Midterm Exam 2
1 Let x0 = −1, x1 = 0, x2 = 1, x3 = 2 and f (x) = x5 − 2x2 + 2. Let P3 (x) be the unique
interpolating polynomial of degree at most 3 interpolates f at x0 , x1 , x2 , x3 .
i (5 points) Construct the Newton form of P3 (x).
ii (5 points) Let w(x) = (x + 1)x(x − 1)(x − 2). Show that
max |w(x)| = 1.
x∈[−1,2]
1
Hint: Use x = t + 2 to be a new variable and find the extreme values of v(t) = w(t + 21 ).
iii (5 points) Prove that
max |f (x) − P3 (x)| ≤ 10.
x∈[−1,2]
iv (5 points) Find another set of interpolation points x̃0 , x̃1 , x̃2 , x̃3 so that the interpolating
polynomial P̃3 (x) has the smallest interpolation error in l∞ -norm. What is the bound
of approximate error maxx∈[−1,2] |f (x) − P̃3 (x)| you can guarantee? Just list the set of
interpolation points x̃i and give the upper bound. No need to prove.
2 (10 points) Let Ln,i (x) denote the Lagrange polynomial associated with n + 1 distinct points
{xi }n0 . Prove that
Xn
f (xi )Ln,i (x) = f (x).
i=0
if and only if f (x) is a polynomial of degree less or equal to n.
3 (10 points) Suppose P (x) is a polynomial of degree 5 and P (x) passes through (−2, 4), (−1, 2),
(1, −2), (2, 6), (3, 9), (4, 12). Use Neville’s algorithm to compute P (0).
4 (10 points) Find a polynomial P of degree at most 4 that satisfies
P (1) = P 0 (1) = 0, P (2) = 1, P 0 (2) = 0, P (3) = 1.
5 (10 points) Let x0 , x1 , x2 be 3 distinct points and P0,0,1, (x), P0,1,2 (x) be the Hermite inter-
polating polynomials of f (x) associating with x0 , x0 , x1 and x0 , x1 , x2 . That is, P0,0,1 (x0 ) =
0
f (x0 ), P0,0,1 (x0 ) = f 0 (x0 ) and P0,0,1 (x1 ) = f (x1 ); P0,1,2 (x0 ) = f (x0 ), P0,1,2 (x1 ) = f (x1 ) and
P0,1,2 (x2 ) = f (x2 ). Show that
(x − x0 )P0,1,2 (x) − (x − x2 )P0,0,1 (x)
P0,0,1,2 (x) =
x2 − x0
6 Let f (x) be a C 2n+2 -function on [a, b] and {xi }ni=0 be n + 1 distinct points.
i (10 points) Show that there exists a unique polynomial P (x) of degree at most 2n + 1 that
satisfies P (xi ) = f (xi ) and P 0 (xi ) = f 0 (xi ) for i = 0, 1, 2 . . . n.
ii (10 points) Let P be the unique polynomial defined in (i) and x ∈ (a, b), x 6= xi . Show that
there exists ξ ∈ (a, b) such that
n
f (2n+2) (ξ) Y
f (x) − P (x) = (x − xi )2 .
(2n + 2)!
i=0
1
7 Let s(x) be a cubic spline interpolant of f relative to the partition
a = x0 < x1 < x2 < x3 < x4 < x5 = b.
Suppose on each interval [xj , xj+1 ], s is written as
s(x) = sj (x) = aj + bj (x − xj ) + cj (x − xj )2 + dj (x − xj )3
and s(xj ) = aj = f (xj ). Let hj = xj+1 − xj .
i (5 points) What equations can you get from continuity of s(x), s0 (x) and s00 (x)?
ii (5 points) Derive the equation that involves cj−1 , cj , cj+1 , aj−1 , aj , aj+1 only by using
equations obtained in (i).
iii (5 points) Suppose s satisfies the nature boundary condition s00 (a) = 0 and s00 (b) = 0. What
equation does the boundary condition translate to in terms of the coefficients? Write down
the complete system for determining the cj .
iv (5 points) Show that the system obtained in (iii) is solvable.
v (10 points) Show that the nature cubic spline s defined in iii satisfies the minimum curvature
property: Let g be any C 2 -function on [a, b] which interpolates f over the partition
a = x0 < x1 < x2 < x3 < x4 < x5 = b.
Then Z b Z b
00 2
[s (x)] dx ≤ [g 00 (x)]2 dx.
a a
Moreover, the equality holds only if g(x) = s(x). You must provide the detail of your proof.
8 (10 points) Suppose f is a C ∞ -function on R. Use Taylor’s expansion to derive the finite difference
formula for approximating the first derivative f 0 (x). Find constants c1 , c2 , c3 , c4 , c5 such that
c1 f (x0 + h) + c2 f (x0 + 12 h) + c3 f (x0 ) + c4 f (x0 − 21 h) + c5 f (x0 − h)
f 0 (x0 ) = + O(hp )
h
and p is as large as possible. You must write down the system of equations for ci ’s and solve the
system to get answers.
9 (10 points) Let Dh denote the second-order central difference approximation for derivative D
obtained using a step size of h. Show that if we apply Richardson extrapolation to approximate
f 0 (x) by Dh f (x) and Dh/2 f (x), then we get the same formula derived from Problem 7.
