DS 288 (AUG) 3:0 Numerical Methods Homework–3

DS 288 (AUG) 3:0 Numerical Methods

Homework–3 1
Due date: September 30, 2021 (Thursday); 7:00 A.M.
1. Exercise Set 3.2, Problem#12 in Text (Page-124). Be sure to read the note about inverse
interpolation directly above the problem. Solve this problem using an iterated interpo-
lation approach (i.e., Neville’s algorithm). Report the relative error of your result and
what value you used for the exact solution. [1.5 points]

2. In some applications one is faced with the problem of interpolating points which lie on
a curved path in the plane, for example in computer printing of enlarged letters. Often
the complex shapes (i.e., alphabet characters) cannot be represented as a function of x
because they are not single-valued. One approach is to use Parametric Interpolation.
Assume that the points along a curve are numbered P1 , P2 , ..., Pn as the curved path
is traversed and let di be the (straight-line) distance between Pi and Pi+1 . Then define
ti = ⌃i=1§
j=1 dj , for for i =1, 2, ..., n (i.e., t1 = 0, t2 = d1 , t3 = d1 +d2 , etc). If Pi = (xi , yi ), one
9.

can consider two sets of data (ti , xi ) and (ti , yi ) for i = 1, 2, ...n which can be interpolated
independently to generate \the functions f (t) and g(t), respectively. Then P (f (t), g(t))
for 0  t  tn is a point in the plane and as t is increased from 0 to tn , P (t) interpolates
the desired shaped (hopefully!). Interpolation of the data given below via this method
should produce a certain letter. Adapt your algorithm from problem (1) to perform the
interpolations on f (t) and g(t) where t is increased from 0.0 to 12.0 in steps of dt (see
data below). Report the value of dt you use to achieve ‘reasonable’ results. Turn in a
plot of your interpolated shape (not the numeric values of P (t)) as well as plots of f (t)
and g(t) individually. [3 points]

ti xi yi
0.0 0.70 2.25
1.0 1.22 1.77
2.0 2.11 1.61
3.0 3.07 1.75
4.0 3.25 2.30
5.0 2.80 2.76
6.0 2.11 2.91
7.0 1.30 2.76
8.0 0.70 2.25
9.0 0.45 1.37
10.0 0.88 0.56
11.0 2.00 0.08
12.0 3.25 0.25

3. Repeat problem (2) with a natural cubic spline. Also report all four coefficients for each
of the cubics which comprise the interpolants for both f (t) and g(t). How does your letter
compare with that produced in problem (2)? Explain any di↵erences. [3.5 points]

4. Consider the oscillograph record of the free-damped vibrations of a structure. From

vibration theory, it is known that for viscous damping (damping proportional to velocity),
1
Posted on: September 2, 2021.

1
DS 288 (AUG) 3:0 Numerical Methods Homework–3
the envelop of such a vibration (i.e., the curve through the peaks of the oscillations) is an
exponential function of the form

2⇡ax
y = be

where x is the cycle number, y is the corresponding amplitude and a is a damping factor.
Using the three data points shown in the figure, determine a and b that result from a best
fit based on the least-squares criterion. Use a linear least squares approach by suitable
change of variable. You may solve this problem “by hand” if you wish.

An alternate approach to this problem would be to construct a nonlinear least-squares

fit using the data directly as given. Would this approach lead to exactly the same a and
b values you determine above (assuming perfect math, i.e., no rounding errors in either
case)?. [2 points]
Hint: see if you can relate the errors at the ith data point between the two fits.

viscous damping

2
 Least
Ans -147
change of
: with variables :
square
-
-

-2ha se
be
Y=
lnb-2-aa.se
lny
=

y
'
= b' + a' a
{ !÷?!aa
b' = tub

NOW ,

Errol E = EM [
'
( b' + a'se ) ]2
yi
-

,
i
izl

For minimum error ,

IIb ,
= 0 2
Em Kyi .
-
Cbi a' RID
-
c- 17=0

i-4
-
①
m
JE
and
{ Yi Cb taxi )
]
'
C-
0 => 2 [ Ri )=o
-

ga .

E- I -

From ① & ②

ZKFEY ;
'
b = -

2- seiyi Eni

MERY -

Exit
 ' MERIT ; -

EXiEYi
a =

MERY -

fait
en
given question .

| / gz.|gg#zg.,s-sgz/ |
Xi Yi Yi '=lnyi Xiii xi2

° " " "" ° °

,, ago

32 5 1.60944 51.50208 1024

I 48 31 6.63986 86.6576 1280

b' = 1280×6.63986-86.6576×48
3×1280 -
4832

lnb = 2.825167

b= 16.8638

NOW ,

3×86.6576 48×6-63986
a'
-

2
3×1280 -
µ 8)

Laa = -
O -

0382425

A = 6.0865×10-3
 (0-0382425) se
16.8638 e-
Y=
→ Non tinea Least Square : -
-

Errol Z( yi
Maki
)2
-

.
E= -
be
Iet

28 =° Em (
Yi
-

bé2aaki)é2aaxi = ☐

☒ i. i -

}±a=o Em (Yi -
be
-2mi
)(xié2aani)=o
i- I
-

egn ① &
@ cannot be solved for
' '

and
'

b' if exactly
this
nonlinear
a we use
,

least method
square .