DIFFERENTIAL GEOMETRY OF CURVES 69Px LxyLn
in
in
i
–1
=0–1() = Σ –1() (3.8)
Example 3.1. Construct a polynomial to interpolate through the data points (0, 0), (1, 2), (3, 2)
and (6, −1) using the Newton’s divided difference and Lagrangian approaches. Perturb point (3, 2) to
(1.5, 4) and observe the change in the curve shape.
Using Newton’s divided difference approach, since there are four data points, the interpolating
polynomial is a cubic, that is
y=α 0 +α 1 (x− x 0 )+α 2 (x− x 0 ) (x− x 1 ) +α 3 (x− x 0 ) (x− x 1 )(x− x 2 )
Now
α 0 = y 0 = 0α
α
110
10=
- = 2 – 0
1 – 0
= 2
y
xx
- = 2 – 0
ααα
220 120
2021=( – ) – ( – )
( – )( – )
=(2 – 0) – 2(3 – 0)
(3 – 0)(3 – 1)
= –^2
3yxx
xxxxααα α
3
30 130 23031
303132=( – ) – ( – ) – ( – )( – )
( – )( – )( – ) =(–1) – 2 (6) +^2
3
(6)(5)
(6)(5)(3)
= 907yxxxxxx
xxxxxxThe polynomial becomes
yx xx = 2 –^2 xx x
3
( – 1) +^7
90
( – 1)( – 3)Using the Lagrangian approach, the polynomial is
yLxy Lxy Lxy Lxy = 03 () + () + 0 13 1 23 () + 2 33 () 3or
y
xx xx xx
xxxxxx
y
xx xx xx
xxxxxx
= y
( – )( – )( – )
( – )( – )( – )
+
( – )( – )( – )
( – )( – )( – )123
010 203
0
023
101213
1+
( – )( – )( – )
( – )( – )( – )
+
( – )( – )( – )
( – )( – )( – )013
202123
2012
303132
3xx xx xx
xxxxxx
y
xx xx xx
xxxxxx
y=( – 1)( – 3)( – 6)
(–1)(–3)(–6)
0 +( – 0)( – 3)( – 6)
(1) (–2)(–5)
2xxx xxx+( – 0)( – 1)( – 6)
(3)(2)(–3)
2 +( – 0)( – 1)( – 3)
(6)(5)(3)
(–1)xxx xxx=( )( – 3)( – 6)
5( )( – 1)( – 6)
9( )( – 1)( – 3)
90xx x xx x xx xOn simplification, the above yields the same result as that from the Newton’s divided differences
method. Moving data point (3, 2) to (1.5, 4) requires re-computing the curve. With the divided
difference approach, only the last two coefficients need to be computed, that is