Math 1005 H Fall 2015

Tutorial 5
1. Determine whether each of the given series converges or diverges. If it converges,
find the sum.
∞
X
(a) 2n (−3)−2n
n=2
Solution: We will use the fact that a geometric series ∞ n
P
n=0 r converges
1
to 1−r when |r| < 1 and diverges when |r| ≥ 1. We have
∞
X ∞
X
−2n
n
2 (−3) = 2n ((−3)2 )−n
n=2 n=2
∞  n
X 2
= .
n=2
9
 
Since  29  < 1, this geometric series converges. The series starts at n = 2.
We must rewrite it to start at n = 0. We give two ways to do this. We
could add and subtract the first two terms:
∞  n ∞  n  0  1
X 2 X 2 2 2
= − −
n=2
9 n=0
9 9 9
1 2
= 2 −1−
1− 9 9
9 2 4
= −1− = .
7 9 63
Alternatively, we could reindex the series:
∞  n ∞  n+2
X 2 X 2
=
n=2
9 n=0
9
∞  n
 2 X
2 2
=
9 9
  n=0
 
4 1
=
81 1− 2
   9
4 9 4
= = .
81 7 63
∞
X 1
(b)
n=1
4n2 −1
Solution: We could try to use the nth term test, but since the terms we
are summing approach 0, it tells us nothing. Using partial fractions, we
 have
1 1
=
4n2 − 1 (2n − 1)(2n + 1)
 
1 1 1
= −
2 2n − 1 2n + 1

(details skipped, you can choose to show more if you want). Thus the nth
partial sum is
"     
1 1 1 1 1 1
sn = 1− + − + − + ...
2 3 3 5 5 7
   #
1 1 1 1
... + − + −
2n − 3 2n − 1 2n − 1 2n + 1
 
1 1
= 1− .
2 2n + 1

This is a telescoping series. Clearly sn → 12 and n → ∞, and so the series

sums to 12 .
Alternatively, we could have used the integral test to see that the series
converges, but it would also require the use of partial fractions, and it
wouldn’t give us the sum of the series.
∞ √
X 2n2 + 1
(c) √
n=1
2n2 − 1
Solution: Dividing numerator and denominator by n, we have
√
q
2n2 + 1 2 + n12
lim √ = lim q
n→∞ 2n2 − 1 n=→∞ 2 − 12
n
√
2
=√
2
= 1.

Since this limit is not 0, the series diverges by the nth term test.

2. Determine whether each of the given series converges or diverges.

∞
X 1
(a) √ √
n=1
n3n
Solution: Recall that the p-series ∞ 1
P
n=1 np converges for p > 1 and diverges
for p ≤ 1. We have
1 1 1
√ √ = 1 1 = 5.
n n3
n2 n3 n6
Thus the series is a p-series with p = 56 . Since 5
6
≤ 1, the series diverges.
 ∞
X 2n + 1
(b)
n=1
3n + 1
Solution: We could try to use the nth term test, but since the terms we
are summing approach 0, it tells us nothing. Note that for n ≥ 1 we have
2n + 1 2n + 1
≤ (we made the denominator smaller)
3n + 1 3n
2n + 2n
≤ (we made the numerator bigger)
3n
2 · 2n
= n
3 
 n
2
=2 .
3
∞  n ∞  n
X 2 X 2
Since 2 = 2 converges (it is a geometric series with
n=0
3 n=0
3
∞
2
X 2n + 1
|r| = 3 < 1), by the comparison test the series must converge.
n=1
3n + 1
∞
X 1
(c) √
n=2
n ln(n)
Solution: We could try to use the nth term test, but since the terms we
1
are summing approach 0, it tells us nothing. The function f (x) = √x ln(x)
is very hard to integrate, so the integral test isn’t working out either. Let’s
1
try the limit comparison test with an = √n ln(n) and bn = n1 (in practice,
one might have to try several sequences bn before finding one that yields
a useful result). Then
an n
lim = lim √
n→∞ bn n→∞ n ln(n)
√
n
= lim
n→∞ ln(n)
1
√
2 n
= lim (by L’Hôpital’s Rule)
n→∞ 1
n
√
n
= lim
n→∞ 2
= ∞.
Since the limit is ∞ and n=1 n1 diverges, the series ∞
P∞
√ 1
P
n=2 n ln(n) diverges
by the limit comparison test.
Note that as an alternate solution we could have used the comparison test
with the inequality
1 1
≤√
n ln(n) n ln(n)
P∞ 1
for n ≥ 2 and the fact that n=2 n ln(n) diverges (as seen in class, or by
using the integral test).
 ∞
X n
3. Show that the series n2
converges and find how large a value of n is needed
n=1
e
so that sn approximates the sum with error no greater than 0.0001.
2
Solution: Let f (x) = exx2 = xe−x . Then f (x) is positive and continuous for
x ≥ 1. Since
2 2 2
f 0 (x) = e−x − 2x2 e−x = e−x (1 − 2x2 ) < 0
for x > √1 ,
f (x) is decreasing for x ≥ 1. We have
2
Z ∞
1 ∞ 1
Z
x
x 2 dx = u
du (letting u = x2 )
1 e 2 1 e
∞
1 
= − u
2e 1
1
= .
2e
Since the integral converges,
R∞ by the integral test the series must also converge.
We know that Rn ≤ n f (x)dx, so in order to guarantee that the error is less
than 0.0001, we need
Z ∞ ∞
1 ∞ 1
Z
x 1  1
Rn ≤ x 2 dx = u
du = − u  = n2 ≤ 0.0001.
n e 2 n2 e 2e n2 2e
Taking reciprocals, we need
2
2en ≥ 10000
2
en ≥ 5000
n2 ≥ ln(5000)
p
n ≥ ln(5000) ≈ 2.918...
Thus we are guaranteed an error of less that 0.0001 if we approximate the sum
by s3 = 1e + e24 + e39 = 0.40488.... This series converges quickly, as only 3 terms
are needed to achieve a small error.

Additional examples if you somehow have extra time

∞
X
4. Determine whether the series 2 · 3−2n 53n+1 converges or diverges. If it con-
n=0
verges, find the sum.
Solution: We have
∞
X ∞
X
−2n 3n+1
2·3 5 = 2 · 5(32 )−n (53 )n
n=0 n=0
∞
X 125n
= 10
n=0
9n
∞  n
X 125
= 10 .
n=0
9
  
This is a geometric series. Since  125
9
 ≥ 1, this series diverges.
∞
X 1
5. Determine whether the series converges or diverges.
n=3
n ln(n) ln(ln(n))
Solution: We could try to use the nth term test, but since the terms we are
summing approach 0, it tells us nothing. Let
1
f (x) = .
x ln(x) ln(ln(x))

For x ≥ 3, f is positive, continuous, and decreasing (this is a little harder to

see, but we can use the fact that the denominator is increasing for x ≥ 3). We
can use the integral test. We must evaluate
Z ∞
dx
.
3 x ln(x) ln(ln(x))
1
We let u = ln(ln(x)) so that du = x ln(x)
. Thus
Z ∞ Z ∞
dx 1
= du
3 x ln(x) ln(ln(x)) ln(ln(3)) u
= ln |u||∞ln(ln(3))

= ∞.

Thus, by the integral test, the series diverges.