Analysis I.

HW 8

Andrzej Novak

Question 5.3.4: 13. Let fn denote the n-th iterate of f, f1 = f, f2 (x) = f (f1 (x)), ...fn (x) =
f (fnl (x)). Express fn0 in terms of f 0 . Show that if a |f 0 (x)| b for all x, then
an |fn0 (x)| < bn .
Proof. By Chain rule:
fn0 (x) = f 0 (fn1 (x))f 0 (fn2 (x))...f 0 (x) = f 0 (x)

n1
Y

f 0 (fi (x))

i=1

|fn0 (x)| = |f 0 (x)|

a

n1
Y

|f 0 (fi (x))|

i=1
0
|fn (x)| bn

Question 5.4.6: 1. Suppose f is a C 2 function on an interval (a, b) and the graph of f

lies above every secant line. Prove that f 00 (x) 0 on the interval.
Proof. We have for some x, y, z (a, b), x < y < z:
f (y)

f (z) f (x)
(y x) + f (x)
zx

f (y) f (x)
f (z) f (x)

yx
zx
Take x = y h, z = y + h:
f (y) f (y h)
f (y + h) f (y h)

h
2h
f (y + h) + f (y h) 2f (y)
2h2
We take h 0 and by the LHospital rule, we know that f 00 (y) 0.
0

Question 5.4.6: 3. If f is C 2 on an interval prove that:

lim

h0

f (x + h) 2f (x) + f (x h)
= f 00 (x)
h2

The expression f (x + h) 2f (x) + f (x h) is called the symmetric second difference.

Proof. If we assume f to be C 2 we have f (x + h) = f (x) + hf 0 (x) + 1/2h2 f 00 (x) + o(h2 )
and f (x h) = f (x) hf 0 (x) + 1/2h2 f 00 (x) + o(h2 ). Therefore:
h2 f 00 (x) + o(h2 )
f (x + h) 2f (x) + f (x h)
=
lim
h0
h0
h2
h2
lim

o(h2 )
= f 00 (x)
h0 h2

= f 00 (x) + lim

Andrzej Novak

Homework 8

Question 5.4.6: 9. Suppose f is a C n function on an interval and Tn (x0 , x) is the same

function of x for all x0 in the interval. What can you say about f ?
Proof. f is constant on the interval.
n
Question 5.4.6: 13. Suppose P
f is a C n function in a neighbourhood of xP
0 and g is C in
n
n
k
a neighbourhood of f (x0 ). Let k=0 ak (xx0 ) be Tn (f, x0 , x), and let j=0 bj (y y0 )j
be Tn (g, y0 , 0) where y0 = f (x0 ). Show that Tn (g f, x0 , x) is obtained from:
X
j
n
n
X
k
b0 +
bj
ak (x x0 )
j=1

k=1

by retaining only the powers (x x0 ) up to n.

Proof.

j
b
X
2
n
Tn = b0 +
bj a0 + a1 (x x0 ) + a2 (x x0 ) ... + an (x x0 )
j=1



 2
 n
2
n
Tn = b0 + b1 a0 + a1 (x x0 ) + a2 (x x0 ) ...an (x x0 ) + b2 .... ... + ...

Tn = b0 + b1 a0 + ... + an (x x0 )

Question 5.4.6: 14. If f (x) =

1
1+x

1


+ b2 ...(x x0 )

n/2

2

show that Tn (f, 0, x) =


n
... + a0 + a1 (x x0 )

Pn

k k
k=0 (1) x .

Proof.
f (x) = (1 + x)1 f 0 (x) = (1)(1 + x)2 , f 00 (x) = 2(1 + x)3
f (n) = n!(1)n (1 + x)1n f (n) (0) = n!(1)n
P
Therefore the zeroth order Taylors polynomial is Tn (f, 0, x) = nk=0 (1)k xk .
Question 5.4.6: 23 (a,b,c).
1. If f is C n on an interval and has n + 1 distinct zeros, prove that f (n) has at least
one zero on the interval.
2. If f is C n on an interval and f (n) never vanishes, then f has at most n zeros on
the interval.
3. A polynomial of degree n has at most n real zero.
Proof.
1. We have from Rolles theorem that between two zeros of f will be at least one zero
of f 0 . If we assume f has n + 1 distinct zeros, than f 0 has n distinct zeros and f (n)
has at least one zero.

Andrzej Novak

Homework 8

2. Assuming to the contrary: from above we know that if f has n + 1 zeros f (n)
vanishes at least once, which contradicts the condition of non-vanishing. Therefore
we have that f with at most n zeros.
3. A polynomial has the form of f (x) = an xn + an1 xn1 + ... + a1 x + a0 . So we have
f n (x) = an n! 6= 0. Therefore a polynomial never vanishes and by (2.) it has at
most n zeros.
Question 5.4.6: 24. If f is C 2 , prove that f cannot have a local maximum or minimum
at an inflection point (note that an inflection point is defined as a point where f 00 changes
sign; it is not enough that f 00 vanish at the point).
Proof. If f is C 2 , we have f 0 , f 00 exist and are continuous. An inflection point of f ,
at f 00 (x0 ) = 0 is also an local extremum of f 0 . Now lets assume to the contrary that
f (x0 ) is also the extremum of f . If it is a maximum we have that f 0 (x ) > 0 and
f 0 (x + ) < 0 and by theorem 5.4.2 f 00 < 0. If it is a minimum we have f 0 (x ) < 0 and
f 0 (x + ) > 0 and by theorem 5.4.2 f 00 > 0. In either case we arrive at contradiction and
therefore a C 2 function cannot have a local max or min at an inflection point.