7/28/2019 Lectures 3 4 5
1/98
users.isc.tuc.gr/~adelis/ENG407.html
Lectures 3-4
Dr A.I. Delis TUC
2012Part 2
1
7/28/2019 Lectures 3 4 5
2/98
Roots of Equations
Motivationart
02
=++= cbxaxxf )( a
acbb
x 2
42 =
3sin log 4 0 ?xBut how about e x x x + + =
pprox ma on so u on pre-compu er :
f(x)
Plot method
Trial and error
x
7/28/2019 Lectures 3 4 5
3/98
An Example of Roots of Equation in Engineering
Parachutists velocity: )()/( tmce
c
gmv = 1
Problem: determine the drag coefficient cfor a parachutist of a given
mass m to attain a prescribed velocity in a set time period, that is,
Given v, m,g, and t, find c.
Approach 1: try to represent c as an explicit function c =f(v, m, g, t)(fails most of time)
Approach 2: express the formula in an explicit form and solve for the
zero root
01 == vec
cf )()(
7/28/2019 Lectures 3 4 5
4/98
Nonlinear Equation
Solvers
Bracketing Graphical Open Methods
Bisection Newton Raphson
a se os on
(Regula-Falsi) Secant
All Iterative
Dr A.I. Delis TUC
2012Part 2
4
7/28/2019 Lectures 3 4 5
5/98
Bracketing Methods
, Chapter 3
wo n a guesses or e rooare required. These guesses
must bracket or be on eithers e o t e root.
==
If one root of a real andcontinuous function, f(x)=0, is
bounded by values
x=x , x =x thenf(xl).f(xu)
7/28/2019 Lectures 3 4 5
6/98
No answer No root
Nice case (one root)
ops wo roo s
work for a while!!)
Dr A.I. Delis TUC
2012Part 2
6
7/28/2019 Lectures 3 4 5
7/98
Two roots( Might
work for a while!!)
function. Need
special method
Dr A.I. Delis TUC
2012Part 2
7
7/28/2019 Lectures 3 4 5
8/98
MANY-MANY roots. What do we do?
f(x)=sin 10x+cos 3x
Dr A.I. Delis TUC
2012Part 2
8
7/28/2019 Lectures 3 4 5
9/98
= ,
1. Pick xl and xu such that they bound the root of
7/28/2019 Lectures 3 4 5
10/98
If f(xl). f[(xl+xu)/2]>0, root lies
in the upper interval, then xl=(xl+xu)/2, go to step 2.
If f(xl). f[(xl+xu)/2]=0, then + ul xxx l u .%100
2
+ ul xx
.s
a
(s tolerance)+
ul xx
or
. a s, s op. erw se repea
the process. %1002
+ ul
u
xx
Dr A.I. Delis TUC
2012Part 2
10
7/28/2019 Lectures 3 4 5
11/98
Dr A.I. Delis TUC
201Part 2
11
7/28/2019 Lectures 3 4 5
12/98
Step 1
Choose x and xu as two guesses for the root such that
7/28/2019 Lectures 3 4 5
13/98
Step 2
Estimate the root, xm of the equation f (x) = 0 as the mid
f(x)
u
x
xm =2
xux
m
13Figure Estimate of x
m
7/28/2019 Lectures 3 4 5
14/98
Step 3
Now check the following
a) If , then the root lies between x
and xm;
= =
( ) ( ) 0ml xfxfthen x
= xm; xu = xu.
0=xx m.algorithm if this is true.
14
7/28/2019 Lectures 3 4 5
15/98
Find the new estimate of the root
xx
m =xu +
2
oldnew xx
Find the absolute relative approximate error
=new
m
ax
m
where
new
rootofestimateprevious=oldmx
15
m
7/28/2019 Lectures 3 4 5
16/98
Compare the absolute relative approximate error with the pre-specified errora
Go to Step 2 using new
.s
Is ?
.
sa >
No Stop the algorithm
Note one should also check whether the number of iterations is more than themaximum number of iterations allowed. If so, one needs to terminate the algorithm
and notify the user about it.
16
7/28/2019 Lectures 3 4 5
17/98
Three ways in which the interval may bracket the root:
(a) True value lies at center
(b) and (c) True value lies near the extreme
Discrepancy between true value and solution never
Dr A.I. Delis TUC
2012Part 2
17
exceeds the interval length
7/28/2019 Lectures 3 4 5
18/98
Error estimate equivalence
Dr A.I. Delis TUC
2012Part 2
18
7/28/2019 Lectures 3 4 5
19/98
Pros
Easy
Cons
Slow Always find root Know a and b that
required to attain an Multiple roots
computed a priori. o accoun s a en o
f(xl
) and f(xu
), if f(xl
) is
c oser o zero, s e y
that root is closer to xl .
Dr A.I. Delis TUC
2012Part 2
19
7/28/2019 Lectures 3 4 5
20/98
Length of the first Interval Lo=b-a
= o
After 2 iterations L2=Lo/4
= k k o
saa
Dr A.I. Delis TUC
2012Part 2
20
o ve pro em . or a proo
7/28/2019 Lectures 3 4 5
21/98
EXAMPLE
If the absolute magnitude of the error is
4
10
=s
and Lo=2, how many iterations will you
have to do to get the required accuracyin the solution?
2 44 = k .2k
Dr A.I. Delis TUC
2012Part 2
21
7/28/2019 Lectures 3 4 5
22/98
Dr A.I. Delis TUC
2012Part 2
22
7/28/2019 Lectures 3 4 5
23/98
Algorithm
Dr A.I. Delis TUC
2012Part 2
23
7/28/2019 Lectures 3 4 5
24/98
Dr A.I. Delis TUC
2010Part 2 (Chapters 5-6-7)
24
7/28/2019 Lectures 3 4 5
25/98
Exam le 1You are working for DOWN THE TOILET COMPANY thatmakes floats for ABC commodes. The floating ball has aspec c grav y o . an as a ra us o . cm. ou areasked to find the depth to which the ball is submerged when
floatin in water.
Figure Diagram of the floating ball
25
7/28/2019 Lectures 3 4 5
26/98
Exam le 1 Cont.
The equation that gives the depth x to which the ball is
010993.3165.0423
=+
xxa) Use the bisection method of finding roots of equations
to find the depth x to which the ball is submerged under
water. Conduct three iterations to estimate the root ofthe above equation.
end of each iteration, and the number of significant
digits at least correct at the end of each iteration.
26
7/28/2019 Lectures 3 4 5
27/98
Example 1 Cont.From the physics of the problem, the ball would besubmerged between x= 0 and x= 2R,
where R= radius of the ball,
that is
20 x
( )055.020 x.
Diagram of the floating ball
27
E l 1 C t
7/28/2019 Lectures 3 4 5
28/98
Example 1 Cont.
Solution
To aid in the understanding of how this
method works to find the root of an
e uation the ra h of f x is shown to the
right,
where
( ) 423 1099331650 -.x.xxf +=
28Graph of the function f(x)
7/28/2019 Lectures 3 4 5
29/98
Example 1 Cont.
Let us assume 00.0
=
=x
.u
Check if the function changes sign between x
and xu .
( ) ( ) ( ) ( )4423
4423
10662.210993.311.0165.011.011.0
10993.310993.30165.000
=+==
=+==
x
fxf l
Hence
( ) ( ) ( ) ( ) ( ( 010662.210993.311.00 44
7/28/2019 Lectures 3 4 5
30/98
Example 1 Cont.
Graph demonstrating sign change between initial limits
30
C
7/28/2019 Lectures 3 4 5
31/98
Example 1 Cont.
055.011.00
=+
=+
=uxxx
Iteration 1
The estimate of the root is22
23
( ) ( ) ( ) ( ) ( )( ) 010655.610993.3055.00......
54>==
=+==
ffxfxf
x
ml
m
Hence the root is bracketed between xm and xu, that is, between 0.055 and 0.11.
,
11.0,055.0 == ul xx,
as we do not have a previous approximation.a
31
E l 1 C t
7/28/2019 Lectures 3 4 5
32/98
Example 1 Cont.
Estimate of the root for Iteration 1
32
E l 1 C t
7/28/2019 Lectures 3 4 5
33/98
Example 1 Cont.
0825.011.0055.0
=+
=+
=uxxx
Iteration 2
22
( ) ( ) ( ) ( )
010655.610622.10825.0055.0
10622.110993.30825.0165.00825.00825.0
54
4423
7/28/2019 Lectures 3 4 5
34/98
Example 1 Cont.
34
Estimate of the root for Iteration 2
E l 1 C t
7/28/2019 Lectures 3 4 5
35/98
Example 1 Cont.
The absolute relative approximate error at the end of Iteration 2a
is
oldnew
xx =new
m
ax
1000825.0
..
=
%333.33=
one o e s gn can g s are a eas correc n e es ma e roo
of xm = 0.0825 because the absolute relative approximate error is
35
.
E l 1 C
7/28/2019 Lectures 3 4 5
36/98
Example 1 Cont.
0825.0055.0=
+=
+=
uxxIteration 3
The estimate of the root is .22
m
5423
( ) ( ) ( ) ( ) ( )( ) 010563.510655.606875.0055.0......
55
7/28/2019 Lectures 3 4 5
37/98
Example 1 Cont.
37Estimate of the root for Iteration 3
Example 1 Cont
7/28/2019 Lectures 3 4 5
38/98
Example 1 Cont.
The absolute relative approximate error at the end of Iteration 3
isa
old
m
new
m xxnew
m
ax
10006875.0
..
=
%20=
Still none of the significant digits are at least correct in theestimated root of the equation as the absolute relative
approximate error is greater than 5%.
38
even more era ons were con uc e an ese era ons are
shown in Table 1.
Table 1 for Example 1
7/28/2019 Lectures 3 4 5
39/98
Table 1 for Example 1
Table 1 Root of f(x)=0 as function of number of iterations for bisection method.
Iteration x
xu xm a % f(xm)
1 0.00000 0.11 0.055 ---------- 6.65510
5
2
3
0.055
0.055
0.11
0.0825
0.0825
0.06875
33.33
20.00
1.622104
5.563105
4
5
0.055
0.06188
0.06875
0.06875
0.06188
0.06531
11.11
5.263
4.48410
2.593105
5
7
8
.
0.06188
0.06188
.
0.06359
0.06273
.
0.06273
0.0623
.
1.370
0.6897
.
3.176106
6.49710
7
9
10
0.0623
0.0623
0.06273
0.06252
0.06252
0.06241
0.3436
0.1721
1.265106
3.0768107
39
Table 1 Cont
7/28/2019 Lectures 3 4 5
40/98
Table 1 Cont.
Hence the number of significant digits at least correct is given
b the lar est value orm for which
105.02
m
a
105.01721.0
2
2
m
m
( ) 23442.0log.
m
..og =m
2=m
The number of si nificant di its at least correct in the estimated
40root of 0.06241 at the end of the 10
th
iteration is 2.
7/28/2019 Lectures 3 4 5
41/98
See also example 3-1: for a MATLAB
realization!!
Dr A.I. Delis TUC
2012Part 2 41
7/28/2019 Lectures 3 4 5
42/98
The False-Position Method
(Regula-Falsi) If a real root isl u
f(x)=0, then we canapproximate the
linear interpolationbetween the points [xl,f x and x f x tofind the x
r
value suchthat l(xr)=0, l(x) is thelinear approximationof f(x).
==
Dr A.I. Delis TUC
2012Part 2 42
False Position Method: derivation
7/28/2019 Lectures 3 4 5
43/98
False-Position Method: derivation
Based on two similar triangles, shown in Figure,one gets:
x x
r l r ux x x x=
The signs for both sides of Eq. (1) is consistent, since:
( ) 0; 0l r lf x x x< >
( ) 0; 0u r uf x x x> 0 then xu=xr == > fu=f(xr)
r need go no further!
Dr A.I. Delis TUC
2012Part 2 46
7/28/2019 Lectures 3 4 5
47/98
. w xl xu u
convergence to be declared. If they are not go back
.
Why this method?
Faster but not alwa s
Always converges for a single root.
Note: Always check by substituting estimated root in the
r .
Dr A.I. Delis TUC2012
Part 2 47
7/28/2019 Lectures 3 4 5
48/98
Dr A.I. Delis TUC2010
Part 2 (Chapters 5-6-7) 48
7/28/2019 Lectures 3 4 5
49/98
An example of slow convergence
10==
Dr A.I. Delis TUC2012
Part 2 49
Table 2 for Example 1
7/28/2019 Lectures 3 4 5
50/98
Table 2 for Example 1
010993.3165.0423
=+=
xxxfTable 2: Root offor False-Position Method.
Iteration x x mx % x1 0.0000 0.1100 0.0660 N/A -3.1944x10-5
2 0.0000 0.0660 0.0611 8.00 1.1320x10-5
3 0.0611 0.0660 0.0624 2.05 -1.1313x10-7
4 0.0611 0.0624 0.0632377619 0.02 -3.3471x10-10
50
2
7/28/2019 Lectures 3 4 5
51/98
m2
m
a
2105.002.0
.
m
21004.0
2)04.0log(
m
)3979.1(2
.
m
3
3979.3
=
mSo
m
The number of significant digits at least correct in the
51
.
is 3.
Open Methods
7/28/2019 Lectures 3 4 5
52/98
p
based on formulas
that re uire onl asingle starting valueof x or two startingvalues that do notnecessarily bracket
e roo
Open methods can
e er verge orconverge (rapidly)
Dr A.I. Delis TUC2012
Part 2 52
Sim le Fixed oint Iteration
7/28/2019 Lectures 3 4 5
53/98
Sim le Fixed- oint IterationRearrange the function so that x is on the
0 == xx
e s e o e equa on:
...2,1,k,given)( 1 == okk xxgx
Bracketing methods are convergent.
Fixed-point methods may sometimes
diver e de endin a on the statin oint(initial guess) and (b) how the function
Dr A.I. Delis TUC2012
Part 2 53
.
Exam le:
7/28/2019 Lectures 3 4 5
54/98
Exam le:
xxxxf 02)( 2 =
xxg 2)( 2 =
or
or
xg
2
1)(+=
x
Dr A.I. Delis TUC2012
Part 2 54
Conver ence
7/28/2019 Lectures 3 4 5
55/98
Conver ence
x=g x can e expresse
as a pair of equations:
y1=x
= x com onent
equations) .
Dr A.I. Delis TUC2012
Part 2 55
Conclusion
7/28/2019 Lectures 3 4 5
56/98
Fixed-point iteration converges if
x)f(x)linetheofslopethethanless(i.e.1)( = xg
When the method converges, the error is
ee a so sec on . pages -
roughly proportional to or less than the error ofthe revious ste therefore it is called linearl
convergent.
100%=approx.current
approx.previous-approx.currenta sa
quadratic convergence (if
n a guess su c en y
close to root)
for functions whose
derivatives cannot beevaluated analytically.
Dr A.I. Delis TUC2012
Part 2 (Chapters 5-6-7) 60
Cases where the NR exhibits poor
7/28/2019 Lectures 3 4 5
61/98
p
convergence
(a) f(x) has an inflection point near root
(b) NR oscillates around local min
(c) Jump from away from root
"ero s ope =
Dr A.I. Delis TUC2012
Part 2 (Chapters 5-6-7) 61
Example 1:NR Method
7/28/2019 Lectures 3 4 5
62/98
The floating ball has a specific gravity of 0.6 and has aradius of 5.5 cm. You are asked to find the depth to which
e a s su merge w en oa ng n wa er.
The equation that gives the depth x to which the ball is submerged under water is given by
423
a) Use the NR method of finding roots of equations to find the depth x to which the ball issubmerged under water. Conduct 3iterations to estimate the root of the above equation.
.. =
62
,significant digits at least correct at the end of each iteration.
Example 1 Cont.
7/28/2019 Lectures 3 4 5
63/98
Solve for ( )xf'
( ) .+x.-xxf -
'
1099331650
2
423=
x-xx .=
Let us assume the initial guess of the root of is
. . This is a reasonable guess (think why
0=xf
0 0.05mx = 0=x
and are not good choices) as the extreme values
of the depth x would be 0 and the diameter (0.11 m) of the
m11.0=x
ball.
63
Example 1 Cont.
7/28/2019 Lectures 3 4 5
64/98
The estimate of the root is
( )' 00
01 =xf
xxx
( ) ( ) 10.993305.0165.005.005.0 2423
+=
10118.10.05
...
3
4
=
( )01242.00.05 =
.=
64
Example 1 Cont.
7/28/2019 Lectures 3 4 5
65/98
65
Estimate of the root for the first iteration for the NR method.
Example 1 Cont.
7/28/2019 Lectures 3 4 5
66/98
The absolute relative approximate error at the end of Iteration 1a
xx1
=x
a
10006242.0
..
=
%90.19=
e num er o s gn cant g ts at east correct s , as you neean absolute relative approximate error of 5% or less for at least one
66
.
Example 1 Cont.
7/28/2019 Lectures 3 4 5
67/98
Iteration 2The estimate of the root is
( )
'
1
12=
xfxx
( ) ( ) 10.993306242.0165.006242.0
423
1
+=
( ) ( )
1097781.3
06242.033.006242.03.
7
2
1090973.8.
5
3
=
06238.0..
=
67
Example 1 Cont.
7/28/2019 Lectures 3 4 5
68/98
68
.
Example 1 Cont.
7/28/2019 Lectures 3 4 5
69/98
The absolute relative approximate error at the end of Iteration 2 is
a
12
=
xx
2
xa
100
06238.0
..=
.=
The maximum value ofm for which is 2.844.
m
a
2105.0
Hence, the number of significant digits at least correct in the
answer is 2.
69
Example 1 Cont.
7/28/2019 Lectures 3 4 5
70/98
Iteration 3The estimate of the root is
( )
'
223 =
xfxx
( ) ( ) 10.993306238.0165.006238.0
423
2
+=
( ) ( )06238.033.006238.03
.
11
2
1091171.8
.06238.0
3
=
06238.0
..
=
=
70
Example 1 Cont.
7/28/2019 Lectures 3 4 5
71/98
Estimate of the root for the Iteration 3 for the NR method.
71
Example 1 Cont.e a so ute re at ve approx mate error at t e en o terat on
7/28/2019 Lectures 3 4 5
72/98
e a so ute re at ve approx mate error at t e en o terat on
isa
10012
=xx
a
06238.006238.0
2
06238.0
=
=
The number of significant digits at least correct is 4, as only4 significant digits are carried through all the calculations.
72
On the Drawbacks of NR1 Divergence at inflection points
7/28/2019 Lectures 3 4 5
73/98
1. Divergence at inflection points
Selection of the initial guess or an iteration value of the root that is close to
root in the Newton-Raphson method.
3
==, .
The Newton-Raphson method reduces to .
.
( )2
33
1
512.01 +=
+
iii
xxx
Next table shows the iterated values of the root of the equation.
ix
.
close to the inflection point of .
Eventually after 12 more iterations the root converges to the exact value of
1=x
.2.0=x
73
NR Drawbacks Inflection Points
7/28/2019 Lectures 3 4 5
74/98
Table : Divergence near inflection point.
Iteration
Number
xi
0 5.0000
1 3.6560
.
3 2.1084
.
5 0.92589
6 30.119
7 19.746
18 0.20003
Divergence at inflection point for
. =+= xx
74
NR Drawbacks Division by Zero
7/28/2019 Lectures 3 4 5
75/98
2. Division by zero
For the equation
( )0104.203.0 623 =+= xxxf
the Newton-Raphson method
reduces to
ii xxxx104.203.0 623 +
=
ii xx .
For , the
denominator will e ual zero.
02.0or0 00 == xx
75
NR Drawbacks Oscillations near
7/28/2019 Lectures 3 4 5
76/98
Results obtained from the Newton-Ra hson method ma
.
oscillate about the local maximum or minimum without
converging on a root but converging on the local maximum or
minimum.
,
and may diverge.
For example for the equation has no real
roots.
022 =+= xxf
76
NR Drawbacks Oscillations near
7/28/2019 Lectures 3 4 5
77/98
5
6f(x)
Iteration
Newton-Raphson method.
4
Number
0 1.0000 3.00
ix i a
2
3
2
1
1
23
0.5
1.750.30357
2.25
5.0632.092
300.00
128.571476.47
0
-2 -1 0 1 2 3
x4
-1.75 -0.3040 0.5 3.142
4
5
6
3.1423
1.2529
0.17166
11.874
3.570
2.029
109.66
150.80
829.88-
Oscillations around local minima for ( ) 22 += xxf78
9
5.73952.6955
0.97678
34.9429.266
2.954
102.99112.93
175.96
77
NR Drawbacks Root Jumping4 Root Jumping
7/28/2019 Lectures 3 4 5
78/98
4. Root Jumping
In some cases where the function is oscillating and has a( )xfnum er o roo s, one may c oose an n a guess c ose o a
root. However, the guesses may jump and converge to some1.5
.1
For example
0
0.5
-2 0 2 4 6 8 10
x( ) 0sin == xxf
Choose
-1
-0.5
- . . . .
539822.74.20 == x
=
instead of-1.5
2831853.62 == x
78
oo ump ng rom n en e oca on o roo or
.( ) 0sin == xxf
The Secant Method
7/28/2019 Lectures 3 4 5
79/98
A slight variation of Newtons method for
evaluate. For these cases the derivative can be
difference.
1 1 xx ii
1
xx iii
,3,2,1)()(
)(1
11 =
=
+ i
xfxfxfx
ii
iiiii
Dr A.I. Delis TUC2012
Part 2
79
Re uires two initial
7/28/2019 Lectures 3 4 5
80/98
estimates of x , e.g, xo,x1. However, because
f(x) is not required to
change signs between
,classified as a
bracketin method.
The secant method has
Newtons method.
Convergence is not
guaranteed for all xo,
f(x).
Dr A.I. Delis TUC2012
Part 2
80
7/28/2019 Lectures 3 4 5
81/98
Dr A.I. Delis TUC2012
Part 2
81
7/28/2019 Lectures 3 4 5
82/98
Dr A.I. Delis TUC2012
Part 2
82
Multiple Roots
7/28/2019 Lectures 3 4 5
83/98
Example: f(x) =(x-3)(x-1)(x-1)(x-1)
Pose some difficulties:
.even roots (can use only open
2. In NR and secant f(x) goes to
3. NR and secant are now linearly
Dr A.I. Delis TUC2012
Part 2
83
Multi le Roots
7/28/2019 Lectures 3 4 5
84/98
None of the methods deal with multiple roots, ,
problems is as follows:
)()(Set
i
i
ixf
xu
=roots at all the same
locations as the
)(
-
ixuxx =
original function
)('')()(')('
'
i
xfxfxfxf
xu
=
)(')(
)]('[ 2
iixfxf
xf
Dr A.I. Delis TUC2012
Part 2
84)('')()]('[ 2
1
iii
iixfxfxf
+
7/28/2019 Lectures 3 4 5
85/98
original equation to determine whether f(xr) 0.
ASSIGMENT 1 PROBLEMS:
Dr A.I. Delis TUC2012 Part 2
85
m f n-Lin r E i n
7/28/2019 Lectures 3 4 5
86/98
0),,,,( 3211 =nxxxxf
0),,,,( 3212 =nxxxxf
0),,,,( =xxxxf
Dr A.I. Delis TUC2012 Part 2
86
ay or ser es expans on o a unc on o more an
i bl
7/28/2019 Lectures 3 4 5
87/98
one variable
)()( ii yyu
xxu
uu
+
+=
ii vv
yx
11111 iiiiii
yx
=+++
The root of the equation occurs at the value of x
i+1 i+1 .
Dr A.I. Delis TUC2012 Part 2
87
uuuu ii
iiii
ii
i11
7/28/2019 Lectures 3 4 5
88/98
yyxxuyyxxiiiii
++=+++ 11
vy
vxvy
vx
v ii
i
iii
i
i
i +
+=
+
++
11 yxyx
-
unknowns that can be solved for.
Dr A.I. Delis TUC2012 Part 2
88
i
i
i
i
7/28/2019 Lectures 3 4 5
89/98
yvyu ii
x
vuv
x
u iiiiii
+1
uv
vu ii
vuvu
xxyyiiiiii
=+
1
xyyx Determinant of
e aco an othe system.
Dr A.I. Delis TUC2012 Part 2
89
Roots of Pol nomials
Ch t 7
7/28/2019 Lectures 3 4 5
90/98
Chapter 7
nxaxaxaax ++++= 2
Follow these rules:
1. For an nth order equation, there are n real or
com lex roots.
2. Ifn is odd, there is at least one real root.
3. If complex root exist in conjugate pairs (that is,
+ i and - i , where i=s rt -1 .
Dr A.I. Delis TUC2012 Part 2
90
7/28/2019 Lectures 3 4 5
91/98
The efficac of bracketin and o en methodsdepends on whether the problem being solved
involves com lex roots. If onl real rootsexist, these methods could be used. However,
open and bracketing methods, also the openmethods could e susceptible to divergence.
Special methods have been developed to find
Mller and Bairstow methods.
Dr A.I. Delis TUC2012 Part 2
91
7/28/2019 Lectures 3 4 5
92/98
Mllers method obtains a root estimate
projecting a parabola to the x axis through three
function values.
Dr A.I. Delis TUC2012 Part 2
92
7/28/2019 Lectures 3 4 5
93/98
The method consists of deriving the
coefficients of arabola that oes throu h thethree points:
1. Write the equation in a convenient form:
cxxbxxaxf ++= )()()( 22
22
Dr A.I. Delis TUC2012 Part 2
93
2. The parabola should intersect the three points
[xo, f(x
o)], [x
1, f(x
1)], [x
2, f(x
2)]. The coefficients of the
l i l b ti t d b b tit ti th
7/28/2019 Lectures 3 4 5
94/98
o o 1 1 2 2polynomial can be estimated by substituting three
points to give
cxxbxxaxf ooo ++= )()()( 22
2
cxxbxxaxf ++= )()()(
2
21
2
211
2. Three equations can be solved for three unknowns,
22222
a, b, c. Since two of the terms in the 3r equation
are zero, it can be immediately solved forc=f(x2).
)()()()( 212
2121
222
xxbxxaxfxf
xxxxaxx ooo
+=
+=
Dr A.I. Delis TUC2012 Part 2
94
xxhxxh ==
7/28/2019 Lectures 3 4 5
95/98
x-xhx-xh 121o1o ==
)()()()( 121
1 xfxfxfxf oo
=
=
)()( 112
11
121
hhahhbhh oooo
o
+=++
11
2
11 hahbh=
and b
211
1xcahha o
==+=
Dr A.I. Delis TUC2012 Part 2
95
quadratic formula:
7/28/2019 Lectures 3 4 5
96/98
quadratic formula:
xx2
23
+=
The error can be calculated as
%10023 xx
a
=3
, .This will result in a largest denominator, and will give root
estimate that is closest to x .
Dr A.I. Delis TUC
2012 Part 2
96
Once x3 is determined the process isi h f ll i i li
7/28/2019 Lectures 3 4 5
97/98
Once x3 is determined, the process isr in h f ll in i lin
1. If only real roots are being located, choose the
estimate, x3.
. o rea an comp ex roo s are es ma e ,
employ a sequential approach just like in secantme o , x1, x2, an x3 o rep ace xo, x1, an x2.
See
Dr A.I. Delis TUC
2012 Part 2
97
7/28/2019 Lectures 3 4 5
98/98
Dr A.I. Delis TUC
2012 Part 2
98