MATA14 Analysis 1 (13 hp)

Wednesday May 27th, 2015

Solutions (version 1.0)

Matematikcentrum
Matematik NF

Question 1.
(a) y = 3x − 1.
(b) xy
(c) (x − 3)(x − 5) (The question was not about completing the square!)
(d) x ∈ (−1, 1).
(e) y = 1 where the expression is defined (that is, for x ∈ {(0, π2 ) + 2kπ : k ∈ Z}).

Question 2.
(a) This is a separable differential equation which can be expressed as
y sin x
2
y0 = .
1+y 1 + cos2 x
Writing y 0 = dy/dx, and separating the variables, we obtain
Z Z
y sin x
2
dy = dx.
1+y 1 + cos2 x
Computing the two indefinite integrals, we get
Z
y 1
2
dy = ln(1 + y 2 ) + C,
1+y 2
and Z Z
sin x du
dx = − = − arctan(cos x) + D.
1 + cos2 x 1 + u2
Thus
1 p
ln(1 + y 2 ) = C − arctan(cos x) =⇒ y = ± eC−2 arctan(cos x) − 1
2
p
= ± De−2 arctan(cos x) − 1.

(b) The first order linear differential equation

1 x
y0 − y= 2
x x + 2x + 2
can be solved using the "integrating factor" method. The integrating factor can be obtained
as follows:
−1
Z
1
dx = − ln x =⇒ e− ln x = .
x x
(We can always ignore the undetermined constant when we have an integrating factor, since
we only need a primitive function of −1/x, and then we can choose our favourite constant,
1
2

which is 0.) Multiplying both sides of the differential equation by the integrating factor, we
get:
1 0 1 1 d y 1
y − 2y = 2 =⇒ = 2
x x x + 2x + 1 dx Zx x + 2x + 2
y dx
=⇒ = .
x x2 + 2x + 2
We compute the indefinite integral by completing the square in the denominator and then
making the change of variable u = x + 1:
Z Z
dx dx
=
x2 + 2x + 2 (x + 1)2 + 1
Z
du
= 2
u +1
= arctan u + C = arctan(x + 1) + C.
The final answer is therefore
 
y = x arctan(x + 1) + C .

Question 3.
(a) That r(x) = O(x3 ) can be defined either as |r(x)| ≤ C|x3 | for x in an interval around
x = 0, or as r(x) = b(x)x3 for some function b(x) which is bounded around x = 0.
2
(b) We use the
expansions e(x ) = 1 + x2 + x4 /2 + O(x6 ), arctan x = x − x3 /3 + O x5 and

sin x = x + O x3 . This gives

2
e(x ) − x arctan x − 1 1 + x2 + x4 /2 + O(x6 ) − x2 + x4 /3 + O(x6 ) − 1
lim = lim
x→0 x3 sin x x→0 x4 + O (x6 )
5x4 /6 + O(x6 )
= lim = 5/6.
x→0 x4 + O (x6 )

Question 4.
(a) According to the divergence test, we know that if a series ∞
P
k=1 ak converges, then we
must have ak → 0 when k → ∞. The statement is therefore false. We can also notice that the
sequence of partial sums Sn = a1 + · · · + an converges to 3.2.
(b) We know that sin(1/k) ≈ 1/k for large k. We then approximate the denominator by k.
Altogether we intuitively have
sin k1 1
1
√ ≈ k = 2,
k+ k k k
P∞ 2 to be convergent. We can check this using the limit
thus we expect the series k=1 1/k
comparison test:
sin k1
 
√
k+ k k 2 sin k1
lim   = lim √
k→∞ 1 k→∞ k + k
k 2

k 2 k1 + O(1/k 2 )


= lim √
k→∞ k+ k
 3

1 + O(1/k) 1+0
= lim √ = = 1.
k→∞ 1 + 1/ k 1+0
(c) Since we have both factorials and powers, we use the quotient test:
(k+1)!

|ak+1 | (k+1)20 20k+1
lim = lim k!
k→∞ |ak |


k→∞ 20 kk 20
(k + 1)! k 20 20k
= lim
k→∞ k! (k + 1)20 20k+1
 k 20 1
= lim (k + 1) · ·
k→∞ k+1 20
 
1
= ∞·1· = ∞.
20
Thus the series is divergent.

Question 5.
(a) Suppose that f is continuous on [a, b] and differentiable on (a, b). Then there exists
c ∈ (a, b) such that
f (b) − f (a)
f 0 (c) = .
b−a
(b) Using the Mean value theorem, there exists c between x1 and x2 such that
f (x2 ) − f (x1 ) = f 0 (c)(x2 − x1 ).
Since we have f 0 (c) ≥ 0 and (x2 − x1 ) ≥ 0, we get that the right hand side is positive, which
implies that f (x2 ) − f (x1 ) ≥ 0 or, equivalently, f (x2 ) ≥ f (x1 ).
(c) We let
x−1
f (x) = ln x − .
x+1
The derivative of this function is
1 (x + 1) − (x − 1) 1 2
f 0 (x) = − 2
= −
x (x + 1) x (x + 1)2
(x + 1)2 − 2x
=
x(x + 1)2
x2 + 1
= .
(x + 1)2
Hence the derivative is positive when x ≥ 1, and the function is strictly increasing. Since
f (1) = 0, we get the desired conclusion.
Question 6.
(a) The definition can be written as
∀ > 0, ∃ δ > 0 such that 0 < |x − a| < δ =⇒ |f (x) − L| < .
To show that this holds for the function f (x) = x2 +3x+1, L = 5 and a = 1, we let a challenge
 > 0 be given. We do the following:
|x2 + 3x + 1 − 5| = |x2 + 3x − 4| = |(x + 4)(x − 1)|.
4

We want to get rid of the factor x + 4, and therefore we set |x − 1| < 1. This gives x ∈ (0, 2),
which implies that |x + 4| < 6. Hence we have
|(x + 4)(x − 1)| < 6|x − 1|.
The right hand side is less than  if |x − 1| < /6. Summing up, we see that if |x − 1| is less
than both 1 and /6, then
|x2 + 3x + 1 − 5| = |(x + 4)(x − 1)| < 6|x − 1| < .
Thus we choose δ = min{1, /6}.
(b) The definition can be written as
∀ω > 0, ∃N > 0 such that n ≥ N =⇒ an > ω.
To show that this definition holds for an = n!/20n , we take a fixed, but unknown, challenge
ω > 0. By definition, we have
n! 1 2 3 n−1 n
an = n = · · ··· · .
20 20 20 20 20 20
We see that from n = 20, an becomes larger and larger. More precisely, if n ≥ 20 we have
21 22 n−1 n
an = a20 · · ··· ·
20 20 20 20
n a20
≥ a20 · 1 · 1 · · · 1 · = · n.
20 20
We now examine when the right hand side is larger than ω:
a20 20ω
· n > ω ⇐⇒ n > .
20 a20
20ω
To sum up, we can say that if n ≥ 20 and n > a20 , then
21 22 n−1 n
·
an = a20 · ··· · > ω.
20 20 20 20
We therefore choose N = max{20, b20ω/a20 c + 1}.
Question 7.
(a) Taylor’s formula with Lagrange error term around x = 0 says that if f, f 0 , f 00 , . . . , f (n+1)
exist and are continuous, then
f 00 (0) 2 f (n) (0) n
f (x) = f (0) + f 0 (0)x + x + ··· + x + R(x),
2! n!
where
f (n+1) (c) n+1
R(x) = x
(n + 1)!
for some c between 0 and x.
(It is also possible to say that f is (n + 1)-times differentiable around x = 0. The continuity
of the (n + 1)’th derivative is not actually needed here.)
(b) We know that
u2
cos(u) = 1 − + r(u),
2
where
cos(c)u4
r(u) = .
4!
 5

This implies that

|u|4
|r(u)| ≤ .
4!
Integrating, we get that
x x x
t4 
Z Z Z
cos(t2 )dt = 1− dt + r(t2 )dt
0 0 2 0

x5
=x− + R(x),
10
where Z x
R(x) = r(t2 )dt.
0
For x > 0, the triangle inequality for integrals gives
Z x Z x 2 4
1 x 8 x9 x9
Z
2 (t )
|R(x)| ≤ |r(t )|dt ≤ dt = t dt = = .
0 0 4! 4! 0 4! · 9 216
Rx R0
(Notice that if x < 0, one can write 0 = − x before using the triangle inequality. This means
that the right hand side is −x9 /4! · 9 for x < 0.)
(c) We just hope that what we did in (b) is enough. Our estimate is then
Z 0.1
1 1 105 − 1 99999
cos(t2 )dt ≈ − 5 = 6
= = 0.099999,
0 10 10 · 10 10 1000000
and this is with an error less than
1 1
|R(0.1)| ≤ 9
≤ = 0.000000000005.
216 · 10 2 · 1011