MAT320/640, Fall 2020 Renato Ghini Bettiol

Solutions to Homework Set 3

1. Decide if each of the statements below is true or false. If it is true, give a complete
proof; if it is false, give an explicit counter-example.
(a) If {xn } is a convergent sequence of real (or complex) numbers, then {|xn |} is also
convergent.
(b) If {|xn |} is a convergent sequence of real (or complex) numbers, then {xn } is also
convergent.
(c) If {xn } is a sequence of real (or complex) numbers that converges to 0, and {yn }
is a sequence of real numbers that diverges to +∞, then the sequence {xn yn }
converges to 1.
(d) If {xn } is a sequence of real numbers that diverges to +∞ and a ∈ R, then
p p
lim log (xn + a) − log xn = 0
n→+∞

Solution:
(a) TRUE: First, for all a, b ∈ C, using the triangle inequality, we have that:
|a| = |a + (b − b)| = |(a − b) + b| ≤ |a − b| + |b| =⇒ |a| − |b| ≤ |a − b|
and, similarly,
|b| = |b + (a − a)| = | − (a − b) + a| ≤ |a − b| + |a| =⇒ |b| − |a| ≤ |a − b|
so, altogether, we have ±(|a| − |b|) ≤ |a − b|, that is, |a| − |b| ≤ |a − b| for all a, b ∈ C.
Now, suppose {xn } converges to x, i.e., for all ε > 0 there exists N ∈ N such that
|xn − x| < ε. Applying the inequality |a| − |b| ≤ |a − b| proved above, we have:
|xn | − |x| ≤ |xn − x| < ε,
for all n ≥ N , that is, {|xn |} converges to |x|.
(b) FALSE: Take xn = (−1)n , so that |xn | = 1. Then {|xn |} converges to 1, but xn
does not converge.
(c) FALSE: Take xn = n2 and yn = n. Then {xn } converges to 0, {yn } diverges to
+∞, but {xn yn } does not converge to 1.
√ √ √ √
(d) TRUE: Using the fact that ( A − B)( A + B) = A − B, we have:
 
xn + a
log
p p log (xn + a) − log xn xn
log (xn + a) − log xn = p √ =p √
log (xn + a) + log xn log (xn + a) + log xn
    
xn + a xn + a
Since {xn } diverges to +∞, we have that converges to 1, so log
xn n o xn
p √
converges to 0. Moreover, the denominators log (xn + a) + log xn diverge to
np √ o
+∞. Thus, the sequence log (xn + a) − log xn converge to 0.
2. Suppose {xn } is a Cauchy sequence in a metric space (X, d), with a subsequence {xnk }
that converges to x ∈ X, i.e., x is a subsequential limit of {xn }. Prove that {xn }
converges to x.
Solution:
Since the subsequence {xnk } converges to x ∈ X, we know tht for all ε > 0, there exists
N1 ∈ N such that if nk ≥ N1 then d(xnk , x) < 2ε . Moreover, since {xn } is Cauchy, for
all ε > 0, there exists N2 ∈ N such that if n, m ≥ N2 , then d(xn , xm ) < 2ε . Choose
` ∈ N such that n` > max{N1 , N2 }. Then, if n ≥ max{N1 , N2 }, we have that
ε ε
d(xn , x) ≤ d(xn , xn` ) + d(xn` , x) < + = ε,
2 2
that is, xn converges to x.

1
3. Given a > 0, define a sequence {xn } of real numbers inductively by setting x1 = ,
a
1
and xn+1 = , i.e.,
a + xn
1
xn = .
1
a+
1
a+
a + ...

(a) Is {xn } monotonic?

1
(b) Prove that {xn } converges to the unique real number L such that L = , i.e.,
a+L
the positive root of the equation x2 + ax − 1 = 0.

Side note: Setting a =

√
1 in the above, the limit of the corresponding sequence {xn } is
1 1+ 5 ∼
L = ϕ , where ϕ = 2 = 1.618 . . . is the so-called golden ratio.
Solution:
(a) No. The sequence {xn } is not monotonic. In fact, for all n ∈ N, we have that:

x2 < x4 < · · · < x2n < · · · < L < · · · < x2n−1 < · · · < x3 < x1 ,

i.e., the subsequence {x2n } is monotonically increasing, the subsequence {x2n−1 } is

monotonically decreasing, and x2n < L < x2n−1 for all n ∈ N.
Proof. Let us first prove that the subsequence {x2n−1 } is monotonically decreasing, by
induction on n. The base case n = 1, i.e., x3 < x1 follows from:
1 1
x3 = 1 < = x1 .
a+ a+0
a+ a1

2
Now, assume by induction that x2n−1 < x2n−3 . Then
1 1 1 1 1
x2n+1 = = =⇒ =a+ >a+
a + x2n a + a+x12n−1 x2n+1 a + x2n−1 a + x2n−3
1
=⇒ x2n+1 < 1 = x2n−1 ,
a+ a+x2n−3

that is, x2n+1 < x2n−1 , which is the next case. This establishes the claim for all n ∈ N.
Similarly, we prove by induction that the subsequence {x2n } is monotonically increasing.
The base case n = 1, i.e., x2 < x4 follows from:
1 1 1 1 1
=a+ =a+ >a+ 1 = .
x2 a a+0 a + a+ 1 x 4
a

Now, assume by induction that x2n−2 < x2n . Then

1 1 1 1 1
x2n+2 = = 1 =⇒ =a+ <a+
a + x2n+1 a + a+x x2n+2 a + x2n a + x2n−2
2n
1
=⇒ x2n+2 > 1 = x2n ,
a+ a+x2n−2

that is, x2n < x2n+2 , which is the next case. This establishes the claim for all n ∈ N.
Finally, let us show that x2n < L < x2n−1 for all n ∈ N, also by induction. The base
case follows from:
1 1
L= < = x1
a+L a
and, using the above inequality,
1 1
x2 = < = L.
a + x1 a+L
Now, assume by induction that x2n < L < x2n−1 . Then
1 1 1 1
=a+ <a+ =a+L =⇒ L= < x2n+1
x2n+1 a + x2n−1 a+L a+L
and, similarly,
1 1 1 1
=a+ >a+ =a+L =⇒ x2n+2 < = L,
x2n+2 a + x2n a+L a+L
that is, x2n+2 < L < x2n+1 , which is the next case. This concludes the proof.
(b) By what we showed in (a), since {x2n } is monotonically increasing and bounded
from above by L, it follows that {x2n } is convergent, say x2n → L, with L ≤ L.
Similarly, since {x2n−1 } is monotonically decreasing and bounded from below by L, it
is convergent, say x2n−1 → L, with L ≥ L. We claim that L = L = L.

3
Letting n → +∞ in the equations
1 1
x2n+1 = 1 , and x2n+2 = 1 ,
a+ a+x2n−1 a+ a+x2n

we find that both L and L are solutions of the equation

1
z= .
a + a+1 1
z

The above equation is equivalent to z 2 + az − 1 = 0, which has a unique positive real

solution z = L. Therefore, L = L = L, and hence {xn } also converges to L because
{xn : n ∈ N} = {x2n : n ∈ N} ∪ {x2n−1 : n ∈ N} and L is its unique limit point.