Académique Documents
Professionnel Documents
Culture Documents
6 x 9 x 4 x 20 x +26=0
It is shown in figure 1 that there are three roots to this equation. That which is labelled root c will be attempted to be
found.
x
-5
-4
-3
-2
-1
0
1
2
3
4
5
f(x)
-23749
-8086
-1993
-238
35
26
-1
2
587
3530
12551
From the changes in sign in table 1, (+, green rows or -, red rows), it can be seen that there are roots in the intervals (2, -1), (0, 1) and (1, 2). These changes are shown as the blue rows of the table. The integer interval which contains
root c can be deduced to be (1, 2). The integer search process of finding this interval is shown for the two bounds here:
f ( 2 )=6 ( 2 ) 9 ( 2 ) 4 ( 2 ) 20 ( 2 ) +26=1921443240+26=2
6
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Root c
Root a
Root b
x
f(x)
1
-1
1.1
-4.83784
1.2
-8.64448
1.3
-12.21532
1.4
-15.28096
1.5
-17.5
1.6
-18.45184
1.7
-17.62948
-14.43232
1.8
x
-8.15896
1.9
1.983
2
1.9831
2
1.9832(1.9, 2)
x is in the interval
1.9833
x=2 to 1 significant
1.9834 figure
x=1.950.05
1.9835
Table 2: the
1.9836
search
1.9837
repeated
with an
1.9838
5 significant
1.9839
achieved
1.984
x
f(x)
1.9
-8.15896
1.91
-7.332245319
1.92
-6.465898701
1.93
-5.559037574
1.94
-4.610767706
1.95
-3.620183125
1.96
-2.586366054
1.97
-1.508386836
-0.385303859 x
f(x) 1.98
1.99
0.7838365091.9833
-0.039441481
2
-0.0278410452
1.98331
-0.016235973
x is in the interval (1.98,1.98332
1.99)
-0.004626265
1.98333
0.00698808x=1.9 to 2 significant figures
1.98334
x=1.9850.005
0.018607064
1.98335
0.030230687
1.98336
0.041858951
1.98337
0.053491855
1.98338
0.065129402
1.98339
0.076771592
1.9834
x
f(x)
1.98
-0.385303859
1.981
-0.270478244
1.982
-0.155191106
1.983
-0.039441481
1.984
0.076771592
1.985
0.193449077
1.986
0.31059194
1.987
0.428201146
1.988
0.546277664
f(x)
1.989
-0.004626265 0.664822462
1.99
0.783836509
-0.003465039
-0.002303767
x is in the interval (1.983, 1.984)
-0.001142448
x=1.98 to 3 significant figures
1.89167E-05
x=1.98350.0005
0.001180328
decimal
0.002341786
process,
0.00350329
until a result
accuracy of
0.00466484
figures is
0.005826437
0.00698808
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
f(1) = -1
f(2) = 2
Figure 2: a graphical illustration of y = f(x), showing that the root is in the interval [1, 2)
For the first integer search, the graph in figure 2 shows again that f(1) is negative and that f(2) is positive, and
therefore, the curve must cross the x-axis somewhere in the interval between these two values i.e. there is a root in the
interval (1, 2).
f(1.98334) > 0
f(1.98333) < 0
Figure 3: graphical illustration of y = f(x) that the root is in the interval [1.98333, 1.98334)
For the final decimal search, the graph in figure 3 shows again that f(1.98333) is negative and that f(1.98334) is
positive, and therefore, the curve must cross the x-axis somewhere in the interval between these two values i.e. there is
a root in the interval (1.98333, 1.98334).
The same process can be repeated with roots a and b to give answers of x = -1.3889 and x = 0.97356 respectively, each
to five significant figures. The original equation has therefore been solved.
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Root b
Root c
Root d (repeated)
f(4) > 0
f(5) > 0
Root a
Figure 4: a graph of the function y = f(x), where f(x) = 1323x5-11277x4+14397x3+68513x2-80476x-153760, showing that there is a repeated root
When the graph for y = f(x) (figure 4) for this equation is plotted, it only touches the x-axis at a single point. It never
crosses it. In this case, on both sides of the root of f(x) = 0 (intersection with the x-axis), f(x) > 0 so the sign is always
remains positive and no change of sign actually occurs.
This is the case because there is a repeated root for the equation above. Because it happens to be an irrational one, and
therefore nonterminating, no amount of repetitions of the decimal search process in the interval (4, 5) will return a
value of zero (and thereby identify the root outright).
x
-3
-2
-1
0
1
2
3
4
5
6
f(x)
-919360
-56700
-31768
-153760
-161280
-63580
18200
9792
42560
612260
Table 3 shows how an integer search suggests that there are just two roots in the interval searched. Not only does it
show that the technique does not suggest that there is a root in the interval, (4, 5), it also does not show that there is a
root in the interval (-2, -1) because there are two distinct roots occurring close together within the said interval, and
both f(-2) and f(-1) are negative so it wouldnt be seen to be necessary to conduct a decimal search on this interval.
The change of sign method fails here because it has failed to identify the root in the intervals (-2,-1) and (4, 5).
Newton-Raphson method
This method works by taking an estimated root and finding where its tangent to the curve crosses the x-axis. The xcoordinate of this intersection is taken and its tangent to the curve is taken and the process is repeated. In most cases
the x-coordinate becomes closer to the root with each repetition.
Taking x0 as the first rough estimate, the point on the curve for this x-value would be (x 0, f(x0)). A general straight line
can be written in the form
y y 0 =m(xx 0)
The tangent to the curve, with the gradient being the derivative of f(x) at x 0 (f(x0)), would be expressed as the
following function:
6
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
yf ( x 0) =f ' ( x ) [ xx 0 ]
The next x-coordinate will be called x1. It is where the line stated above crosses the x-axis, in other words, when y = 0.
Substituting this into the function above, as well as changing x to x1 to make it the first iteration, it can be rearranged
to give the following:
x 1=x 0
f (x 0)
f '(x )
When this is generalised for when x0 = xn, the inductive formula is given as:
x n+1=x n
f ( x n)
f ' ( x n)
x +3 x 13 x +4=0
Root k
Root l
Root m
Figure 5: a graph of the function y = f(x), where f(x) = x3+3x2-13x+4
Because f(x) = 0 where the curve intersects the x-axis, it can be seen in figure 5 that there are three distinct roots to the
equation, labelled as k, l and m. They are in the intervals (-6, -5), (0, 1) and (2, 3) respectively. Without a graph, these
intervals could have been obtained using an integer search using the change of sign method.
As previously explained, the iterative formula for the Newton-Raphson method is:
x n+1=x n
f ( x n)
f ' ( x n)
Because here, f(x) = x3+3x2-13x+4, f(x) = 3x2+6x-13. Substituting these into the iteration:
3
x n+1=x n
x + 3 x 13 x +4
3 x 2 +6 x13
This will be used, first, to find an approximation for root k, in the interval (-6, -5). The value for x 0 will be -6 as it is
the closest integer to the root according to the graph.
x 0=6
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
x 1=x 0
3
2
x 03 +3 x 0213 x0 + 4
(6 ) +3 (6 ) 13 (6 ) + 4
=
(
6
)
=5.559322 *
2
3 x 02 +6 x 013
3 (6 ) +6 (6 )13
x 2=x 1
3
2
x 13 +3 x1213 x 1+ 4
(5.56 ) +3 (5.56 ) 13 (5.56 ) +4
=
(
5.56
)
=5.498338 *
2
3 x 12+ 6 x113
3 (5.56 ) +6 (5.56 )13
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
x0
x1
x2
x3
x4
-6
-5.559322
-5.498338
-5.497205
-5.497204
f(xn)
-26
-2.827377677
-0.050643003
-1.73165 x10-5
-2.01794 x10-12
f(xn)
59
46.362252
44.705122
44.674551
44.674541
Table 4: iterations for root k. All values of xn have been found using the iterative formula shown above
From the fact that the root is close to -5.497204, and xn for values where n>3 round to this value (to seven significant
figures), it can be suggested that it lies in the interval [-5.4972035, -5.4972045), or that root k is x=5.4972040.0000005, which can also be expressed as x = -5.4972 to five significant figures, since both bounds round
to this value. It has taken four iterations to obtain this answer.
The verification for these bounds is as follows:
3
Graphical explanation
The tangent to the curve at the initial value of x, x0 = -6 is taken (figure 6). From this x-coordinate, the tangent to the
curve is taken and the process is repeated three times (figure 7). Vertical lines are drawn to show where the next
iteration of x is on the curve.
x0 = -6
x1 -5.56
x2 -5.50
Root k
Root k
Figure 7: only x1 and x2 have been labelled as due to the resolution, it would y =
be impractical to label further iterations of x
Further roots
Table 5 shows that for the root l in the interval (0, 1), it has taken four iterations of this method to obtain the answer x
= 0.33681 to five significant figures.
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
x0
x1
x2
x3
x4
0
0.3076923
0.3365018
0.3368101
0.3368101
f(xn)
4
0.313154301
0.003280019
3.80968 x10-7
5.32907 x10-15
f(xn)
-13
-10.8698
-10.6413
-10.6388
-10.6388
2
2.181818
2.160698
2.160394
2.160394
f(xn)
-2
0.30353118
0.004248277
8.7684 x10-7
3.55271 x10-14
f(xn)
11
14.3719
13.97004
13.96428
13.96428
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Root d
Root e
Figure 8 shows that there are two roots to this equation. The Newton-Raphson method will be applied to try to find
root d with an estimate by eye of x0 = -2.
The iteration formula is given by the following:
x n+1=x n
-2
1.4166667
1.0002484
0.3829392
1.4726444
1.0521629
0.5130953
1.7942253
1.3195768
0.9027871
-0.026547
1.6845599
1.2323494
0.8028538
-1.400101
-1.643706
-1.660733
-1.660956
-1.660956
f(x)
0.41
-3.472899
-1.390559
-1.36681
-3.970262
-1.52211
-1.221009
-8.193961
-2.749496
-1.218257
-1.900147
-6.466232
-2.235909
-1.128521
-0.602357
-0.035315
-0.00045
-8.04 x10-8
-2 x10-15
f'(x)
-0.12
-8.33993
-2.25261
1.254293
-9.44218
-2.8236
0.953072
-17.2632
-6.59684
-1.31089
1.110478
-14.2992
-5.2059
-0.51228
-2.47268
-2.07403
-2.02059
-2.01986
-2.01986
Table 7
Instead of giving the correct result for root f, the application of the method has instead converged towards x = -1.6610
to five significant figures root e. Figure 9 shows graphically why this has happened. Because x = -2 is close to a
turning point, the gradient of the tangent near it is changing at a quick rate as it turns around the local maxima. It is
therefore not relatively consistent, where this method works best. Its shallow gradient can be illustrated by the large
number of iterations required to obtain an answer (at least compared to the previous example).
6
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Because the tangent of x = -2 has a negative gradient as it is on the other side of the turning point to root d, it crosses
the x-axis to the left of this initial x value and so begins to converge towards the root just less than x = -2, instead of
the one just more than it, which was the one desired.
Figure 9: y = f(x), showing graphically all the iterations of the formula required to find the root
In figure 9, the tangent to the curve at x = 2 crosses the y-axis on the right side of the turning point, and so successive
iterations converge towards the root on the right side of the turning point (root e) instead of that on the left (root d).
This is what causes the failure for finding root d, with x 0 = -2.
x n+1=g ( x n )
The application of this produces a converging sequence of numbers towards the value at which the line y = x crosses y
= g(x) on a graph. Algebraically, this, through simultaneously solving through equating these two equations, gives, at
the point where x = g(x), the x value for which f(x) = 0. It therefore finds the root.
99 x +5 x 7 x+ 2=0
This can be arranged into many different new equations for x = g(x). One of these is:
1
5 x 27 x +2
1
2
x=
=( ( 5 x 7 x +2 ) )3
99
99
3
Figure 10 graphically shows that there is one root to the equation; it is in the interval (0, 1).
Therefore, taking x0 = 0 (judging by eye), the application of the iterative formula provides the sequence shown in table
8. At the 17th iteration, the same result is given as the previous iteration, to six significant figures. This value rounds to
0.20047 to five significant figures.
The verification for this is as follows, by taking the values of f(x) for the upper and lower bounds of this value:
3
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Root
x0
x1
x2
x3
x4
x5
x6
x7
x8
x9
x10
x11
x12
x13
x14
x15
x16
x17
0.000000
0.272353
0.167400
0.213859
0.194787
0.202838
0.199481
0.200888
0.200300
0.200546
0.200443
0.200486
0.200468
0.200475
0.200472
0.200474
0.200473
0.200473
x 0=0
5( 0)27(0)+2
x 1=
=0.272353 *
99
3
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
5( 0.27)27(0.27)+2
x 2=
=0.167400 *
99
3
Figure 11 graphically shows how the rearrangement of f(x) = 0 converges towards the root. The two lines mentioned
have been plotted and the iterative formulas graphical illustration is shown in blue. The blue line shows convergence
in a rotating pattern like a cobweb.
From the point on the curve y = g(x0), a horizontal line is drawn across to meet the line y = x, so that the x value is
found where both lines have the same y value. This new value for x becomes x 1. From here, a horizontal line, again, is
drawn across from the curve at the point which has the x value of x 1 and the same process is repeated to find
subsequent values of xn, where n > 1. These values of x converge towards the value of x where the two lines intersect,
the same as the root of f(x). The labelled graphical values of x n correspond with those in the table.
x2
x0
x4
x5
x3
x1
Figure 11: graphical illustration of y = g(x) (red line) and y = x (green line) with the rearrangement iterations shown in blue
(
g ( x )=
'
10 x7
99
)((
5 x 27 x+2
99
3
))
2
3
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
1
3
10
7
( 0.2 )
99
99
( )(( )
g' ( 0.2 ) =
1
( 5 ( 0.2 )27 ( 0.2 ) +2 )
99
( ))(( )
2
3
0.418 |0.418|< 1
x=
99 x +5 x +2
7
99 x n3 +5 x n2+ 2
x n+1=
7
Table 9 shows iterations of this formula, again starting with x 0 = 0. It can be deduced that xn does not find the root
because the values seem to reoccur from the seventh and eighth iterations onwards, according to this degree of
accuracy.
Figure 12 graphically illustrates this failure, showing that the first couple of iterations form a cobweb pattern, but then
goes round and round on itself, forming a stationary rectangle. This is the observation that the values for each iteration
of xn change from 0.013857 and 0.285814 and keep doing so, seemingly forever.
x0
x1
0.285714
x2
0.014161
x3
0.285817
x4
0.013846
x5
0.285814
x6
0.013857
x7
0.285814
x8
0.013857
x9
0.285814
x10
0.013857
x11
x1428
0.285814
0.013857
x1429
0.285814
x1430
0.013857
x1431
0.285814
Using the rearrangement method with h(x) fails because at the root, x 0.2, h(x) has a greater magnitude that 1:
'
h ( x )=
( 10297 x ) x
7
'
h ( 0.2 )=
1.41|1.41|>1
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Because this value is negative, the iteration has a cobweb pattern, but it also does not converge towards the root.
x0
x2
x1
Figure 12: graphical illustration of y = h(x) (red line) and y = x (blue line) with the rearrangement iterations shown in green. It is impractical to
show labels for further iterations
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
f(x)
-238
-174.41884
-122.52448
-80.70832
-47.51296
-21.625
-1.86784
12.80552
23.31968
30.48404
35
f(x)
-0.022685742
-0.006189195
0.010302388
0.026789005
0.04327066
0.059747353
0.076219084
0.092685855
0.109147667
0.12560452
0.142056416
x
-1.4
-1.39
-1.38
-1.37
-1.36
-1.35
-1.34
-1.33
-1.32
-1.31
-1.3
x
-1.3889
-1.38889
-1.38888
-1.38887
-1.38886
-1.38885
-1.38884
-1.38883
-1.38882
-1.38881
-1.3888
f(x)
-1.86784
-0.187924509
1.442231859
3.023594836
4.557118054
6.043743125
7.484399706
8.880005574
10.2314667
11.53967732
12.80552
f(x)
-0.006189195
-0.004539813
-0.002890481
-0.001241199
0.000408034
0.002057217
0.00370635
0.005355434
0.007004468
0.008653453
0.010302388
x
-1.39
-1.389
-1.388
-1.387
-1.386
-1.385
-1.384
-1.383
-1.382
-1.381
-1.38
f(x)
-0.187924509
-0.022685742
0.142056416
0.306302934
0.47005478
0.633312923
0.796078328
0.958351961
1.120134786
1.281427764
1.442231859
x
-1.38887
-1.388869
-1.388868
-1.388867
-1.388866
-1.388865
-1.388864
-1.388863
-1.388862
-1.388861
-1.38886
f(x)
-0.001241199
-0.001076273
-0.000911348
-0.000746424
-0.0005815
-0.000416576
-0.000251653
-8.67305E-05
7.81915E-05
0.000243113
0.000408034
Newton-Raphson method
Using x0 = -1, the Newton-Raphson method takes seven iterations to find the same result to five significant figures,
shown in table 11.
x0
x1
x2
x3
x4
x5
x6
x7
-1
-2.029411765
-1.679330827
-1.473319358
-1.398239983
-1.38899275
-1.3888625
-1.388862474
Table 11: iterations to solve the chosen root of f(x) using the Newton-Raphson formula
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
Because Newton-Raphson does not give an interval which the root is in, it is necessary to verify the assumed result.
This is shown below, using the interval bounds for the approximation of x = -1.3889 to 5 significant figures.
f (1.38885 )=0.002057217> 0
f (1.38895 )=0.0144368<0
Rearrangement into x = g(x) method
A rearrangement of f(x) as an iterative formula is the following:
9 x n4 + 4 x n3 +20 x n26
x n+1=
6
1
5
Iterations are shown in table 12. The root is shown to be x = -1.3889 to five significant figures.
x0
x1
x2
x3
x4
x5
x6
x7
x8
x9
x10
x11
x12
x13
x14
-1
-1.46868
-1.34256
-1.40875
-1.37891
-1.39351
-1.38662
-1.38993
-1.38835
-1.38911
-1.38875
-1.38892
-1.38884
-1.38888
-1.38886
Again, for the same reasons as for the Newton-Raphson method, this answer must be verified using the upper and
lower bounds of the approximation interval.
f (1.38885 )=0.002057217> 0
f (1.38895 )=0.0144368<0
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
The rearrangement method took longer to converge and it can often take a large number of iterations to converge to an
accuracy of five significant figures. Rearrangement is also quite easy and quick to do.
The decimal search was the slowest in finding the roots. Although it only took six iterations, without extremely
complex spreadsheet formulas, each iteration needed to be set up individually on spreadsheet software through
identifying the two bounds from the previous iteration. Simple and reliable though this method is, it is much more
time-consuming than the other two methods. Each iteration also included nine unique calculations involving x.
Technologies
Software-based spreadsheet
Using spreadsheet software like Google Docs Spreadsheet, Microsoft Excel or Numbers, all three methods are quite
easy and quick to apply. Spreadsheets have the ability to quickly copy formulas and reapply them in a very short
amount of time. This is done by keyboard shortcuts, context menus or by dragging or double clicking a cells fill
handle. They can also be made to make the processes automatic through the use of complex spreadsheet formulas, a
very strong advantage for bulk calculations of roots of equations. A disadvantage is that all mathematical formulas
must be typed in using a non-dedicated QWERTY keyboard on desktops and laptops, which can be very clunky and
tedious. This is even worse on tablets and smartphones, where the user must scroll through many pages of symbols on
their keyboard to find commonly used mathematical symbols.
Using purely software-based spreadsheets, the decimal search method is easiest to use as it does not require any sort of
mathematical calculation, like rearranging or differentiating, just simply typing in a spreadsheet formula for the
function f(x).
Scientific calculator
Iterative methods, like Newton-Raphson or rearranging for x = g(x), can be easily done on scientific calculators using
the answer button, or equivalent. A formula can be typed in, replacing x n with the answer button, and pressing the
equals button gives the result for the next iteration.
The decimal search can, at a push, also be used, but it would take a very long time a very many button presses,
because each calculation for f(x) must be typed in separately nine separate polynomial calculations for each
iteration.
A disadvantage of using a scientific calculator is the fact that the small display and linear input makes it easy to forget
to close brackets, resulting in an error.
Newton-Raphson is, in my opinion, the easiest method to use with just a scientific calculator as it is iterative and the
calculation would normally be relatively straightforward to work out and type in.
Graphical calculator
In the same way as scientific calculators, iterative formulas are easy to apply using graphical calculators. Some also
have the advantage of being able to differentiate with respect to a variable like x, meaning the user doesnt have to if
they are lazy.
Iterative formulas can also be put into tables, much like a spreadsheet. This makes it easier to spot convergence. Using
the table calculator function, the decimal search can also easily be used.
I consider the decimal search the easiest method to use because it is easiest to type in just a simple polynomial
equation, as opposed to complex iterative formulas of the other two methods.
Graphing software
Advanced graphing software like Autograph can be used to apply the Newton-Raphson and rearrangement methods in
a graphical way.
For Newton-Raphson, Autograph requires for the graph of y = f(x) to be selected, then bringing up the context menu
and selecting the Newton-Raphson option brings up a dialog box. This enables iterations of the formula to be plotted
with an x0 value.
The x = g(x) method can be graphically plotted in the same way, but first, the line y = x must be plotted.
Maurice Yap 6946 Core 3 Mathematics Coursework 4752/02 Methods for Advanced Mathematics
In my opinion, both these methods are equally easy to apply using Autograph software, but the Newton-Raphson
method is the most reliable, so it would probably be my preference.
* For presentation purposes, the values printed are rounded, and not exact values. Subsequent uses of this
number are shown rounded further, however, unrounded exact results have been used in the calculations.