Scribd
Upload
1 views

Tutorial 3

real analysis tutorial 3

Uploaded by

Shannamae Yang
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
1 views

Tutorial 3

real analysis tutorial 3

Uploaded by

Shannamae Yang
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Numerical Analysis I.

Tutorial 3

1. TUTORIAL PROBLEMS

1. Let g(x) have a fixed point p that satisfies g ′ (p) = g ′′ (p) = 0. Assume
that the fixed point iteration scheme with the initial value p0 converges to p.
Prove that the convergence is at least of order 3.

2. Consider the Newton method for finding a solution p of the equation


f (x) = 0 by finding a fixed point of a function g. Prove that if f ′ (p) ̸= 0, the
scheme converges (if it does) to p at least quadratically.

3. Consider the iteration scheme for computing c1/3 (with c > 0):
(
pn+1 = αpn + βcp−2 2 −5
n + γc pn , n ≥ 0,
given p0 sufficiently close to c1/3 .

Determine α, β, γ so that the iteration converges to c1/3 with the highest


order of convergence. With this choice of α, β and γ, what is the order of
convergence?

4. Suppose that f (x) is sufficiently smooth on [a, b] and p is the unique


root of f (x) = 0 with multiplicity 2, i.e., f (x) = (x − p)2 h(x) and h(p) ̸= 0.
Assume the initial approximation p0 is close to the root.
(i) Show that the Newton’s method
f (x)
pn+1 = g(pn ), g(x) = x − (1)
f ′ (x)
converges only linearly.
(ii) Show that the following modified Newton’s iteration scheme
f (x)f ′ (x)
pn+1 = g(pn ), where g(x) = x − (2)
[f ′ (x)]2 − f (x)f ′′ (x)
converges at least quadratically (Hint: use µ(x) = f (x)/f ′ (x)).
(iii) Show that the following modified iteration scheme
f (x)
pn+1 = g(pn ), where g(x) = x − 2 (3)
f ′ (x)
converges at least quadratically.
(iv) If f (x) has a unique root in [a, b] with multiplicity m > 2, does the
iteration scheme (2) converge at least quadratically? What about the
scheme (3)?

1
2. PROGRAMMING

Task 1 The following code finds a fixed point of a function g. It will


return the approximated fixed point pn when |pn − pn−1 | < ε. The code will
return an error message if we have not found such pn after N iterations.

def fixedpoint(p0,N,epsilon):
i=1
while i<=N:
p = g(p0)
if abs(p-p0)<epsilon:
fp=p
return fp
else:
p0=p
i=i+1
print("The method failed after",N,"iterations")

Define the function


cos2 (2x) − x2
g(x) = x + .
4 cos(2x) sin(2x) + 2x

The iteration scheme for this function is the Newton’s method to solve

f (x) = cos2 (2x) − x2 = 0.

Run the programme with the following initial values p0: 0.02, 0.021, 0.93,
0.94, 1.13, 1.14, 1.475, 1.476 and note the result. This shows you how the
result depends sensitively on the initial value p0.
Task 2. Modify the code so that we will stop when |f (p)| < ε instead.
You will need to define also the function f . This shows that solving the
equation f (x) = 0 is equivalent to finding fixed points for function g above.

You might also like