Description
1. Set up and plot the Runge function f(x) = 1
1 + ax2
on [−1, 1] in Matlab for a = 30. Report
your plot.
2. For n + 1 interpolation points x0, x1, …, xn, implement the following forms of the interpolating polynomial.
a). Write a Matlab function that solves the interpolation problem by the Vandermonde matrix.
b). Write a Matlab function that computes barycentric form of the interpolating polynomial.
c). Write a Matlab function that computes the divided differences and the Newton form
of the interpolating polynomial.
d). Set n ≥ 1000 or even larger, e.g. n = 10000. Find and click the button “Run and Time” in
Matlab and compare the execution time of the three functions. Report the results and your
observations.
1
MACM 316 COMPUTING ASSIGNMENT #5
3. Observe Runge phenomenon (we have observed a bit in class).
a). Set the interpolation points to be equidistant on [−1, 1], and n = 10, 15, 25, 40. Plot the
four polynomials (with respect to the four values of n) together with the Runge function. Observe the oscillation at the ends of the interval [−1, 1] and report your plots and observations.
b). Set the interpolation points to be Chebyshev on [−1, 1], and n = 10, 15, 25, 40. Plot the four
polynomials (with respect to the four values of n) together with the Runge function. Observe
the results and report your plots and observations.
c). This question is exploratory. Can you find some other distributions of interpolation points
(other than equidistant or Chebyshev), that give rise to a better interpolating polynomial than
the equidistant points or the Chebyshev points? Try a different distribution of interpolation
points, e.g. randomly distributed, and report your observations.
4. Write a Matlab function to compute and plot the natural spline of the Runge function
with n = 5, 10. Report your plot and observations.
2 MACM 316
