝑌2 ]=[𝐶2 ] ↔ 𝑌2
𝑥 𝑥 𝑥 𝑌3 𝐶3 ↔ 𝑌3
11+2(1.053)-(1.080)
X1= = 2.085
6 By Back Substitution:
5+2(2.085)-2(1.080)
X2= = 1.053 Unknowns
7

X3=
-1-2.085-2(1.080 )
= 1.080 𝑥 𝑥 𝑥 𝑋1 𝑌1 ↔ 𝑋1
-5
[0 𝑥 𝑥 ] [𝑋2 ]=[𝑌2 ] ↔ 𝑋2
.
. 0 0 𝑥 𝑋3 𝑌3 ↔ 𝑋3
.
.
9th Iteration ☑ Sample Problem
X1=2.000 Solve the following simultaneous equations using LU
Decomposition.
X2=1.000
𝑥1 + 𝑥2 − 𝑥3 = 4
X3=1.000
𝑥1 − 2𝑥2 + 3𝑥3 = −6
III. Direct Methods: LU Decomposition
2𝑥1 + 3𝑥2 + 𝑥3 = 7
 In many applications where linear systems appear, one needs
to solve Ax = b for many different vectors b. Gaussian
elimination with pivoting is the most efficient and accurate way
to solve a linear system. Most of the work in this method is
spent on the matrix A itself. If we need to solve several
different systems with the same A, and A is big, then we would
like to avoid repeating the steps of Gaussian elimination on A
for every different b. This can be accomplished by the LU
decomposition, which in effect records the steps of Gaussian
elimination.

[𝐴]𝑛𝑥𝑚 [𝑋 ]𝑛𝑥1 = [𝐶 ]𝑛𝑥1

Since,

[𝐴] = [𝐿][𝑈]

Lower Triangular Upper Triangular

Matrix Matrix

Substitute:

[𝐿][𝑈][𝑋 ] = [𝐶 ]
[𝐿]−1 [𝐿][𝑈][𝑋 ] = [𝐿]−1 [𝐶 ]
[𝐼][𝑈][𝑋 ] = [𝐿]−1 [𝐶 ]
[𝑈][𝑋 ] = [𝑌]— 

[𝐿]−1 [𝐶 ] = [𝑌]
FRANCH MAVERICK LORILLA 5