Download as pdf or txt
Download as pdf or txt
You are on page 1of 61

MATH 1023 - Honors Calculus I (Solution to Exercises)

1.1.1 (a) If limn→∞ xn = l, then limn→∞ xn − l = limn→∞ xn − limn→∞ l = l − l = 0.


(b) If limn→∞ (xn − l) = 0, then limn→∞ xn = limn→∞ ((xn − l) + l) = limn→∞ (xn − l) + limn→∞ l =
0 + l = l.
So we may conclude that limn→∞ xn = l if and only if limn→∞ (xn − l) = 0.
1.1.2 Given that limn→∞ xn exists and limn→∞ yn exists, we may assume that limn→∞ xn = a and limn→∞ yn =
b. Also we have that limn→∞ xn yn = limn→∞ xn · limn→∞ yn = ab.
So if limn→∞ xn yn = 0, then we must have ab = 0.
If neither a nor b is 0, then we must have ab 6= 0. So if ab = 0, then a = 0 or b = 0, that is limn→∞ xn = 0
or limn→∞ yn = 0.
If we don’t know the convergence of xn and yn , we may have {xn } = {1, 0, 1, 0, . . .} and {yn } =
{0, 1, 0, 1, . . .}. In this case, {xn yn } = {0, 0, 0, 0, . . .}, so limn→∞ xn yn = 0 while neither limn→∞ xn nor
limn→∞ yn exists.
1.1.3 1
2 2 1
n+2 1+ n limn→∞ 1 + limn→∞ n limn→∞ 1 + 2 limn→∞ n 1+2·0
lim = lim 3 = 3 = 1 = =1
n→∞ n − 3 n→∞ 1 − limn→∞ 1 − limn→∞ limn→∞ 1 − 3 limn→∞ 1−3·0
n n n

2
1
n+2 + 22 limn→∞ n1 + 2 limn→∞ n12 0+2·0
lim 2
= lim n n3 = 1 = 1−3·0 =0
n→∞ n − 3 n→∞ 1 − 2 limn→∞ 1 − 3 limn→∞ n2
n

3
3 2
2n2 − 3n + 2 2− n + n2 limn→∞ 2 − 3 limn→∞ n1 + 2 limn→∞ n12 2−3·0+2·0 2
lim = lim 4 1 = 1 1 = =
n→∞ 3n2 − 4n + 1 n→∞ 3 − + limn→∞ 3 − 4 limn→∞ n + limn→∞ n2 3−4·0+0 3
n n2

4
n3 + 4n2 − 2 1 + n4 − n23 limn→∞ 1 + 4 limn→∞ n1 − 2 limn→∞ n13 1+4·0−2·0 1
lim 3
= lim 1 3 = 1 1 = 2−0+3·0 = 2
n→∞ 2n − n + 3 n→∞ 2 − 2 + 3 limn→∞ 2 − lim n→∞ n2 + 3 limn→∞ n3
n n

5
(n + 1)(n + 2) (1 + n1 )(1 + n2 ) (limn→∞ 1 + limn→∞ n1 )(limn→∞ 1 + 2 limn→∞ n1 )
lim = lim =
n→∞ 2n2 − 1 n→∞ 2 − n12 limn→∞ 2 − limn→∞ n12
(1 + 0)(1 + 2 · 0) 1
= =
2−0 2
6
2n2 − 1 2 − n12 limn→∞ 2 − limn→∞ n12
lim = lim =
n→∞ (n + 1)(n + 2) n→∞ (1 + 1 )(1 + 2 ) (limn→∞ 1 + limn→∞ n1 )(limn→∞ 1 + 2 limn→∞ n1 )
n n
2−0
= =2
(1 + 0)(1 + 2 · 0)

7
(n2 + 1)(n + 2) (1 + n12 )(1 + n2 ) (limn→∞ 1 + limn→∞ n12 )(limn→∞ 1 + 2 limn→∞ n1 )
lim = lim =
n→∞ (n + 1)(n2 + 2) n→∞ (1 + 1 )(1 + 22 ) (limn→∞ 1 + limn→∞ n1 )(limn→∞ 1 + 2 limn→∞ n12 )
n n
(1 + 0)(1 + 2 · 0)
= =1
(1 + 0)(1 + 2 · 0)

1
8
(2 − n)3 ( n2 − 1)3 (2 limn→∞ n1 − limn→∞ 1)3
lim = lim =
n→∞ 2n3 + 3n − 1 n→∞ 2 + 32 − 13 limn→∞ 2 + 3 limn→∞ n12 − limn→∞ 1
n n n3
(2 · 0 − 1)3 1
= =−
2+3·0−0 2
9
3 3 1 3
(n2 + 3)3 (1 + n2 ) (limn→∞ 1 + 3 limn→∞ n2 ) (1 + 3 · 0)3
lim = lim 2 2 = 1 2 = =1
n→∞ (n3 − 2)2 n→∞ (1 − (limn→∞ 1 − 2 limn→∞ (1 − 2 · 0)2
n3 ) n3 )

1.1.4 1
√ 1+ √2 limn→∞ 1 + limn→∞ √2 limn→∞ 1 + 2 limn→∞ √1
n+2 n n n 1+2·0
lim √ = lim = = = =1
n→∞ n − 3 n→∞ 1 − √3 limn→∞ 1 − limn→∞ √3n limn→∞ 1 − 3 limn→∞ √1n 1−3·0
n

2
√ √1 + 2 limn→∞ √1n + 2 limn→∞ n1
n+2 n n 0+2·0
lim = lim = 1 = 1−3·0 =0
n→∞ n−3 n→∞ 1 − 3 limn→∞ 1 − 3 limn→∞ n
n

3
√ √2 −3+ 2
2 limn→∞ √1 − limn→∞ 3 + 2 limn→∞ 1
2 n − 3n + 2 n n n n 2·0−3+2·0 3
lim √ = lim = = =
n→∞ 3 n − 4n + 1 n→∞ √3 −4+ 1 3
limn→∞ n
√ − limn→∞ 4 + limn→∞ n1 3·0−4+0 4
n n

4
√ √ 1
+ √4n − n2 1
limn→∞ n2/3 + 4 limn→∞ √1n − 2 limn→∞ n1
3
n+4 n−2 n2/3
lim √ = lim 2 =
n→∞ 23n−n+3 n→∞
n2/3
− 1 + n3 1
2 limn→∞ n2/3 − limn→∞ 1 + 3 limn→∞ n1
0+4·0−2·0
= =0
2·0−1+3·0
5
√ √ (1 + √1n )(1 + √2n ) (limn→∞ 1 + limn→∞ √1n )(limn→∞ 1 + 2 limn→∞ √1 )
( n + 1)( n + 2) n
lim = lim =
n→∞ 2n − 1 n→∞ 2 − n1 limn→∞ 2 − limn→∞ n1
(1 + 0)(1 + 2 · 0) 1
= =
2−0 2
6
2n − 1 2 − n1 limn→∞ 2 − limn→∞ n1
lim √ √ = lim =
n→∞ ( n + 1)( n + 2) n→∞ (1 + √1 )(1 + √2 ) (limn→∞ 1 + limn→∞ √1n )(limn→∞ 1 + 2 limn→∞ √1 )
n n n
2−0
= =2
(1 + 0)(1 + 2 · 0)

7
√ (1 + √1n )(1 + n2 ) (limn→∞ 1 + limn→∞ √1n )(limn→∞ 1 + 2 limn→∞ n1 )
( n + 1)(n + 2)
lim √ = lim =
n→∞ (n + 1)( n + 2) n→∞ (1 + 1 )(1 + √2 ) (limn→∞ 1 + limn→∞ n1 )(limn→∞ 1 + 2 limn→∞ √1n )
n n
(1 + 0)(1 + 2 · 0)
= =1
(1 + 0)(1 + 2 · 0)

2
8
√ 2 3 1 3
(2 − 3 n)3 (√ 3 n − 1) (2 limn→∞ √3 n − limn→∞ 1)
lim √ = lim 2 =
n→∞ 2 3 n + 3n − 1 n→∞ 2/3
n
+ 3 − n1 2 limn→∞ n2/31
+ limn→∞ 3 − limn→∞ 1
n
(2 · 0 − 1)3 1
= =−
2·0+3−0 3
9
√ (1 + 3 3 1 3
3 n) (limn→∞ 1 + 3 limn→∞ 3 n)
√ √
( 3 n + 3)3 (1 + 3 · 0)3
lim √ = lim = = =1
n→∞ ( n − 2) 2 n→∞ (1 − √2 )2 (limn→∞ 1 − 2 limn→∞ √1n )2 (1 − 2 · 0)2
n

1.1.5 1
a a
n+a 1+ n limn→∞ 1 + limn→∞ n limn→∞ 1 + a limn→∞ n1 1+a·0
lim = lim b
= = 1 = 1=b·0 =1
n→∞ n + b n→∞ 1 +
n limn→∞ 1 + limn→∞ nb limn→∞ 1 + b limn→∞ n

2
√ √1 + a limn→∞ √1n + a limn→∞ n1
n+a n n 0+a·0
lim = lim = 1 = =0
n→∞ n+b n→∞ 1 + b limn→∞ 1 + b limn→∞ n 1+b·0
n

3
1 a
n+a n + n2 limn→∞ n1 + a limn→∞ n12 0+a·0
lim = lim = = =0
n→∞ n2 b c
+ bn + c n→∞ 1 + n + n2 limn→∞ 1 + b limn→∞ n1 + c limn→∞ 1
n2
1+b·0+c·0

4
√ √1 + a limn→∞ √1 + a limn→∞ 1
n+a n n n n 0+a·0
lim √ = lim = = =0
n→∞ n + b n + c n→∞ 1 + √b +
n
c
n
limn→∞ 1 + b limn→∞ √1n + c limn→∞ n1 1+b·0+c·0

5
√ √ (1 + √an )(1 + √bn ) (limn→∞ 1 + a limn→∞ √1n )(limn→∞ 1 + b limn→∞ √1 )
( n + a)( n + b) n
lim = lim =
n→∞ cn + d n→∞ c + nd limn→∞ c + d limn→∞ n1
(1 + 0)(1 + 2 · 0) 1
= =
2−0 2
6
cn + d c + nd limn→∞ c + d limn→∞ n1
lim √ √ = lim =
n→∞ ( n + a)( n + b) n→∞ (1 + √a )(1 +
n
√b )
n
(limn→∞ 1 + a limn→∞ √1n )(limn→∞ 1 + b limn→∞ √1 )
n
2−0
= =2
(1 + 0)(1 + 2 · 0)

7
an3 + b a + nb3 limn→∞ a + b limn→∞ n13 a+b·0 a
lim √ = lim = = = 6
n→∞ (c n + d)6 n→∞ (c + √d )6 (lim n→∞ c + d limn→∞ √1 )6 (c + d · 0) 6 c
n n

8
√ a b 1 1
(a 3 n + b)2 ( n5/12 + n3/4 )2 (a limn→∞ n5/12 + b limn→∞ n1/3 )2 (a · 0 + b · 0)2
lim √ = lim d
= 1 = =0
n→∞ (c n + d)3 n→∞ (c + n1/2 )3 (limn→∞ c + d limn→∞ n1/2 )3 (c + d · 0)3

3
9
√ b 1
(a 3 n + b)2 (a + n1/3 )2 (limn→∞ a + b limn→∞ n1/3
)2 (a + b · 0)2 a2
lim √ 3
= lim d
= 1 = 3
= 3
n→∞ (c n + d) n→∞ (c +
n1/2
)3 (limn→∞ c + d limn→∞ n1/2
)3 (c + d · 0) c

1.1.6 • If p < q, then q − p > 0 and q − p + k > 0 for any positive integer k. So
ap np + ap−1 np−1 + · · · + a1 n + a0
lim
n→∞ bq nq + bq−1 nq−1 + · · · + b1 n + b0
ap ap−1 a1 a0
nq−p + nq−p+1 + · · · + nq−1 + nq
= lim b b1 b0
n→∞ bq + q−1
n + · · · + nq−1 + nq
1 1 1 1
ap limn→∞ nq−p + ap−1 limn→∞ nq−p+1 + · · · + a1 limn→∞ nq−1 + a0 limn→∞ nq
= 1
limn→∞ bq + bq−1 limn→∞ n + · · · + b1 limn→∞ nq−1 + b0 limn→∞ n1q
1

ap · 0 + ap−1 · 0 + · · · + a1 · 0 + a0 · 0
=
bq + bq−1 · 0 + · · · + b1 · 0 + b0 · 0
=0

• If p = q, then
ap np + ap−1 np−1 + · · · + a1 n + a0
lim
n→∞ bq nq + bq−1 nq−1 + · · · + b1 n + b0
a a1 a0
ap + q−1n + · · · + nq−1 + nq
= lim
n→∞ b + bq−1 + · · · + b1 + b0
q n nq−1 nq
ap + ap−1 limn→∞ n1 + · · · + a1 limn→∞ nq−1
1
+ a0 limn→∞ n1q
=
limn→∞ bq + bq−1 limn→∞ n1 + · · · + b1 limn→∞ nq−1
1
+ b0 limn→∞ 1
nq
ap + ap−1 · 0 + · · · + a1 · 0 + a0 · 0
=
bq + bq−1 · 0 + · · · + b1 · 0 + b0 · 0
ap
=
bq

1.1.7 1 Using the result from Ex 1.1.6, because 1 < 2, so


1010 n
lim =0
n→∞ n2 − 10

2
55 (2n + 1)2 − 1010 55 · 4n2 + 55 · 4n + 55 − 1010 55 · 4
lim = lim = = 2 · 54 = 1250
n→∞ 10n2 − 5 n→∞ 10n2 − 5 10
Because both the highest power on the numerator and the denominator are 2, then according to
Ex 1.1.6, we have the result above.
3
55 (2n + 1)2 − 1010 55 · 4n2 + 55 · 4n + 55 − 1010 55 · 4
lim 2
= lim 2
= = 2 · 54 = 1250
n→∞ 10n − 5 n→∞ 10n − 5 10
Because both the highest power on the numerator and the denominator are 1, then according to
Ex 1.1.6, we have the result above.
1.1.8 1
n n n2 − n − n2 − n −2n
lim − = lim = lim 2 =0
n→∞ n + 1 n − 1 n→∞ (n + 1)(n − 1) n→∞ n − 1

Because the highest power on the numerator is 1 and the highest power on the denominator is 2,
and 1 < 2, by Ex 1.1.6, we have the result above.

4
2
n2 n2 n3 − n2 − n3 − n2 −2n2
lim − = lim = lim 2 = −2
n→∞ n + 1 n − 1 n→∞ (n + 1)(n − 1) n→∞ n − 1

Because both the highest power on the numerator and the denominator are 2, then according to
Ex 1.1.6, we have the result above.
3
n n n3/2 − n − n3/2 − n −2n
lim √ −√ = lim = lim = −2
n→∞ n+1 n − 1 n→∞ n−1 n→∞ n − 1

Because both the highest power on the numerator and the denominator are 1, then according to
Ex 1.1.6, we have the result above.
4
n+a n+c n2 + (a + d)n + ad − n2 − (b + c)n − bc (a + d − b − c)n + ad − bc
lim − = lim = lim =0
n→∞ n + b n + d n→∞ n2 + (b + d)n + bd n→∞ n2 + (b + d)n + bd

Because the highest power on the numerator is 1 and the highest power on the denominator is 2,
and 1 < 2, by Ex 1.1.6, we have the result above.
5
n2 + a n2 + c n3 + dn2 + an + ad − n3 − bn2 − cn − bc (d − b)n2 + (a − c)n + ad − bc
lim − = lim 2
= lim =d−b
n→∞ n + b n+d n→∞ n + (b + d)n + bd n→∞ n2 + (b + d)n + bd

Because both the highest power on the numerator and the denominator are 2, then according to
Ex 1.1.6, we have the result above.
6
√ √
n+a n+c n3/2 + dn + a n + ad − n3/2 − bn − c n − bc
lim √ −√ = lim √
n→∞ n+b n + d n→∞ n + (b + d) n + bd

(d − b)n + (a − c) n + ad − bc
= lim √ =b−d
n→∞ n + (b + d) n + bd

Because both the highest power on the numerator and the denominator are 1, then according to
Ex 1.1.6, we have the result above.
7
n3 + a n3 + c n5 + dn3 + an2 + ad − n5 − bn3 − cn2 − bc
lim 2
− 2 = lim
n→∞ n + b n + d n→∞ n4 + (b + d)n2 + bd
(d − b)n3 + (a − c)n2 + ad − bc
= lim =0
n→∞ n4 + (b + d)n2 + bd

Because the highest power on the numerator is 3 and the highest power on the denominator is 4,
and 3 < 4, by Ex 1.1.6, we have the result above.
8
n2 + a n2 + c n5 + dn2 + an3 + ad − n5 − bn2 − cn3 − bc
lim 3
− 3 = lim
n→∞ n + b n + d n→∞ n6 + (b + d)n3 + bd
(a − c)n3 + (d − b)n2 + ad − bc
= lim =0
n→∞ n6 + (b + d)n3 + bd

Because the highest power on the numerator is 3 and the highest power on the denominator is 6,
and 3 < 6, by Ex 1.1.6, we have the result above.

5
9
√ √
n+a n+c n5/6 + dn1/2 + an1/3 + ad − n5/6 − bn1/2 − cn1/3 − bc
lim √ − √ = lim
n→∞ 3 n + b 3
n + d n→∞ n2/3 + (b + d)n1/3 + bd
(d − b)n1/2 + (a − c)n1/3 + ad − bc
= lim =0
n→∞ n2/3 + (b + d)n1/3 + bd

Because the highest power on the numerator is 1/2 and the highest power on the denominator is
2/3, and 1/2 < 2/3, by Ex 1.1.6, we have the result above.
1.1.9 1
n2 + a1 n + a0 n2 + c1 n + c0 n3 + (a1 + d)n2 + (a1 d + a0 )n + a0 d − n3 − (c1 + b)n2 − (c1 b + c0 )n − c0 b
lim − = lim
n→∞ n+b n+d n→∞ n2 + (b + d)n + bd
2
(a1 + d − c1 − b)n + (a1 d + a0 − c1 b − c0 )n + a0 d − c0 b
= lim
n→∞ n2 + (b + d)n + bd
=a1 + d − c1 − b

Because both the highest power on the numerator and the denominator are 2, then according to
Ex 1.1.6, we have the result above.
2
n2 + a1 n + a0 n 2 + c1 n + c0 n2 + a1 n + a0 n2 + c1 n + c0
lim − = lim − lim =1−1=0
n→∞ n2 + b1 n + b0 n2 + d1 n + d0 n→∞ n2 + b1 n + b0 n→∞ n2 + d1 n + d0

Because both the highest power on the numerator and the denominator are 2, then according to
Ex 1.1.6, those two limits both equal to 1.
3
 2  2  2  2
n+a n+c n+a n+c
lim − = lim − lim = 12 − 12 = 0
n→∞ n+b n+d n→∞ n + b n→∞ n + d

Because both the highest power on the numerator and the denominator are 1, then according to
Ex 1.1.6, those two limits both equal to 1.
4
2  2 2
n2 + a
 2
n + a n2 + c
 2
n + a n2 + c
 
n +c
lim − = lim + −
n→∞ n+b n+d n→∞ n+b n+d n+b n+d
 2
n + a n2 + c

lim − = d − b because of Ex1.1.9(1).
n→∞ n+b n+d
 2
√ √ n + a n2 + c √

2 2
For sufficiently large n, n + a ≥ n(n + b) and n + c ≥ n(n + d), so + ≥ 2 n.
n+b n+d
This means that for big n, if b 6= d, then
 2
n + a n2 + c
 2
n + a n2 + c √

n+b + n+d
− ≥ 2|d − b| n
n+b n+d

so the limit does not exist.


If b = d, then
2 2
n2 + a n2 + c (n2 + a)2 − (n2 + c)2 2(a − c)n2 + a2 − c2
 
lim − = lim = lim = 2(a − c)
n→∞ n+b n+d n→∞ (n + b)2 n→∞ n2 + 2bn + b2

6
1.1.10 1 According to Ex1.1.6, we have
(
np + a 0 if p < q,
lim q =
n→∞ n + b 1 if p = q.

If p > q, then for large n, np + a ≥ (nq + b) · n(p−q)/2 , so


np + a
≥ n(p−q)/2
nq + b
and it does not converge.
2 If p = q, then
anp + bnq + c (a + b)np + c
lim q p
= lim =1
n→∞ an + bn + c n→∞ (a + b)np + c

If p 6= q, without loss of generality, we may assume that p > q, then


anp + bnq + c a
lim =
n→∞ anq + bnp + c b
Because both the highest power on the numerator and the denominator are p, then according to
Ex 1.1.6, we have the result above.
3
np + a np + c np+q + anq + dnp + ad − np+q − cnq − bnp − bc
lim − = lim
n→∞ nq + b nq + d n→∞ n2q + (b + d)nq + bd
(d − b)n + (a − c)nq + ad − bc
p
= lim
n→∞ n2q + (b + d)nq + bd

Because p, q > 0, we must have 2q > q.


So if p < 2q, then the limit is 0.
If p = 2q, then the limit is d − b.
If p > 2q, then the limit does not exist.
4

2p p
n + a1 n + a2 0,
 if p < q,
lim 2q = 1, if p = q,
n→∞ n + b1 nq + b2 
does not exist , if p > q,

by Ex 1.1.6 and Ex 1.1.10(1).


1.1.11 If limn→∞ |xn | = 0, consider the inequality

−|xn | ≤ xn ≤ |xn |,

then limn→∞ −|xn | = − limn→∞ |xn | = 0.


By the Sandwich rule, limn→∞ xn = 0.
1.1.12 1 For sufficiently big n, 3n − 4 > n > 0, so √ 1 < √1 . Therefore,
3n−4 n

1 1
0≤ √ ≤√
3n − 4 n

By limn→∞ √1 = 0 and the Sandwich rule, we get limn→∞ √ 1 = 0.


n 3n−4

7
2n+3 1
2 For sufficiently big n, 2n2 + 3n ≤ 16n2 − 8n + 1, so (4n−1)2 ≤ n. Therefore

2n + 3 1
0≤ ≤√
4n − 1 n

√1 2n+3
By limn→∞ n
= 0 and the Sandwich rule, we get limn→∞ 4n−1 = 0.
a
3 For sufficiently big n, because a > 0, an + b ≥ 2 n. Therefore

1 1
0≤ ≤ pa
an + b 2n

and
r
1 2 1
lim p a = lim √ = 0
n→∞
2n
a n→∞ n

By the Sandwich rule, limn→∞ √ 1 = 0.


an+b

4 If c 6= 0, for sufficiently big n, because a > 0, then an2 + bn ≥ a2 (n + dc )2 . Therefore


√ r
an + b a 1
0≤ ≤ ·√
n + dc 2 n

pa an + b
By limn→∞ · √1 = 0 and the Sandwich rule, lim = 0. Then
2 n n→∞ n + dc
√ √
an + b 1 an + b
lim = lim =0
n→∞ cn + d c n→∞ n + dc

If c = 0, then

an + b 1√
= an + b
cn + d d
it does not converge.
1.1.13 1 Because − n1 ≤ cos n
n ≤ 1
n and limn→∞ 1
n = 0, limn→∞ − n1 = 0, by the Sandwich rule
cos n
lim =0
n→∞ n
(−1)n
2 Because − n1 ≤ n ≤ 1
n and limn→∞ 1
n = 0, limn→∞ − n1 = 0, by the Sandwich rule

(−1)n
lim =0
n→∞ n

sin n
3 Because − n1 ≤ n ≤ 1
n and limn→∞ 1
n = 0, limn→∞ − n1 = 0, by the Sandwich rule

sin n
lim =0
n→∞ n
1
4 Because − √n−2 ≤ √cos n
n−2
≤ √1
n−2
and limn→∞ √1
n−2
1
= 0 (by Ex1.1.12(3)), limn→∞ − √n−2 = 0,
by the Sandwich rule
cos n
lim √ = 0
n→∞ n

8
1 1 1
5 Because 0 ≤ n+(−1)n ≤ n−1 and limn→∞ n−1 = 0, by the Sandwich rule

1
lim =0
n→∞ n + (−1)n

1 cos n 1 1 1
6 Because − n+(−1) n ≤ n ≤ n+(−1)n and limn→∞ n+(−1)n = 0 (previous exercise), limn→∞ − n+(−1)n =

0, by the Sandwich rule


cos n
lim =0
n→∞ n + (−1)n

1 cos n √1 √1 1
7 Because − √n−2 ≤√ ≤ n−2
and limn→∞ n−2
= 0 (Ex 1.1.12(3)), limn→∞ − √n−2 = 0,
n+(−1)n 2
then by the Sandwich rule
cos n
lim p =0
n→∞ n + (−1)n 2

1 cos n √1 √1 1
8 Because − √n−1 ≤ √ √ ≤ n−1
and limn→∞ n−1
= 0 (Ex1.1.12(3)), limn→∞ − √n−1 = 0,
n+sin n
then by the Sandwich rule
cos n
lim p √ =0
n→∞ n + sin n
n
1
9 Because − √n−1 ≤ √ (−1) ≤ √1
n−1
and limn→∞ √1
n−1
1
= 0 (Ex1.1.12(3)), limn→∞ − √n−1 = 0,
n+(−1)n
then by the Sandwich rule

(−1)n
lim p =0
n→∞ n + (−1)n
√ √
2 3
1 2+(−1)n 3 5 532
10 Because for sufficiently big n, − n2/3 = −√
3
≤ √
3 2
n −2 cos n
≤ √
3
= n 2/3 , and
n2 − 12 n2 n2 − 12 n2

3 √
3
2 5 2
limn→∞ − n2/3 = 0, limn→∞ n2/3
= 0, then by the Sandwich rule

2 + (−1)n 3
lim √ =0
n→∞ 3 n2 − 2 cos n

2 sin n+(−1)n cos n


11 Because for sufficiently big n, − √n−1 ≤ √ ≤ √2 and limn→∞ √2 = 0 (Ex 1.1.6),
n+(−1)n n−1 n−1
2
limn→∞ − √n−1 = 0, then by the Sandwich rule,

sin n + (−1)n cos n


lim √
n→∞ n + (−1)n

| sin n+cos n| 2 2
12 Because 0 ≤ n ≤ n and limn→∞ n = 0, then by the Sandwich rule,

| sin n + cos n|
lim =0
n→∞ n
√ √
3 n+2 4 n √4 √4
13 Because for sufficiently big n, 0 ≤ 2n+(−1)n 3 ≤ 2n−n = n
and limn→∞ n
= 0, then by the
Sandwich rule

3 n+2
lim =0
n→∞ 2n + (−1)n 3

9
√ √ √
2 n n sin n+cos n 2 n
14 Because for sufficiently big n, − √4n = − n− n ≤
n−1 ≤ n− n = √4 ,
n
and limn→∞ √4
n
= 0,
2 2
limn→∞ − √4n = 0, then by the Sandwich rule

n sin n + cos n
lim =0
n→∞ n−1

n+sin n
15 Because for sufficiently big n, n−1
n+1 ≤ n+cos 2n ≤ n+1
n−1 , and limn→∞ n−1
n+1 = limn→∞ n+1
n−1 = 1 (Ex
1.1.6), then by the Sandwich rule

n + sin n
lim =1
n→∞ n + cos 2n

√ √ √ √ √
16 Because for sufficiently big n, √n−1
n+1
≤ √ n+sin n
n−cos n
≤ √n+1 ,
n−1
and limn→∞ √n−1
n+1
= limn→∞ √n+1
n−1
=1
(Ex 1.1.6), then by the Sandwich rule

n + sin n
lim √ =1
n→∞ n + cos n

1 (−1)n (n+1)
17 Because − n+1 = − nn+1
2 −1 ≤ n2 +(−1)n+1 ≤ n+1
n2 −1 = 1
n+1 , and limn→∞ 1
n+1
1
= 0, limn→∞ − n+1 = 0,
then by the Sandwich rule
(−1)n (n + 1)
lim =0
n→∞ n2 + (−1)n+1

(n+10)2 −1010 (−1)n (n+10)2 −1010 (n+10)2 +1010 (n+10)2 −1010


18 Because for sufficiently big n, 10n2 +5 ≤ 10(−1)n n2 −5 ≤ 10n2 −5 , and limn→∞ 10n2 +5 =
2 10
1 (n+10) +10 1
10 , limn→∞ 10n2 −5 = 10 (Ex 1.1.6), then by the Sandwich rule

(−1)n (n + 10)2 − 1010 1


lim n 2
=
n→∞ 10(−1) n − 5 10
√ √ √
n+a n+a n+a
1.1.14 1 Because for sufficiently big n, 0 ≤ n+(−1)n b ≤ n+|b| , and limn→∞ n+|b| = 0 (Ex 1.1.6), then by
the Sandwich rule

n+a
lim =0
n→∞ n + (−1)n b
1 1 1 1
2 Because for sufficiently big n, 0 ≤ √
3
n2 +an+b
≤ √
3
= √
3 1
, and limn→∞ √
3 1
= 0,
n2 − 14 n2 − 41 n2 2n
2
2n
2

then by the Sandwich rule


1
lim √
3
=0
n→∞ n2 + an + b

n+c+d

2 2n
√ √ 1 √ √ 1
3 Because for sufficiently big n, 0 ≤ √
3 2
n +an+b
≤ √
3 1 2
= 2 2 3 2 n1/6 and limn→∞ 2 2 3 2 n1/6 = 0,
2n
then by the Sandwich rule

n+c+d
lim √ =0
n→∞ 3 n2 + an + b

4
n n
n + (−1)n a 1 + a (−1)
n 1 + a limn→∞ (−1)
n
lim = lim n = (−1)n
=1
n→∞ n + (−1)n b n→∞ 1 + b (−1) 1 + b lim
n n→∞ n

by Ex1.1.13(2).

10
5
1 (−1)n
(−1)n (an + b) n (a + b · n ) 0·a
lim 2 = lim n = =0
n→∞ n + c(−1)n+1 n + d n→∞ 1 − c · (−1) + d 1−c·0+d·0
2 n n

6
n
(−1)n (an + b)2 + c (an + b)2 + (−1)n c (a + nb )2 + (−1) · c
lim = lim = lim (−1)n
n n
= a2
n→∞ (−1)n n2 + d n→∞ n2 + (−1)n d n→∞ 1+ · d
n n

7
√ √
cos n
cos n + a n + na
lim = lim
n→∞ n + b sin n n→∞ 1 + b · sin n
n

Because − n1 ≤ cosn n ≤ n1 and limn→∞ n1 = 0, limn→∞ − n1 = 0, then by the Sandwich rule

limn→∞ cosn n = 0. So the original limit is
√ √
cos n
cos n + a n + na 0+a·0
lim = lim = =0
n→∞ n + b sin n n→∞ 1 + b · sin n 1+b·0
n
√ √ √
−|a|−1
8 Because for sufficiently big n, − 2(|a|+1)

n
= √ 1
≤ cos n+a

n+b sin n
≤ √|a|+11 = 2(|a|+1)

n
, and
n− 2 n n− 2 n
limn→∞ √1 = 0, then by the Sandwich rule
n

cos n + a
√ =0
n + b sin n

9 Because − n1 ≤ sin n
n ≤ 1
n and limn→∞ 1
n = 0, then by the Sandwich rule limn→∞ sin n
n = 0. So

an + b sin n a + b sinn n a + b limn→∞ sin n


n a
lim = lim = =
n→∞ cn + d sin n n→∞ c + d sin n c + d limn→∞ sin n c
n n


sin n
1.1.15 1 Because − n1p ≤ np ≤ 1
np , and limn→∞ 1
np = 0 when p > 0, then by the Sandwich rule

sin n
lim =0
n→∞ np
sin(n+1)
2 Because for sufficiently big n, − n2p = − np −11 np ≤ np +(−1)n ≤ 1
np − 12 np
= 2
np , and limn→∞ 1
np = 0,
2
then by the Sandwich rule
sin(n + 1)
lim =0
n→∞ np + (−1)n
|a|+|b| |a|+|b|
3 Because for sufficiently big n, − 2(|a|+|b|)
np = np − 12 np
≤ a sin n+b
np +c ≤ np − 21 np
= 2(|a|+|b|)
np , and limn→∞ 1
np =
0, then by the Sandwich rule
a sin n + b
lim =0
n→∞ np + c
|a| |a|
4 Because for sufficiently big n, − 2|a|
np = np − 21 np
≤ a cos(sin n)
np −b sin n ≤ np − 12 np
= 2|a|
np , and limn→∞ 1
np = 0,
then by the Sandwich rule
a cos(sin n)
lim =0
n→∞ np − b sin n

11
√ √
1.1.16 For a < 0, the sequence n+a−
n satisfies
√ √ √ √
a a ( n + a − n)( n + a + n) √ √
√ <√ √ = √ √ = n+a− n<0
n n+a+ n n+a+ n
√ √
By limn→∞ √an = 0 and the Sandwich rule, we get limn→∞ ( n + a − n) = 0.
r
n+2
1.1.17 The sequence − 1 satisfies
n
q q
n+2 n+2
( − 1)( n + 1)
2
r
n+2 n 2
0< −1= q =q n ≤
n n+2
+1 n+2
+1 n
n n
q q
2 n+2 n+2
By limn→∞ n = 0 and the Sandwich rule, we get limn→∞ n − 1 = 0, so limn→∞ n = 1.
r
n+a n+a
1.1.18 • If a > b, then 1 < < . And limn→∞ n+a
n+b = 1 (Ex 1.1.6). By the Sandwich rule
n+b n+b
r
n+a
lim =1
n→∞ n+b
q
n+b
• If a < b, then by previous argument limn→∞ n+a = 1. So
r
n+a 1
lim = q =1
n→∞ n+b n+b
limn→∞ n+a

r
n+a
Then we may conclude that lim = 1.
n→∞ n+b
|a−b| |a−b| √ √ |a−b| |a−b|
1.1.19 1 The sequence satisfies − √ n+a
≤ − √n+a+ √
n+b
≤ n+a− n+b ≤ √ √
n+a+ n+b
≤ √
n+a
. And
|a−b| |a−b|
limn→∞ √ n+a
= 0 (Ex1.1.12(3)), limn→∞ − √ n+a
= 0. Then by the Sandwich rule
√ √
lim n+a− n+b=0
n→∞

2

n+a 1 1 1
lim √ √ = lim √ √ = √ √ =
n→∞ n + c + n + d n→∞ √ n+c + √n+d n+c
limn→∞ √n+a + n+d
limn→∞ √n+a 2
n+a n+a

By (Ex 1.1.18)
3
√ √
√ √ √ n+b √ n+b
n+a+ n+b 1+ n+a
1 + limn→∞ n+a
lim √ √ = lim √ √ = √ √ =1
n→∞ n + c + n + d n→∞ n+c √n+d √ n+c √n+d

n+a
+ n+a
limn→∞ n+a
+ limn→∞ n+a

By (Ex 1.1.18)
4
√ √ √ √
n+a n+b n+a n+b
lim √ √ = lim √ lim √ =1
n→∞ n+c n+d n→∞ n+c n→∞ n+d
By (Ex 1.1.18)

12
5
√ √
√ √n+a + √b limn→∞ √n+a + limn→∞ √b
n+a+b n+c n+c n+c n+c
lim √ = lim = =1
n→∞ n + c + d n→∞ 1+ √d limn→∞ 1 + d
limn→∞ √n+c
n+c

By (Ex 1.1.18 and 1.1.12(3))


6
√ √ √ √
√ √ √ √ ( n + a − n + b)( n + a + n + b)
lim n( n + a − n + b) = lim n · √ √
n→∞ n→∞ n+a+ n+b

(a − b) n
= lim √ √
n→∞ n+a+ n+b
a−b
= lim √n+a √
n→∞ √
n
+ √n+b
n
a−b a−b
= √ √ =
limn→∞ √n+a
+ limn→∞ √n+b 2
n n

By (Ex 1.1.18)
7
√ √ √ √
√ √ √ √ ( n + a − n + b)( n + a + n + b)
lim n + c( n + a − n + b) = lim n + c · √ √
n→∞ n→∞ n+a+ n+b

(a − b) n + c
= lim √ √
n→∞ n+a+ n+b
a−b
= lim √n+a √
n→∞ √ n+b
n+c
+ √n+c
a−b a−b
= √ √ =
limn→∞ √n+a + n+b
limn→∞ √n+c 2
n+c

By (Ex 1.1.18)
8
√ √ √ √ √ √ √
lim n+a+ n + b − 2 n + c = lim ( n + a − n + c) + lim ( n + b − n + c) = 0
n→∞ n→∞ n→∞

By (Ex 1.1.19(1))
r
n
9 The sequence 2
satisfies
n +n+1
s s
n + 12
r
n 1
0< < 1 = 1
n2 + n + 1 2
n +n+ 4 n+ 2
s
1
Because lim 1 = 0 (Ex 1.1.12(3)), the Sandwich rule gives us
n→∞ n+ 2
r
n
lim =0
n→∞ n2 + n + 1

10 Because for sufficiently big n, (n + a) n ≤ n2 + bn + c, we have
r
n+a 1
0≤ ≤ √
na + bn + c 4
n

13
1
Because limn→∞ n1/4
= 0, the sandwich rule gives us
r
n+a
lim =0
n→∞ n2 + bn + c

11
√ √ √ √
p p ( n2 + an + b − n2 + cn + d)( n2 + an + b + n2 + cn + d)
n2 + an + b − n2 + cn + d = √ √
n2 + an + b + n2 + cn + d
(a − c)n + b − d
=√ √
n + an + b + n2 + cn + d
2

a − c + b−d
=q qn
n2 +an+b 2
n2 + n +cn+d
n2
r
n2 + an + b
Consider the sequence .
n2
r
n2 + an + b n+a
• If a > 0, for sufficiently big n, n2 < n2 + an + b < n2 + 2an + a2 , so 1 < 2
< .
r n n
2
n + an + b
Because limn→∞ n+a n = 1, the Sandwich rule gives us n→∞
lim = 1.
n2
2 2 2 2
• If a < 0, r for sufficiently big n, n + 2an + a < n + an + b < n , and n + a > 0.
2
n + an + b n+a
So 1 > 2
> . Because limn→∞ n+a n = 1, the Sandwich rule gives us
r n n
n2 + an + b
lim = 1.
n→∞ n2
r
n2 + an + b
In conclusion, lim = 1.
n→∞
r n2
n2 + cn + d
Similarly, lim = 1.
n→∞ n2
Then by the arithmetic rule
p p
lim n2 + an + b − n2 + cn + d
n→∞
a − c + b−d
= lim q qn
n→∞ n2 +an+b n2 +cn+d
n2 + n2

limn→∞ a − c + limn→∞ b−d


= q qn
n2 +an+b 2
limn→∞ n2 + limn→∞ n +cn+d
n2
a−c
=
2
12
√ √ √ √ p p a+b−c−d
lim n + a n + b − n + c n + d = lim n2 + (a + b)n + ab − n2 + (c + d)n + cd =
n→∞ n→∞ 2
By (Ex 1.1.19(11))
n n n n n
13 The sequence √ satisfies =√ <√ < 1. And limn→∞ n+1 =
n2 + n + 1 n+1 n2 + 2n + 1 n2 + n + 1
1. Then by the Sandwich rule, we have
n
lim √ =1
n→∞ n2 + n + 1

14
14 • If b > 0, then for sufficiently big n, n2 < n2 + bn + c < n2 + 2bn + b2 , and n + a > 0.
n+a n+a n+a
So < √ < . Because limn→∞ n+a
n+b = 1 and limn→∞ n
n+a
= 1, the
n+b 2
n + bn + c n
Sandwich rule gives us
n+a
lim √ =1
n→∞ n2 + bn + c

• If b < 0, then for sufficiently big n, n2 > n2 + bn + c > n2 + 2bn + b2 , and n + a > 0.
n+a n+a n+a
So > √ > . Because limn→∞ n+a
n+b = 1 and limn→∞ n
n+a
= 1, the
n+b 2
n + bn + c n
Sandwich rule gives us
n+a
lim √ =1
n→∞ n2 + bn + c

In conclusion,
n+a
lim √ =1
n→∞ n2
+ bn + c
s r s
n2 − 2|a|n + a2 n2 + an + b n2 + 2|a|n + a2
15 For sufficiently big n, ≤ ≤ . That is
n2 + 2|c|n + c2 n2 + cn + d n2 − 2|c|n + c2
r
n − |a| n2 + an + b n + |a|
≤ 2

n + |c| n + cn + d n − |c|
n−|a| n+|a|
Because limn→∞ n+|c| = 1 and limn→∞ n−|c| = 1, the Sandwich rule gives us
r
n2 + an + b
lim =1
n→∞ n2 + cn + d

1.1.20 1
√ √ √ √
√ √ ( n + a sin n − n + b cos n)( n + a sin n + n + b cos n)
n + a sin n − n + b cos n = √ √
n + a sin n + n + b cos n
a sin n − b cos n
=√ √
n + a sin n + n + b cos n

|a| + |b| a sin n − b cos n |a| + |b|


And for sufficiently big n, − p ≤√ √ ≤p .
n − |a| n + a sin n + n + b cos n n − |a|
|a| + |b| |a| + |b|
Because lim p = 0 (Ex 1.1.12(3)), and lim − p = 0, the Sandwich rule gives us
n→∞ n − |a| n→∞ n − |a|
√ √
lim n + a sin n − n + b cos n = 0
n→∞

r s r s
n + a sin n n − |a| n + a sin n n + |a|
2 The sequence satisfies ≤ ≤ for sufficiently big n.
n + b cos n n + |b| n + b cos n n + |b|
s s
n − |a| n + |a|
And lim = 1, lim = 1. (Ex 1.1.18) Then the Sandwich rule gives us
n→∞ n + |b| n→∞ n − |b|
r
n + a sin n
lim =1
n→∞ n + b cos n

15
s s s s
n + (−1)n a n − |a| n + (−1)n a n + |a|
3 The sequence satisfies ≤ ≤ for sufficiently big n.
n + (−1)n b n + |b| n + (−1)n b n + |b|
s s
n − |a| n + |a|
And lim = 1, lim = 1. (Ex 1.1.18) Then the Sandwich rule gives us
n→∞ n + |b| n→∞ n − |b|
s
n + (−1)n a
lim =1
n→∞ n + (−1)n b

4

√ √n+a + sin n

n + a + sin n n+c n+c
√ = (−1)n
n + c + (−1)n 1+ √
n+c

n+a
lim √ = 1 because of Ex 1.1.18.
n→∞ n+c
sin n 1 sin n 1
The sequence √ satisfies − √ <√ <√ for sufficiently big n. And limn→∞ √1 =
n+c n+c n+c n+c n+c

1 sin n
0, (Ex 1.1.12(2)), limn→∞ − √n+c = 0. Then the Sandwich rule gives us lim √ = 0.
n→∞ n+c
n n
(−1) 1 (−1) 1
The sequence √ satisfies − √ <√ <√ for sufficiently big n. And limn→∞ √1 =
n+c n+c n+c n+c n+c

1 (−1)n
0, (Ex 1.1.12(2)), limn→∞ − √n+c = 0. Then the Sandwich rule gives us lim √ = 0.
n→∞ n+c
So the original limit is

√ √n+a + sin n

n + a + sin n n+c n+c
lim √ = lim (−1)n
=1
n→∞ n + c + (−1)n n→∞ 1+ √
n+c

5 – If a = b, then the limit of the sequence is 0.


p √ √ √ √ √
– If a > b, then the sequence n + (−1)n ( n + a− n + b) satisfies n − 1( n + a− n + b) ≤
p √ √ √ √ √
n + (−1)n ( √n + a −√ n + b) ≤ √ n + 1( n + a − n + b). √ √ √
And limn→∞ n − 1( n + a − n + b) = (a − b)/2, limn→∞ n + 1( n + a − n + b) =
(a − b)/2 (Ex 1.1.19(7)). By the Sandwich rule, we have
p √ √ a−b
lim n + (−1)n ( n + a − n + b) =
n→∞ 2
p √ √ √ √ √
– If a < b, then the sequence n + (−1) n ( n + a− n + b) satisfies n − 1( n + a− n + b) ≥
p √ √ √ √ √
n
n + (−1) ( √n + a −√ n + b) ≥ √ n + 1( n + a − n + b). √ √ √
And limn→∞ n − 1( n + a − n + b) = (a − b)/2, limn→∞ n + 1( n + a − n + b) =
(a − b)/2 (Ex 1.1.19(7)). By the Sandwich rule, we have
p √ √ a−b
lim n + (−1)n ( n + a − n + b) =
n→∞ 2
√ √ √ √ √
6 The 2 an + sin n− n2√+ bn + cos n satisfies n2 + an − 1− n2 + bn + 1 ≤ n2 + an + sin n−
√ sequence n +√
n2 + bn + cos 2
√ n ≤ n + an √ + 1 − n2 + bn − 1. √ √
And limn→∞ n + an − 1 − n2 + bn + 1 = (a − b)/2, limn→∞ n2 + an + 1 − n2 + bn − 1 =
2

(a − b)/2 (Ex 1.1.19(11)). by the Sandwich rule,


p p a−b
lim n2 + an + sin n − n2 + bn + cos n =
n→∞ 2

16
r r r r
n2 + an + sin n n2 + an − 1 n2 + an + sin n n2 + an + 1
7 The sequence satisfies ≤ ≤ for
n2 + bn + cos n n2 + bn + 1 n2 + bn + cos n n2 + bn − 1
sufficiently big n.
r r
n2 + an − 1 n2 + an + 1
And lim 2
= 1, lim = 1 by Ex 1.1.19(15).
n→∞ n + bn + 1 n→∞ n2 + bn − 1
So the Sandwich rule gives us
r
n2 + an + sin n
lim =1
n→∞ n2 + bn + cos n
√ s √
n2 + an + b n2 + an + b n2 + an + b
8 For sufficiently big n, the sequence n
satisfies 2 2
≤ ≤
n + (−1) c n + 2|c|n + c n + (−1)n c
s
n2 + an + b
.
n − 2|c|n + c2
2
s s
n2 + an + b n2 + an + b
And lim = 1, lim = 1 by Ex 1.1.19(15). Then the Sandwich
n→∞ n2 + 2|c|n + c2 n→∞ n2 − 2|c|n + c2
rule gives us

n2 + an + b
lim =1
n→∞ n + (−1)n c

1.1.21 1
√ √ √ 2 √ √ √ 2

3

3 ( 3 n + a − 3 n + b)( 3 n + a + 3 n + a 3 n + b + 3 n + b )
n+a− n+b= √ 2 √ √ √ 2
3
n+a + 3n+a3n+b+ 3n+b
a−b
=√ 2 √ √ √ 2
3
n+a + n+a3n+b+ 3n+b
3

Then for sufficiently big n, we have



3
1 1 1 2
0≤ √ 2 √ √ √ 2 ≤ √ ≤ q = √
3 3 3 3
n+a + n+a n+b+ n+b
3
n+a 3
n − 12 n
3
n


3 1
2
Because limn→∞ √
3 n = 0, by the Sandwich rule, lim √ 2 √ √ √ 2 = 0, so
n→∞ 3
n+a + 3
n+a3n+b+ 3n+b
a−b
lim √ 2 √ √ √ 2 = 0.
n→∞ 3
n+a + n+a3n+b+ 3n+b
3

√ √
3
That is lim 3 n + a − n + b = 0
n→∞
r
3 n+a n+a n+a
2 • If a > b, then the sequence satisfies 1 ≤ ≤ . And lim = 1. By the
n+b n+b n→∞ n + b
Sandwich rule, we have that
r
3 n+a
lim =1
n→∞ n+b
r
3 n+a n+a n+a
• If a < b, then the sequence satisfies 1 ≥ ≥ . And lim = 1. By the
n+b n+b n→∞ n + b
Sandwich rule, we have that
r
3 n+a
lim =1
n→∞ n+b

17
r
n+a
• If a = b, then 3 = 1. So it is a constant sequence with limit 1.
n+b
r
n+a
So lim 3 = 1.
n→∞ n+b
3


3 √
3

3
3
n2 (a − b)
n2 ( n + a − n + b) = √ 2 √√ √ 2
3
n+a + n+a3n+b+ 3n+b
3

a−b
=q q q q q q
3 n+a 3 n+a 3 n+a 3 n+b 3 n+b 3 n+b
n n + n n + n n

r r
3 n+a 3 n+b
Because lim = 1, lim = 1 (Ex 1.1.21(2)).
n→∞ n n→∞ n


3 √ √
3 a−b
lim n2 ( 3 n + a − n + b) =
n→∞ 3
by Arithmetic Rule.
4

√ 3 √ √ n(a − b) 3
q q
3
3
n( n+a− n + b) = p√ 2 p√p √ p√ 2
3 3
n+a + n+a3 n+b+ 3 n+b
a−b
=q √ q √ q √
3 n+2a n+a2 3 n+(a+b) n+ab 3 n+2b n+b2
n + n + n
r √
n+c n+d
3
Consider lim .
n→∞ n
r √ r
3 n + c n+d 3 n + d
• If c = 0, then lim = lim = 1 (Ex 1.1.21(2))
n→∞ n n→∞ n
• If c > 0, then for sufficiently big n,
r √ √
3 n + c n+d n+c n+d c d
1≤ ≤ =1+ √ +
n n n n

c d
Because lim 1 + √ + = 1, by the Sandwich rule we have
n→∞ n n
r √
3 n + c n+d
lim =1
n→∞ n

• If c < 0, then for sufficiently big n,


r √ √
3 n + c n+d n+c n+d c d
1≥ ≥ =1+ √ +
n n n n

c d
Because lim 1 + √ + = 1, by the Sandwich rule we have
n→∞ n n
r √
3 n + c n+d
lim =1
n→∞ n

18
r √
n+c n+d
3
So lim = 1.
n→∞ n
Then by Arithmetic rule
√ 3 √ √ a−b
q q
3
lim 3
n( n+a− n + b) =
n→∞ 3
1.1.22 1
5 2 5 2 5
1− limn→∞ 1 − limn→∞
  
n−2 n n
lim = lim 1 = 1 =1
n→∞ n+1 n→∞ 1+ n limn→∞ 1 + limn→∞ n

by Arithmetic Rule.
2 Because 0 < n−2
n+1 < 1, we have
 6  5.4  5
n−2 n−2 n−2
< <
n+1 n+1 n+1
 6  5
n−2 n−2
And = 1, = 1. The Sandwich rule gives us
n+1 n+1
 5.4
n−2
lim =1
n→∞ n+1
n+1
3 Because n−2 > 1, we have
    √2  2
n+1 n+1 n+1
< <
n−2 n−2 n−2
   2   √2
n+1 n+1 n+1
And = 1, = 1. The Sandwich rule gives us lim = 1. So
n−2 n−2 n→∞ n−2
  −√2
n−2
lim =1
n→∞ n+1
n+a
4 Without loss of generality, we may assume that a > b and p > 0. So n+b > 1 and
 p  [p]+1
n+a n+a
1≤ ≤
n+b n+b
 [p]+1
n+a
lim = 1 by Arithmetic Rule. So
n→∞ n+b
 p
n+a
lim =1
n→∞ n+b

1.1.23 1 Without loss of generality we may assume p > 0, so for sufficiently big n
[p]+1
√ p  √

[p]+1 |a|
n + a sin n n + |a| 1+ √ n
0≤ √ ≤ √ =
n + b cos 2n n − |b| 1 − √|b|n
 [p]+1
|a|
1+ √
n
Because limn→∞ √1 = 0, the Arithmetic rule gives lim   = 0. So the Sandwich
n n→∞ |b|
1− √
n
rule gives us
√ p
n + a sin n

lim =0
n→∞ n + b cos 2n

19
2 Without loss of generality we may assume p > 0, so
p
n2 + an + b

• If a > 0, then for sufficiently big n, the sequence satisfies
n2 + (−1)n c
p [p]+1
n2 + an + b n2 + an + b
 
1≤ ≤
n2 + (−1)n c n2 − |c|
[p]+1
n2 + an + b

Then lim = 1 by Arithmetic rule. So the Sandwich rule gives us
n→∞ n2 − |c|
p
n2 + an + b

lim =1
n→∞ n2 + (−1)n c

 2 p
n + an + b
• If a > 0, then for sufficiently big n, the sequence satisfies
n2 + (−1)n c
p [p]+1
n2 + an + b n2 + an + b
 
1≥ ≥
n2 + (−1)n c n2 + |c|
[p]+1
n2 + an + b

Then lim = 1 by Arithmetic rule. So the Sandwich rule gives us
n→∞ n2 + |c|
p
n2 + an + b

lim =1
n→∞ n2 + (−1)n c
Then we may conclude that
p
n2 + an + b

lim =1
n→∞ n2 + (−1)n c
3 Without loss of generality we may assume that p > 0. Then for sufficiently big n
p  p
n + 12 n

n+a 3
0≤ 2
≤ 2 1 2
= p
n + bn + c n − 2n n
1
Because limn→∞ np = 0 when p > 0, the Sandwich rule gives us
 p
n+a
lim =0
n→∞ n2 + bn + c

1.1.24 If xn ≤ 1, then
• If 0 < p < 1, then xn < xpn < 1. So limn→∞ xn = 1 and the Sandwich rule gives us limn→∞ xpn = 1.
[p]+1 [p]+1
• If p > 1, then 1 > xpn > xn . Because limn→∞ xn = 1, the Arithmetic rule gives us limn→∞ xn =
1. Then by the Sandwich rule limn→∞ xpn = 1.
So limn→∞ xpn = 1.
√ √ √
sufficiently big n, then n a ≤ n xn ≤ n b for sufficiently
1.1.25 If a ≤ xn ≤ b for some constants a, b > 0 and √
√ √
big n. Because limn→∞ n a = 1, and limn→∞ n b = 1, the Sandwich rule gives us limn→∞ n xn = 1.
1 1 √
1.1.26 1 For sufficiently big n, 1 ≤ n 2n ≤ n n . Because limn→∞ n n = 1, the Sandwich rule gives us
1
limn→∞ n 2n = 1.
√ 2 1
 1
2
2 Because limn→∞ n n = 1, by the Arithmetic rule limn→∞ n n = limn→∞ (n n )2 = limn→∞ n n =
1.

20
c √ [c]+1
3 Without loss of generality we may assume that c > 0, then 1 < n n < ( n n) . Because
√ √ [c]+1
limn→∞ n = 1 and the Arithmetic rule, we have limn→∞ ( n)
n n
= 1. Then the Sandwich
c
rule gives us limn→∞ n n = 1.
c c
4 Without loss of generality, we may assume that c > 0. The sequence (n + 1) n satisfies n n ≤
c c 1 c c 1
(n + 1) n ≤ (n + n) n = (2c ) n · n n . Because limn→∞ n n = 1 and limn→∞ (2c ) n = 1, the Arithmetic
1 c c
c n
rule gives us limn→∞ (2 ) · n n = 1. So limn→∞ (n + 1) n = 1 by the Sandwich rule.
c 1 c
5 Without loss of generality, we may assume that c > 0. The sequence (an+b) n satisfies ((a/2)c ) n n n =
c c c 1 c c
(an− 12 an) n ≤ (an+b) n ≤ (an+an) n = ((2a)c ) n ·n n for sufficiently big n. Because limn→∞ n n = 1
1 1 c
and limn→∞ k n = 1 for k > 0, the Arithmetic rule gives us limn→∞ ((a/2)c ) n n n = 1 and
1 c c
limn→∞ ((2a)c ) n · n n = 1. So limn→∞ (an + b) n = 1 by the Sandwich rule.
c
6 Without loss of generality, we may assume that c > 0. The sequence (an2 + b) n satisfies
1 2c 1 c c c 1 2c
((a/2)c ) n n n = (an2 − an2 ) n ≤ (an2 + b) n ≤ (an2 + an2 ) n = ((2a)c ) n · n n
2
c 1
for sufficiently big n. Because limn→∞ n n = 1 and limn→∞ k n = 1 for k > 0, the Arithmetic rule
gives us
2c
lim n n = 1
n→∞
1 c
lim ((a/2)c ) n n n = 1
n→∞
1 c
lim ((2a)c ) n · n n = 1
n→∞
c
So limn→∞ (an2 + b) n = 1 by the Sandwich rule.
√ c √ c
7 Without loss of generality, we may assume that c > 0. The sequence ( n + 1) n satisfies ( n) n ≤
√ c √ √ c 1 c/2 c/2 1
( n + 1) n ≤ ( n + n) n = (2c ) n · n n . Because limn→∞ n n = 1 and limn→∞ (2c ) n = 1, the
1 c/2 √ c
Arithmetic rule gives us limn→∞ (2c ) n · n n = 1. So limn→∞ ( n + 1) n = 1 by the Sandwich rule.
1 1
8 Without loss of generality, we may assume that c > 0. The sequence (n−2) n+3 satisfies (n−2) n+n ≤
1 1 1
(n − 2) n+3 ≤ (n − 2) n for sufficiently big n. Then by Ex 1.1.26(5), limn→∞ (n − 2) n+3 = 1.
c
9 Without loss of generality, we may assume that c > 0. The sequence (an + b) n+d satisfies (an +
c
c c
n− 1 n
b) n+n ≤ (an + b) n+d ≤ (an + b) for sufficiently big n. Then by Ex 1.1.26(5), limn→∞ (an +
2
c/2 2c c
b) n = 1 and limn→∞ (an + b) n = 1. So the Sandwich rule gives us limn→∞ (an + b) n+d = 1.
cn
10 Without loss of generality, we may assume that c > 0. The sequence (an + b) n2 +dn+e satisfies
cn cn cn
2 1 2
(an + b) n2 +n2 +n2 ≤ (an + b) n2 +dn+e ≤ (an + b) n − 2 n for sufficiently big n. Then by Ex 1.1.26(5),
c/3 2c
limn→∞ (an + b) n = 1 and limn→∞ (an + b) n = 1.
cn
So the Sandwich rule gives us limn→∞ (an + b) n2 +dn+e = 1.
c
11 Without loss of generality, we may assume that c > 0. The sequence (an2 + b) n+d satisfies (an2 +
c
c c
n− 1 n
b) n+n ≤ (an2 + b) n+d ≤ (an2 + b) for sufficiently big n. Then by Ex 1.1.26(6), limn→∞ (an2 +
2
c/2 2c c
b) n = 1 and limn→∞ (an2 + b) n = 1. So the Sandwich rule gives us limn→∞ (an2 + b) n+d = 1.
cn
12 Without loss of generality, we may assume that c > 0. The sequence (an2 + b) n2 +dn+e satisfies
cn cn cn
n2 − 1 n2
(an2 + b) n2 +n2 +n2 ≤ (an2 + b) n2 +dn+e ≤ (an2 + b) for sufficiently big n. Then by Ex
2
c/3 2c
1.1.26(5), limn→∞ (an2 + b) n = 1 and limn→∞ (an2 + b) n = 1.
cn
So the Sandwich rule gives us limn→∞ (an2 + b) n2 +dn+e = 1.
√ √ √ √
1.1.27 1 The sequence n n + √sin n satisfies n n − 1√< n n + sin n < n n + 1 for sufficiently big n. By
Ex1.1.26(5) limn→∞ n n − 1 = 1, limn→∞ n n + 1 = 1.

Then the Sandwich rule gives us limn→∞ n n + sin n = 1

21
√ p √ p
2 The sequence n an + b sinp n satisfies n an − |b| <pn an + b sin n < n an + |b| for sufficiently big n.
By Ex1.1.26(5) limn→∞ n an − |b| = 1, limn→∞ n an + |b| = 1.

Then the Sandwich rule gives us limn→∞ n an + b sin n = 1
p √ √ √
3 The sequence n n + (−1) n sin n satisfies n n − 1 < n n + sin n < n n + 1 for sufficiently big n. By
√ √
Ex1.1.26(5) limn→∞ n n − 1 = 1, limn→∞ p n
n + 1 = 1.
Then the Sandwich rule gives us limn→∞ n + (−1)n sin n = 1.
n

p p p p
4 The sequence n an + (−1)n b sin n satisfiesp
n
(−1)n b sin n < n an + |b| for suf-
an − |b| < n an +p
ficiently big n. By Ex1.1.26(5) limn→∞ n an − |b| = 1, limn→∞ n an + |b| = 1.
p
Then the Sandwich rule gives us limn→∞ n an + (−1)n b sin n = 1.
1
1 1 1 1
5 The sequence (n − cos n) n+sin n satisfies
√ (n − 1) n+n < (n
√ − cos n) n+sin n < (n + 1) n− 2 n for sufficiently
n n
big n. By Ex1.1.26(5) limn→∞ n − 1 = 1, limn→∞ n + 1 = 1.
1
Then the Sandwich rule gives us limn→∞ (n − cos n) n+sin n = 1.
n n n n
2 − 1 n2
6 The sequence (an+b sin n) n2 +c cos n satisfies (an−|b|) n2 +n2 < (an+b sin n) n2 +c cos n < (an+|b|) n 2
c/2 2c
for sufficiently big n. By Ex1.1.26(5) limn→∞ (an − |b|) n = 1, limn→∞ (an + |b|) n = 1.
n
Then the Sandwich rule gives us limn→∞ (an + b sin n) n2 +c cos n = 1.
√ q p q √ √ √ p
1.1.28 1 The sequence n np + sin n satisfies n 12 n n = n np − 21 np < n np + sin n < n np + np = n 2n n .
q p √ p
By Ex 1.1.26(3) and the Arithmetic rule limn→∞ n 12 n n = 1 and limn→∞ n 2n n = 1.

The Sandwich rule gives us limn→∞ n np + sin n = 1
√ √
2 Without loss of generality, we may assume that p > q. The sequence n
n p + nq satisfies n np <
√ √ p √ p
n
np + nq < n np + np . By Ex 1.1.26(3) and the Arithmetic rule limn→∞ n n = 1 and limn→∞ n 2n n =
1.

The Sandwich rule gives us limn→∞ n np + nq = 1
√ p √
3 Without loss of generality, we may assume that p > q. The sequence n+2
n + nq satisfies n+2 np <
√ p
n+2

q < n+2 np + np . By Ex 1.1.26(9) and the Arithmetic rule lim
p
n + n√ n→∞ n n+2 = 1 and
p
limn→∞ n+2 2n n+2 = 1.

The Sandwich rule gives us limn→∞ n+2 np + nq = 1
√ √
n−2
4 Without
√ loss of generality,
√ we may assume that p > q. The sequence n−2 np + nq satisfies p
np <
n−2 p q
n + n√ < n−2 p p
n + n . By Ex 1.1.26(9) and the Arithmetic rule limn→∞ n n−2 = 1 and
p
limn→∞ n−2 2n n−2 = 1.

The Sandwich rule gives us limn→∞ n−2 np + nq = 1
√ √
1.1.29 1 For sufficiently big n, 5n > 5n − 4n >√5n−1 + 4 · 5n−1 − 4 · 4n−1 > 5n−1 , so 5 > n 5n − 4n > 5/ n 5.
5
By limn→∞ √ n
5
= 5, we get limn→∞ n 5n − 4n = 5.
n n √ q
2 For sufficiently big n, 54 < 16 , then 1 − 3 · 54 > 12 . So 5 > n 5n − 3 · 4n > 5 · n 12 . By
q √
limn→∞ n 12 = 1, and the Sandwich rule, we get limn→∞ n 5n − 3 · 4n = 5.
√ √ √
3 The sequence n 5n − 3 · 4n + 2n √ satisfies n 5n − 3 · 4n < n 5n − 3 · 4n + 2n < 5. By Ex1.1.29(2)
and the Sandwich rule, limn→∞ n 5n − 3 · 4n + 2n = 5.
n n n n
4 For sufficiently big n, 54 < 19 and 25 < 13 , so 1 − 3 · 45 − 25 > 31 .
r
n 1
√ q
So 5 · ≤ n 5n − 3 · 4n − 2n ≤ 5. By limn→∞ n 13 = 1 and the Sandwich rule, we have that
√ 3
lim n 5n − 3 · 4n − 2n = 5.
n→∞
n n q √ q
5 For sufficiently big n, 58 < 81 , then 14 − 85 > 18 . So 8 · n 18 < n 42n−1 − 5n < n 14 · 8. By
q q √
limn→∞ n 18 = 1 and limn→∞ n 14 = 1, and the Sandwich rule, we get that limn→∞ n 42n−1 − 5n =
8.

22
n n q p
6 For sufficiently big n, 58 < 18 , then 14 − 85 > 18 . So 8 · n 18 < n 42n−1 + (−1)n 5n <
√ q q q
n
42n−1 + 42n−1 = n 12 · 8. By limn→∞ n 18 = 1 and limn→∞ n 12 = 1, and the Sandwich rule, we

get that limn→∞ n 42n−1 − 5n = 8.
q
7 For sufficiently big n, 5n−1 > 5n−1 − 4n−1 > 5n−2 + 4 · 5n−2 − 4 · 4n−2 > 5n−2 , so 5 n 15 >
√n
√ 5 5

n
5n−1 − 4n−1 > 5/ n 25. By limn→∞ √ n
5
= 5, lim n→∞ √
n
25
= 5, we get lim n→∞ 5n−1 − 4n−1 =

5. So limn→∞ n+1
5n − 4n = 5.
1 1 1 √ 1
8 For sufficiently big n, 1 < (5n −4n ) n < (5n −4n ) n−2 < (5n ) n−2 = n−2 25·5. By limn→∞ (5n −4n ) n =
√ 1
5 and limn→∞ n−2
25 · 5 = 5, And the Sandwich rule, we get that limn→∞ (5n − 4n ) n−2 = 5.
1 n+1 n+1 n+1 √
9 For sufficiently big n, (5n −4n ) n < (5n −4n ) n2 +n < (5n −4n ) n2 +1 < (5n ) n2 . By limn→∞ n 5n − 4n =
n+1
5 and the Sandwich rule, we get that limn→∞ (5n − 4n ) n2 +1 = 5.
√ √ √ √ √ √ √
1.1.30 1 The sequence n an + bn satisfies n an ≤ n an√+ bn ≤ n an + an = n 2 · n an . By limn→∞ n 2 = 1
and the Sandwich rule, we get that limn→∞ n an + bn = a.
n n q √ √
2 Because a > b > 0, ab ≤ ab , so 1 − ab > a−ba . That is a · n a−b
a <
n
an − bn < n an . By
q √
limn→∞ n a−b a = 1 and the Sandwich rule, we get that limn→∞
n
an − bn = a.
p √ p √
3 The sequence n an + (−1)n bn satisfies n apn − bn ≤ n an + (−1)n bn ≤ n an + bn . Then by Ex1.1.30(1)

and (2), and the Sandwich rule limn→∞ n an + (−1)n bn = a.



n

4 limn→∞ an b2n+1 = limn→∞ ab2 n b = ab2 .
√ √
n+2 n+2 √ √ q
5 The sequence n+2 an + bn satisfies n+2√a 2 < n+2 an + bn < n+2 an + an = a · n+2 a22 . By
a
√ q √
limn→∞ n+2 a = 1 and limn→∞ n+2 a22 = 1, and the Sandwich rule, we get limn→∞ n+2 an + bn = a.
n n √ p
6 Because a > b > 0, ab ≤ ab , so 1− ab > a−b a . The sequence n−2 an − bn satisfies a n−2 a(a − b) =
q √ √ √ √ √
an · a−b
n−2 n−2
n−2
a <
n−2
an − bn < n−2 an = a · a2 . By limn→∞ n−2 a2 − ab = 1, limn→∞ a2 =

n−2
1, and the Sandwich rule, we get that limn→∞ an − bn = 1.
n n n n n
7 The sequence (an + bn ) n2 −1 satisfies (an ) n2 −1 < (an + bn ) n2 −1 < (an + an ) n2 −1 = (2an ) n2 −1 .
n n2 1 1
• If a > 1, then a < (an ) n2 −1 < a n2 −n = a · a n−1 . By limn→∞ a n−1 = 1 and the Sandwich rule,
n
we get that limn→∞ (an ) n2 −1 = a.
n n2 1 1
• If 0 < a < 1, then a > (an ) n2 −1 > a n2 −n = a · a n−1 . By limn→∞ a n−1 = 1 and the Sandwich
n
rule, we get that limn→∞ (an ) n2 −1 = a.
1 n 1 1 1
Also 2 n < 2 n2 −1 < 2 n−1 . By limn→∞ 2 n = 1 and limn→∞ 2 n−1 = 1 and the Sandwich rule, we get
n
that limn→∞ 2 n2 −1 = 1.n
So limn→∞ (an + bn ) n2 −1 = a by the Sandwich rule.
n n n
8 Because a > b > 0, ab ≤ ab , so 1 − ab > a−b
a . The sequence (an − bn ) n2 −1 satisfies (an ·
n n n n n
a−b n2 −1
a ) < (an − bn ) n2 −1 < (an ) n2 −1 . By limn→∞ (an ) n2 −1 = a, limn→∞ ( a−b
a ) n2 −1 = 1 and the
n
Sandwich rule, we get that limn→∞ (an − bn ) n2 −1 = a.
n n n n
9 The sequence (an − (−1)n bn ) n2 −1 satisfies (an − bn ) n2 −1 ≤ (an − (−1)n bn ) n2 −1 ≤ (an + bn ) n2 −1 .
n n
By limn→∞ (an + bn ) n2 −1 = a, limn→∞ (an − bn ) n2 −1 = a, and the Sandwich rule, we get that
n
limn→∞ (an − (−1)n bn ) n2 −1 = a.

n
1.1.31 Without
√ loss √
of generality, we may
√ assume that a >
√ b > c >√0, then the sequence
√ an + bn + cn satisfies
n n n n n n n n n n n n n n
a = a < a + b + c √< a + a + a = 3a = 3a. By lim √ n→∞ 3 = 1 and the Sandwich
rule, we get that limn→∞ n an + bn + cn = a. So in general limn→∞ n an + bn + cn = max{a, b, c}

23
1.1.32 For a ≥ 1,
√ a−1 a−1 a−1
0< n
a−1= √ √ √ ≤ =
( n a)n−1 + ( n a)n−2 + · · · + n a + 1 1 + 1 + ··· + 1 + 1 n
a−1 √ √
By limn→∞ n = 0 and the Sandwich rule, limn→∞ n a − 1 = 0, that is limn→∞ n a = 1.
1
1.1.33 (a) First assume 0 < a < 1 and we write a = 1+b where b > 0. And

n2 n2 n2
0 < n2 an = = <
(1 + b)n 1 + nb + n(n−1) 2
b + n(n−1)(n−2) 3
b + · · · + bn n(n−1)(n−2) 3
b
2 6 6

n2
By limn→∞ n(n−1)(n−2) 3 = 0 and the Sandwich rule, we get that limn→∞ n2 an = 0.
6 b
If −1 < a < 0, then 0 < |a| < 1 and limn→∞ |n2 an | = limn→∞ n2 |a|n = 0. By Ex 1.1.11, we get
limn→∞ n2 an = 0.
√ √
(b) If 0 < a < 1, then we also have 0 < a < 1. So limn→∞ n( a)n = 0 by the given condition. So
√ √ √ √
lim n2 an = lim n( a)n · n( a)n = lim n( a)n · lim n( a)n = 0
n→∞ n→∞ n→∞ n→∞

If −1 < a < 0, then 0 < |a| < 1 and limn→∞ |n2 an | = limn→∞ n2 |a|n = 0. By Ex 1.1.11, we get
limn→∞ n2 an = 0.
1 1
1.1.34 If 0 < a < 1, then we also have 0 < a 6 < 1. So limn→∞ n(a 6 )n = 0 by the given condition. So

1
 1
6
lim n6 an = lim (n(a 6 )n )6 = lim n(a 6 )n =0
n→∞ n→∞ n→∞

Then 0 < n5.4 an < n6 an , by the Sandwich rule we have that limn→∞ n5.4 an = 0.
If −1 < a < 0, then 0 < |a| < 1 and limn→∞ |n5.4 an | = limn→∞ n5.4 |a|n = 0. By Ex 1.1.11, we get
limn→∞ n5.4 an = 0.
1.1.35 • If a = 5.4, then for n ≥ 6, we have

(5.4)n 5.4 · 5.4 · 5.4 · 5.4 · 5.4 5.4 5.4 5.4 5.4 · 5.4 · 5.4 · 5.4 · 5.4 5.4 5.46 1
0< = · · ··· ≤ · = ·
n! 1·2·3·4·5 6 7 n 1·2·3·4·5 n 5! n
5.46 1 (5.4)n
By limn→∞ 5! · n = 0 and the Sandwich rule, we get that limn→∞ n! = 0.
• If a = −5.4, then
(−5.4)n n n

lim
= lim (5.4) = 0 ⇒ lim (−5.4) = 0
n→∞ n! n→∞ n! n→∞ n!

by Ex 1.1.11.
1.1.36 • If a = 5.4, then for n ≥ 30, we have

(5.4)n 5.430 5.4 5.4 5.430 5.4


0< √ = √ · √ ··· √ ≤ √ ·√
n! 30! 31 n 30! n
5.430 (5.4)n
By limn→∞ √ · √1 = 0 and the Sandwich rule, we get that limn→∞ √ = 0.
30! n n!
• If a = −5.4, then
(−5.4)n n
(−5.4)n

lim √
= lim (5.4)
√ = 0 ⇒ lim √ =0
n→∞ n! n→∞ n! n→∞ n!
by Ex 1.1.11.

24
n!
1.1.37 The sequence satisfies
(2n)(2n − 1) · · · (n + 1)
n! 1
0< <
(2n)(2n − 1) · · · (n + 1) n+1
1 n!
By limn→∞ n+1 = 0 and the Sandwich rule, we have that limn→∞ (2n)(2n−1)···(n+1) = 0. So

n!an n! an n! an
lim = lim · = lim · lim =0·0=0
n→∞ (2n)! n→∞ (2n)(2n − 1) · · · (n + 1) n! n→∞ (2n)(2n − 1) · · · (n + 1) n→∞ n!

1.1.38 1
 n  n
n+1 1 1 1 1
lim = lim n + n = lim n + lim n = 0 + 0 = 0
n→∞ 2n n→∞ 2 2 n→∞ 2 n→∞ 2

2
 2
n2 1
lim n = lim n2 =0
n→∞ 2 n→∞ 2
by Ex 1.1.33.
3
√  √ 99
lim n99 (0.99)n = lim (n( 0.99)n )99 = 0.99)n ) = 09 9 = 0
99 99
lim (n(
n→∞ n→∞ n→∞

by Eg 1.1.12 and the Arithmetic rule.


(n2 +1)1001
4 The sequence 1.001n−2 satisfies

(n2 + 1)1001 (n2 + n2 )1001 n2002


0< n−2
< n−2
= 21001 · 1.0012 ·
1.001 1.001 1.001n
n n2002
By lim  n = 0, the Arithmetic rule gives us lim 21001 · 1.0012 · . So by the
n→∞
(1.001)
1
2002
n→∞ 1.001n
(n2 + 1)1001
Sandwich rule, lim = 0.
n→∞ 1.001n−2
5
  n  n   n  n
n + 2n 1 2 1 2
lim n
= lim n · + = lim n · + lim =0+0=0
n→∞ 3 n→∞ 3 3 n→∞ 3 n→∞ 3

6
  n  n   n n
n2n + (−3)n

1 3 1 3
lim = lim n · + − = lim n · + lim − =0
n→∞ 4n n→∞ 2 4 n→∞ 2 n→∞ 4
n3n
7 The sequence (1+2n )2 satisfies
 n
n3n n3n 3
0< n 2
< n 2 =n·
(1 + 2 ) (2 ) 4
3 n n3n

by limn→∞ n · 4 = 0 and the Sandwich rule, we get that limn→∞ (1+2n )2 = 0.
8
5 n
n 5 n
n
− 6n · 96 − limn→∞ 6n · 69
 
5n − n6n+1 9 limn→∞ 9 0−0
lim 2n−1 = lim 8 n
= n = 1 =0
− 23n+1 1
limn→∞ 13 − limn→∞ 2 · 89

3 −0
n→∞ 3
3 −2· 9
n→∞

25
1.1.39

1.2.3.4. Consider these 4 sub-questions together, we may assume that p > 0, |a| < 1.
1 If 0 < a < 1, then the sequence np an satisfies n[p] an < np an < n[p]+1 an . By limn→∞ n[p] an =
 1
[p]  1
[p]
limn→∞ n(a [p] )n = limn→∞ n(a [p] )n = 0[p] = 0 (Arithmetic rule), and limn→∞ n[p]+1 an =
 1
[p]+1  1
[p]+1
limn→∞ n(a [p]+1 )n = limn→∞ n(a [p] )n = 0[p]+1 = 0 (Arithmetic rule), the Sand-
wich rule gives us limn→∞ np an = 0.
If −1 < a < 0, then limn→∞ |np an | = limn→∞ np |a|n = 0. By Ex 1.1.11, limn→∞ np an = 0.
So limn→∞ np an = 0.
n
2 limn→∞ anp = limn→∞ n1p · an = 0.
p
3 Because ann > np , and np gets larger and larger as n approaches infinity. This sequence does
not have a limit.
4 Because limn→∞ np an = 0. Then for any k > 0, when n is sufficiently big, we have np an < k1 .
That is np1an > k. So limn→∞ np1an does not exist.
np np np np
5 If p > 0, the sequence n! satisfies 0 < <
n·(n−1)·(n−2)···(n−[p]) . By limn→∞ n·(n−1)·(n−2)···(n−[p])
n! =
p
0, and the Sandwich rule, we get that limn→∞ nn! = 0.
p
If p < 0, then −p > 0. So limn→∞ nn! = limn→∞ n−p 1 1
· n! = limn→∞ n−p 1 1
· limn→∞ n! =0·0=0
6
np an np an−[p]−1
lim = lim · · a[p]+1
n→∞ n! n→∞ n · (n − 1) · (n − 2) · · · (n − [p]) (n − [p] − 1)!

np an−[p]−1
=a[p]+1 · lim · lim
n→∞ n · (n − 1) · (n − 2) · · · (n − [p]) n→∞ (n − [p] − 1)!

=0

np np ( a1 )n
lim = =0
n→∞ n!an n!
by the previous question.
np np np
8 For sufficiently big n, 0 < √ < p < [p]+1
. By
n! n(n − 1)(n − 2) · · · (n − 2[p] − 1) (n − 2[p] − 1)
np np
limn→∞ (n−2[p]−1) [p]+1 = 0 and the Sandwich rule, we get that limn→∞

n!
= 0.
9 For sufficiently big n,

np a n np an−2[p]−2
0 < √ <p ·p · a2[p]+2
n! n(n − 1)(n − 2) · · · (n − 2[p] − 1) (n − 2[p] − 2)!
2
np a a([a ]+1)
< [p]+1
·p ·p · a2[p]+2
(n − 2[p] − 1) n − 2[p] − 2 ([a2 ] + 1)!
np a
by limn→∞ (n−2[p]−1)[p]+1
= 0, limn→∞ √ = 0 and the Sandwich rule, we get that
n−2[p]−2

np an
lim √ = 0.
n→∞ n!

26
10 For sufficiently big n,

np an np an−3[p]−3
0< √3
<p · p · a3[p]+3
n! 3
n(n − 1)(n − 2) · · · (n − 3[p] − 2) 3
(n − 3[p] − 3)!
3
np a a([a ]+1)
< [p]+1
·p · p · a3[p]+3
(n − 3[p] − 2) n − 3[p] − 3 3
([a3 ] + 1)!
np a
by limn→∞ (n−3[p]−2)[p]+1
= 0, limn→∞ √
3
= 0 and the Sandwich rule, we get that
n−3[p]−3

np an
lim √ = 0.
n→∞ 3 n!

11 For sufficiently big n, we have


n!an a
0< <
(2n)! 2n
a n!
By limn→∞ 2n = 0 and the Sandwich rule, we get that limn→∞ nn (2n)! = 0.
12 The sequence satisfies
n!np an n!np an np an
0< < 2
=
(2n)! (n!) n!
np an n!np an
By limn→∞ n! = 0 (Ex 1.1.39(6)), and the Sandwich rule, we get that limn→∞ (2n)! = 0.

1.1.40 1
n2 + 3n + 5n n2 3n 5n n2 5n
 
3n
lim lim + + = lim + lim + lim =0
n→∞ = n→∞ n! n! n! n→∞ n! n→∞ n! n→∞ n!

2
2 n 2 n
n 3n 5
n2 + 3n + 5n n! + n! + n! limn→∞ nn! + limn→∞ 3n 5
n! + limn→∞ n!
lim = lim 2 n = 2 n =0
n→∞ n! − n2 + 2n n→∞ 1 − n + 2 limn→∞ 1 − limn→∞ nn! + limn→∞ 2n!
n! n!

3
n2 + n3n + 5! n2 n3n n2 n3n
 
5! 5!
lim = lim + + = lim + lim + lim =0
n→∞ n! n→∞ n! n! n! n→∞ n! n→∞ n! n→∞ n!

4
n2 3n+5 + 5 · (n − 1)! 2 n
n2 3n
 
5 n 3 5 5
lim = lim 3 · + = 35 · lim + lim =0
n→∞ (n + 1)! n→∞ n! n(n + 1) n→∞ n! n→∞ n(n + 1)

5
n2 1 n2 1
n2 + n! + (n − 1)! n! +1+ n limn→∞ n! + limn→∞ 1 + limn→∞ n
lim n = lim 3n 1 = 3n 1 = −1
n→∞ 3 − n! + (n − 1)! n→∞ −1+ limn→∞ − limn→∞ 1 + limn→∞
n! n n! n

6
2n( 12 )n n! 2( 34 )n n! 2n( 12 )n n! 2( 34 )n n!
2n n! + 3n (n − 1)! (2n)! + (2n)! limn→∞ (2n)! + limn→∞ (2n)!
lim n = lim 2n( 54 )n n!
= 2n( 54 )n n!
=0
n→∞ 4 (2n − 1)! − 5n n! n→∞
1 − (2n)! limn→∞ 1 − limn→∞ (2n)!

27
 n n
1.1.41 For any positive integer 0 < k ≤ n, we have that k(n − k) > n
2. So Πnk=1 k(n − k) > . That is
n √ 2
(n!)2 > n2 . So n n! > n2 . Then
p

r
1 2
0< √
n
<
n! n
q
2 1
By limn→∞ n = 0 and the Sandwich rule, we get that limn→∞ √
n
n!
= 0.

1.1.42 •
n! n(n − 1)(n − 2) · · · 1 1
0< = ≤
nn n · n · n···n n
1 n!
By limn→∞ n = 0 and the Sandwich rule, we get that limn→∞ nn = 0.

(n!)2 n! n · (n − 1) · · · 1 1
0< = = <
(2n)! (2n)(2n − 1) · · · (n + 1) (2n)(2n − 1) · · · (n + 1) n+1

1 (n!)2
By limn→∞ n+1 = 0 and the Sandwich rule, we get that limn→∞ (2n)! = 0.

(n!)k (n!)k−1 1
0< = <
(kn)! (kn)(kn − 1) · · · (n + 1) n+1

1 (n!)2
By limn→∞ n+1 = 0 and the Sandwich rule, we get that limn→∞ (2n)! = 0.
l
1.1.44 • If limn→∞ xn = l > 0, then for sufficiently big n,< xn < 2l. So limn→∞ xn = 1 by Ex 1.1.25.
2
q
• Consider the sequence xn = n1 . We have that limn→∞ n1 = 0, while limn→∞ n n1 = limn→∞ √ 1
n n =

1.
n √
• Consider the sequence xn = 0.32
n . Then limn→∞ xn = 0 while limn→∞
n x 0.32
n = limn→∞ √n n = 0.32.

√ q n n n
1.1.45 1 n
5n − n4n = 5 n 1 − n 45 . Because limn→∞ n 54 = 0, for sufficiently big n, we have n 45 <
q q q
1 n 1 4 n n 1
n

2 . Then 5 2 < 5 1 − n 5 < 5 for sufficiently big n. By lim n→∞ 5 2 = 5 and the Sandwich

n n n
rule, we get that limn→∞ 5 − n4 = 5.

q n
n+2 1−n( 4 ) n
2 n+2 5n − n4n = 5 25
5
. Because limn→∞ n 54 = 0, for sufficiently big n, we have
q n
n+2 1−n( 4 )
q q q
n
n 45 < 12 . Then 5 n+2 50 1 1 1
 5 n+2 n+2
<5 25 < 5 25 for sufficiently big n. By lim n→∞ 5 50 =
q
1

5, limn→∞ 5 n+2 25 = 5 and the Sandwich rule, we get that limn→∞ n+2 5n − n4n = 5.
√ q n  n
3 n−2 5n − n4n = 5 n−2 25 · 1 − n 45 . Because limn→∞ n 45 = 0, for sufficiently big n, we
4 n
 1
q
n−2 25 n−2
q
4 n
  √
n−2
have n 5 < 2 . Then 5 2 < 5 25 · 1 − n 5 < 5 25 for sufficiently big n. By
q √ √
limn→∞ 5 n−2 25 2 = 5, limn→∞ 5
n−2
25 = 5 and the Sandwich rule, we get that limn→∞ n−2 5n − n4n =
5.
p q n n
4 n 5n − (−1)n n4n = 5 n 1 − n − 45 . Because limn→∞ n − 54 = 0, for sufficiently big n, we have
n q q n q q
− 12 < n − 45 < 12 . Then 5 n 12 < 5 n 1 − n 54 < 5 n 32 for sufficiently big n. By limn→∞ 5 n 12 =
q p
5, limn→∞ 5 n 32 = 5 and the Sandwich rule, we get that limn→∞ n 5n − (−1)n n4n = 5.

28
4 n 4 n
< 12 . So
 
5 Because limn→∞ n 5 = 0, for sufficiently big n, we have n 5
  n−1
1 n2 +1 1 n2 −n
− n+1 n−1 n−1
· 5 · 5 n2 +1 = ( · 5n ) n2 +1 < (5n − n4n ) n2 +1 < 5 n2 +1 < 5
2 2
n n+1 2n n 2n
And 5 2n2 < 5 n2 +1 < 5 n2 . By limn→∞ 5 2n2 = 1 and limn→∞ 5 n2 = 1 and the Sandwich rule, we
n+1
get that limn→∞ 5 n2 +1 = 1.
 n−1
Similarly, limn→∞ 12 n2 +1 = 1. So by the Sandwich rule,
n−1
lim (5n − n4n ) n2 +1 = 5
n→∞
n n
6 Because limn→∞ n − 45 = 0, for sufficiently big n, we have − 21 < n − 45 < 12 . So
  n−1   n−1   n+1   n+1
1 n2 +1 − nn+1 1 n n2 +1 n+(−1)n 3 n2 +1 3 n2 +1 n−1
n n n n
5· ·5 2 +1
= ·5 < (5 − (−1) n4 ) n +1 <
2
·5 =5· · 5 n2 +1
2 2 2 2
  n−1
1 n2 +1 − n+1
By similar argument to the previous question, we have lim = 1, lim 5 n2 +1 = 1,
n→∞ 2 n→∞
  n+1
3 n +1
2 n−1
lim = 1, lim 5 n2 +1 = 1. Then by the Sandwich rule, we get that
n→∞ 2 n→∞

n+(−1)n
lim (5n − (−1)n n4n ) n2 +1 =5
n→∞

1.1.46 1
r r s  n
n 1 n n 1 n 4
5 − n4n = 5 1 − n2
n n 5
4 n
n
= 0, for sufficiently big n, 12 < 1 − n2 54 < 1. So

Because limn→∞ n2 5
r r s  n
5 n 1 n n 1 n 4 5
√n
< n
5 − n4 = 5 1−n 2 < √
2n n n 5 n
n
5 5
By limn→∞ √
n
2n
= 5, limn→∞ √
n n = 5, and the Sandwich rule, we get that
r
n 1 n
lim 5 − n4n = 5
n→∞ n
2
r r s  n
n 1 n n n n 1 n 4
5 − (−1) n4 = 5 1 − n2 −
n n 5
n n
Because limn→∞ n2 − 45 = 0, for sufficiently big n, 12 < 1 − n2 − 54 < 32 . So
r r s  n r
5 n 1 n n 1 n 4 n 3

n
< n
5 − (−1) n4 = 5n 1−n −2 <5·
2n n n 5 2n
q
5 3
By limn→∞ √ n
2n
= 5, lim n→∞ 5 · n
2n = 5, and the Sandwich rule, we get that
r
n 1 n
lim 5 − (−1)n n4n = 5
n→∞ n

29
3
r r s  n
n+2 1 n n n+2 1 n+2 4
5 − n4 sin n = 5 1− n2 sin n
n 25n 5

4 n
n
= 0, for sufficiently big n, 12 < 1 − n2 54 sin n < 32 . So

Because limn→∞ n2 5
r r r r r
n+2 1 n+2 1 n+2 1 n n n+2 1 n+2 3
5· < 5 − n4 sin n < 5 ·
25n 2 n 25n 2
q q q
n+2 1 n+2 1 n+2 3
By limn→∞ 25n = 1, limn→∞ 2 = 1, limn→∞ 2 = 1 and the Sandwich rule, we get
that
r
1 n n+2
lim 5 − n4n sin n = 5
n→∞n
5 n 5 n
 
4 Because limn→∞ 16 = 0, then for sufficiently big n, 0 < 16 < 1. So

n2  n  2n−2
− nn+1 n−1 n−1
− n+1 n2 +1
16 · 16 2 +1
( − 1) n2 +1 < (n2 42n−1 − 5n ) n2 +1 < 16 · 16 n2 +1
4 2
Because
n n+1 2n
16 2n2 < 16 n2 +1 < 16 n2
 2  n−1
n n2 +1 n 2
1< −1 < (n2 ) n2 = n n
4
 n  2n−2
n2 +1 2
1< < nn
2
n+1
 2  n−1  2n−2
n2 +1
so limn→∞ 16 n2 +1 = 1, limn→∞ n4 − 1 = 1, limn→∞ n2 n2 +1 = 1 by the Sandwich rule.
And then
n−1
lim (n2 42n−1 − 5n ) n2 +1 = 16
n→∞

5
s  
n  n
r
n 3n−1 n 1 1 3
n2n + 4n+1 + =4· n +4+ ·
n 2 3n 4

1 n 1
n
· 43 = 0. For sufficiently big n, we have

Because limn→∞ n 2 = 0 and limn→∞ 3n
s  
n  n

1 1 3 n
4<4· n n +4+ · <4 6
2 3n 4

By limn→∞ 4 n 6 = 4 and the Sandwich rule, we get that
r
n 3n−1
lim n2n + 4n+1 + = 4.
n→∞ n
6
q
8 n−2
s 
 n−2 n−2
64n3
r
n−2 32n−1 n−2 8 81 9 + 27
n23n + =9· 64n + 2 =9· √
n2 9 3n n−2
n2

30
8 n−2
n−2
= 0, so for sufficiently big n, 0 < 64n3 98

Because limn→∞ 64n3 9 < 1. Then

n−2
r √
n−2
27 n−2 32n−1 28
√ <
9 · n−2 n23n + < 9 · √
n2 n2 n−2
n2

n−2
√ √
n−2
By limn→∞ 27 = 1, limn→∞ n−2 28 = 1, limn→∞ n2 = 1 and the Sandwich rule, we get
that
r
n−2 32n−1
n23n + =9
n2
7
r s  n−2
n−2 n − 1 2n−1 n−2 8 27(n − 1)
23n + 2 3 =9· 64 · +
n +1 9 n2 + 1
n−2
Because limn→∞ 64· 98 = 0 and limn→∞ 27(n−1) 13 26n
n2 +1 = 0. Then for sufficiently big n, n = n2 +n2 <
27(n−1) 8 n−2
+ 27(n−1)

n2 +1 < 64 · 9 n2 +1 < 1. So
r s  n−2
n−2 13 n−2 8 27(n − 1)
< 64 · + <1
n 9 n2 + 1
q q n−2 27(n−1)
By limn→∞ n−2 13n = 1 and the Sandwich rule, we have that lim n→∞
n−2
64 · 89 + n2 +1 = 1.
That is
r
n − 1 2n−1
lim n−2 23n + 2 3 =9
n→∞ n +1
8
 n−1  n  n−1
32n−1
 2
 n2
3n
n 1
1− n 8 1
n2 + =9 · n· + 2
n2 9 3n
8 n 1 1 8 n 1
 
Because limn→∞ n · 9 = 0 and limn→∞ 3n2 = 0, for sufficiently big n, 3n2 < n· 9 + 3n2 < 1,
so
 n1 − 1  n−1
32n−1
  2
1 1 n2
3n
n 1
9 1 ·< n2 + <9 1
9 n 3n2 n2 9n
1
 n1 1
 1 1
n2
By limn→∞ 3n2 = 1, limn→∞ 3n2 = 1, limn→∞ 9 n = 1 and the Sandwich rule
 n−1
32n−1
 2n
lim n23n + =9
n→∞ n2
9
1  n  n12
32n−1
  
n2 1 8 1
3n
n2 + =9 · n·
n + 2
n2 9 3n
8 n 1 1 8 n 1
 
Because limn→∞ n · 9 = 0 and limn→∞ 3n2 = 0, for sufficiently big n, 3n2 < n· 9 + 3n2 < 1,
so
1 1
32n−1
   
1 1 n2 n2 1
3n
9 · n < n2 + < 9n
3n2 n2
1 1
1

n2
By limn→∞ 3n2 = 1, limn→∞ 9 n = 1 and the Sandwich rule
1
32n−1
 
n2
lim n23n + =1
n→∞ n2

31
1.1.47 1
s  n
p
n n+1 n n b
a +b =a a+
a

b n
n
= 0, for sufficiently big n, 0 < ab < 1. So

Because limn→∞ a
√ p
n

a n a < an+1 + bn < a n a + 1
√ √
By limn→∞ a n a = a, limn→∞ a n a + 1 = a and the Sandwich rule,
p
n
lim an+1 + bn = a
n→∞

2
s  n
p
n n b
an+1 + (−1)n bn =a a+ −
a
n n
Because limn→∞ − ab = 0, for sufficiently big n, − 12 a < − ab < 12 a. So
r r
n 1 n 3
p
n
a a< a n+1 n n
+ (−1) b < a a
2 2
q q
By limn→∞ a n 12 a = a, limn→∞ a n 32 a = a and the Sandwich rule,
p
n
lim an+1 + (−1)n bn = a
n→∞

3
s  n−2
p
n−2 n−2 b
an+1 + (−1)n bn =a a3 + b2 −
a
n−2 n−2
Because limn→∞ − ab = 0, for sufficiently big n, − 2b12 a3 < − ab < 1 3
2b2 a . So
r r
n−2 1 3 n−2 3 3
p
a a < n−2 an+1 + (−1)n bn < a a
2 2
q q
By limn→∞ a n−2 12 a3 = a, limn→∞ a n−2 32 a3 = a and the Sandwich rule,

n−2
p
lim an+1 + (−1)n bn = a
n→∞

4
s  n

n b
4an − 5bn = a n 4 − 5 ·
a

b n b n
 
Because limn→∞ 5 · a = 0, then for sufficiently big n, 0 < 5 · a < 3. So

n

n
a < 4an − 5bn < a 4

By limn→∞ a n 4 = a, and the Sandwich rule

n
lim 4an − 5bn = a
n→∞

32
5 • If a > b2 , then
s  2 n
p
n b
n
4a + 5b2n+1 = a n 4 + 5b ·
a
 2 n  2 n
b b
Because limn→∞ 5b · a = 0, for sufficiently big n, 0 < 5b · a < 1. Then
p
n

n
a< 4an + 5b2n+1 < a 5

By limn→∞ a n 5 = a and the Sandwich rule
p
n
lim 4an + 5b2n+1 = a
n→∞

• If a < b2 , then
r  
p
n a n 2
+ =b n 4 2
4an 5b2n+1
+ 5b
b
n n
Because limn→∞ 4 ba2 = 0, for sufficiently big n, 0 < 4 ba2 < 1. So
p √
b2 < 4an + 5b2n+1 < b2
n n
5b + 1

2 n
By limn→∞ b 5b + 1 = b2 and the Sandwich rule
p
4an + 5b2n+1 = b2
n
lim
n→∞

n n
√ √n
6 • If
√ 0 < b < 1, then lim
√ n→∞ b = 0. For
√ sufficiently big n, 0 < b < 1. So n
a < √ a + bn <
n
a + 1. By limn→∞ n a = 1, limn→∞ a + 1 = 1, and the Sandwich rule, limn→∞ n a + bn =
n

1.

• If b > 1, then n a + bn = b n 1 + ban . Because limn→∞ ban = 0, for sufficiently big n, 0 < ban < 1.
p
√ √ √ √
So b < n a + bn < b n 2. By limn→∞ n 2 = 1 and the Sandwich rule, limn→∞ n a + bn = b.

7 • If
√ 0 < b < √ 1, then limn→∞ bn = √0. For sufficiently√ big n, 0 < bn < 1. So n an <
n
an + bn√< an + 1. By limn→∞ n an = 1, limn→∞ an + 1 = 1, and the Sandwich rule,
n n

limn→∞ n an + bn = 1.

• If b > 1, then n an + bn = b n 1 + an limn→∞ an 0 < an
p
√ √ bn . Because
√ bn = 0, for sufficiently big n,
√ bn <
n n n n
1. So b < an + bn < b 2. By limn→∞ 2 = 1 and the Sandwich rule, limn→∞ an + bn =
b.
8
s  n−2
p
n−2 n−2 b
nan + (n2 + 1)bn =a na2 + (n2 + 1)b2
a

b n−2 b n−2
 
Because limn→∞ (n2 + 1)b2 a = 0, for sufficiently big n, 0 < (n2 + 1)b2 a < 1. So

n−2 n−2
p p
n−2
a na2 < nan + (n2 + 1)bn < a na2 + 1

n−2

By limn→∞ a na2 = a, limn→∞ a n−2 na2 + 1 = a and the Sandwich rule,
p
lim n−2 nan + (n2 + 1)bn = a
n→∞

33
9
r s  n+2
n+2 1 n n+1 n+2 1 n b
a + nb =a +
n a2 n b a

b n+2
n
n+2
= 0, for sufficiently big n, 0 < nb ab

Because limn→∞ b a < 1. So
r r r
n+2 1 n+2 1 n n+2 1
a < a + nbn+1 < a +1
a2 n n a2 n
q q
By limn→∞ a n+2 a21n = a, limn→∞ a n+2 a21n + 1 = a, and the Sandwich rule
r
n+2 1 n
lim a + nbn+1 = a
n→∞ n
10
 n  nn+1
2 +1

n n
n+1 b n−1
(a + b ) =a·a n2 +11+ n2 +1
a
b n
n
= 0, for sufficiently big n, 0 < ab < 1. So

Because limn→∞ a
n−1 n+1 n−1 n+1
a · a n2 +1 < (an + bn ) n2 +1 < a · a n2 +1 · 2 n2 +1

By
n−1 1
1 < a n2 +1 < a n
n+1 2
1 < 2 n2 +1 < 2 n
n−1 n+1
limn→∞ a n2 +1 = 1 and limn→∞ 2 n2 +1 = 1. So the Sandwich rule gives us
n+1
lim (an + bn ) n2 +1 = a
n→∞

11
 n  n2n−1 
n n
n b 1
((n + 1)a + b ) =a·a n2 −1· n+1+ n2 −1
a
b n n
= 0, for sufficiently big n, 0 < ab < 1. So
 
Because limn→∞ a
1 n n 1 n
a · a n2 −1 · (n + 1) n2 −1 < ((n + 1)an + bn ) n2 −1 < a · a n2 −1 · (n + 2) n2 −1

By
1 2
1 < a n2 −1 < a n2
n 2
1 < (n + 1) n2 −1 < (n + 1) n2
n 2
1 < (n + 2) n2 −1 < (n + 2) n2
1 n n
limn→∞ a n2 −1 = 1, limn→∞ (n + 1) n2 −1 = 1 and limn→∞ (n + 2) n2 −1 = 1. Then the Sandwich rule
gives us
n
lim ((n + 1)an + bn ) n2 −1 = a
n→∞

34
12
  n  n21−1
n n
1 n b
(a + b ) =a
n2 −1 1+n2 −1
a
b n
n
= 0, for sufficiently big n, 0 < ab < 1. So

Because limn→∞ a

2 1 1 1
a n < (an + bn ) n2 −1 < a n 2 n2 −1
2 1 1
By limn→∞ a n = 1, limn→∞ a n = 1, limn→∞ 2 n2 −1 = 1, and the Sandwich rule,
1
lim (an + bn ) n2 −1 = 1
n→∞

13
n
  n  (−1)
(−1)n (−1)n n b n2 −1
n n n
(a + (−1) b ) n2 −1 =a n2 −1 1+ −
a
n
n  n   (−1)
b n n2 −1
Because limn→∞ − ab = 0, for sufficiently big n, −1 < − ab
< 1, then 1 < 1 + − a <
(−1)n
1 1
 
n n2 −1
2 n2 −1 . By limn→∞ 2 n2 −1 = 1 and the Sandwich rule, limn→∞ 1 + − ab = 1.
And
−n (−1)n n n
2 2
a− n = a n
2 − 1 n2 2 − 1 n2
2 <a n2 −1 < an 2 = an
2
By limn→∞ a n = 1 and the Sandwich rule, we have that
(−1)n n
lim a n2 −1 =1
n→∞

So
(−1)n
lim (an + (−1)n bn ) n2 −1 = 1
n→∞

1.1.48 1 Without loss of generality, we may assume that a > b > c > 0. Then
s  n
p b  c n
n 2 n n n n 2
n a + nb + 2c = a n + n · +2·
a a

b n c n b n
  
Because limn→∞ n · a = 0 and limn→∞ 2 · a = 0, for sufficiently big n, 0 < a < 1,
n
0 < 2 · ac < 1. So
p
n
p
n
a< n2 an + nbn + 2cn < a n2 + 2
√ √
By limn→∞ n n2 + 1 = 1 and the Sandwich rule, limn→∞ n n2 an + nbn + 2cn = a.

In conclusion limn→∞ n n2 an + nbn + 2cn = max{a, b, c}.
p √
2 The limit of n an (bn + 1) + ncn is equal to the limit of n dn + an + ncn where d = ab. Without
loss of generality, we may assume that a > d > c > 0, then
s  n
√n n d  c n
n n n
d + a + nc = a 1 + +n
a a

35
d n c n d n
  
Because limn→∞ a = 0 and limn→∞ n · a = 0, for sufficiently big n, 0 < a < 1, 0 <
n
n · ac < 1. So

n

n
a< dn + an + ncn < a 3

n
By limn→∞ 3 = 1 and the Sandwich rule,

n
lim dn + an + ncn = a
n→∞

In general
p
n
lim an (bn + 1) + ncn = max{a, ab, c}
n→∞

3 Without loss of generality, we may assume that a > b > c > 0, then
s  n
p b  c n
n
(n + sin n)a + b + n c = a n n + sin n +
n n 2 n + n2
a a
n n b n
Because limn→∞ ab = 0 and limn→∞ n2 · ac
  
= 0, for sufficiently big n, 0 < a < 1,
c n
2

0 < n · a < 1. So
p √
a < n (n + sin n)an + bn + n2 cn < a n n + 2

By limn→∞ n n + 2 = 1 and the Sandwich rule,
p
lim n (n + sin n)an + bn + ncn = a
n→∞

In general
p
n
lim (n + sin n)an + bn + ncn = max{a, b, c}
n→∞
p √
4 The limit of n an (n + bn (1 + ncn )) is equal to the limit of n nan + dn + nf n where d = ab, and
f = abc. Without loss of generality, we may assume that a > d > f > 0, then
s  n  n
p d f
n
nan + dn + nf n = a n n + +n
a a
n   n n
Because limn→∞ ad = 0 and limn→∞ n · fa = 0, for sufficiently big n, 0 < ad < 1,
 n
0 < n · fa < 1. So
p √
a < n nan + dn + nf n < a n n + 2

By limn→∞ n n + 2 = 1 and the Sandwich rule,

n
lim nan + dn + ncn = a
n→∞

In general
p
n
lim an (n + bn (1 + ncn )) = max{a, ab, abc}
n→∞

1.1.49
 
n p−1 n 1 1 1
p(n) = ap p + an−1 n + · · · + a1 n + a0 = p ap + ap−1 + · · · + a1 p−1 + p
n n n
1 p a 1 ap
Because limn→∞ nk
= 0 when k is a positive integer. So for sufficiently big n, − 2pap−k < nk
< 2pap−k .
So
 
1 1 1 1
ap + ap−1 + · · · + a1 p−1 + p > (ap − ap ) > 0
n n n 2
That is p(n) > 0 for sufficiently big n.

36
1.1.50 Without loss of generality, we may assume that a > b > c > 0, then
s  n
p b  c n
n
p(n)an + q(n)bn + r(n)cn = a n p(n) + q(n) + r(n)
a a
n n b n
Because limn→∞ q(n) ab = 0 and limn→∞ r(n) · ac

= 0, for sufficiently big n, 0 < q(n) a < 1,
n
0 < r(n) · ac < 1. So


p p p
a n p(n) < n p(n)an + q(n)bn + r(n)cn < a n p(n) + 2
Assume the leading term in p(n) is ap np , then by argument in Ex 1.1.49,
1
ap np < p(n) < p(n) + 2 < pap np
2
So
r
1 n
p p p p
aap np < a n p(n) < n p(n)an + q(n)bn + r(n)cn < a n p(n) + 2 < a n pap np
2
q √
By limn→∞ a n 12 ap np = a, limn→∞ a n pap np = a and the Sandwich rule
p
n
lim p(n)an + q(n)bn + r(n)cn = a
n→∞

In general
p
n
lim p(n)an + q(n)bn + r(n)cn = max{a, b, c}
n→∞

1.1.51 1
r

n p n sin n
anp + b sin n = n n a+b
np
Because limn→∞ b sin n sin n
np = 0, for sufficiently big n, −1 < b np < 1. So for sufficiently big n,
r
p
n
√n
√n n 3
p p p
an /2 < an − 1 < an + b sin n < anp
2
p q
By limn→∞ n anp /2 = 1 and limn→∞ n 32 anp = 1 and the Sandwich rule

n
lim anp + b sin n = 1
n→∞
q
bn
2 Without loss of generality, we may assume that p > q, then limn→∞ an p = 0. So for sufficiently big
bnq
n, anp < 1 or bnq < anp . Then
√n

n

n
2bnq < anp + bnq < 2anp
√ √
Because limn→∞ n 2bnq = 1 and limn→∞ n 2anp = 1 and the Sandwich rule,
√n
lim anp + bnq = 1
n→∞
q
bn
3 Without loss of generality, we may assume that p > q, then limn→∞ an p = 0. So for sufficiently big
q
bn q p
n, anp < 1 or bn < an . Then

n+2

n+2

n+2
2bnq < anp + bnq < 2anp
√ √
Because limn→∞ n+2 2bnq = 1 and limn→∞ n+2 2anp = 1 and the Sandwich rule,

n+2
lim anp + bnq = 1
n→∞

37
bnq
4 Without loss of generality, we may assume that p > q, then limn→∞ anp = 0. So for sufficiently big
bnq q p
n, anp < 1 or bn < an . Then


n−2

n−2 n−2

2bnq < anp + bnq < 2anp

n−2 n−2

Because limn→∞ 2bnq = 1 and limn→∞ 2anp = 1 and the Sandwich rule,
n−2

lim anp + bnq = 1
n→∞

bnq
5 Without loss of generality, we may assume that p > q, then limn→∞ anp = 0. So for sufficiently big
bnq q p
n, anp < 1 or bn < an . Then


n2

n2

n2
2bnq < anp + bnq < 2anp

n2

n2
Because limn→∞ 2bnq = 1 and limn→∞ 2anp = 1 and the Sandwich rule,

n2
lim anp + bnq = 1
n→∞

bnq
6 Without loss of generality, we may assume that p > q, then limn→∞ anp = 0. So for sufficiently big
bnq q p
n, anp < 1 or bn < an . Then

1 1 1
(2bnq ) n2 −1 < (anp + bnq ) n2 −1 < (2anp ) n2 −1
1 1
Because limn→∞ (2bnq ) n2 −1 = 1 and limn→∞ (2anp ) n2 −1 = 1 and the Sandwich rule,
1
lim (anp + bnq ) n2 −1 = 1
n→∞


xn
1.1.52 If
≤ c for 0 < c < 1, then
xn−1

xn xn−1 xn−2
· · · x2 |x1 | ≤ cn−1 |x1 |

0 ≤ |xn | = · ·
xn−1 xn−2 xn−3 x1

By limn→∞ cn−1 = 0 and the Sandwich rule, limn→∞ |xn | = 0. Then limn→∞ xn = 0 by Ex 1.1.11.
xn
1.1.53 If limn→∞ xn−1 = l and |l| < 1, without loss of generality we may assume that 0 < l < 1. Then for
xn l+1
sufficiently big n, (assume when n > N ), 0 < xn−1 < 2 . Then for n > N ,
 n−N  n  −N
xn xn xn−1 xN +1 l+1 l+1 l+1
0 < xn = · · ··· · xN < · xN = C · , C= xN
xn−1 xn−1 xn−2 xN 2 2 2
n
By limn→∞ l+1
2 = 0 and the Sandwich rule,

lim xn = 0
n→∞

1.1.54 1
xn (2n)(2n − 1)
lim = lim a = 4a
n→∞ xn−1 n→∞ n2

So if |a| < 41 , then limn→∞ xn = 0

38
2
xn n2
lim = lim a=0
n→∞ xn−1 n→∞ (3n)(3n − 1)(3n − 2)

So for any real number a, limn→∞ xn = 0


3
xn n3 a
lim = lim a=
n→∞ xn−1 n→∞ (3n)(3n − 1)(3n − 2) 27
So if |a| < 27, then limn→∞ xn = 0
4
p
xn (2n)(2n − 1)
lim = lim a = 2a
n→∞ xn−1 n→∞ n

So if |a| < 21 , then limn→∞ xn = 0


5
xn √
lim = lim na2n−1
n→∞ xn−1 n→∞

The limit equal to 0 if |a| < 1. So if |a| < 1, then limn→∞ xn = 0


6
xn a2n−1
lim = lim
n→∞ xn−1 n→∞ n
The limit equal to 0 if |a| < 1. So if |a| < 0, then limn→∞ xn = 0
7
xn a2n−1
lim = lim √
n→∞ xn−1 n→∞ n
The limit equal to 0 if |a| < 1. So if |a| < 1, then limn→∞ xn = 0
8
xn
lim = lim np a
n→∞ xn−1 n→∞

The limit equal to 0 if p < 0. And the limit does not exist if p > 0. So if p < 0, then limn→∞ xn = 0
9
xn a
lim = lim p
n→∞ xn−1 n→∞ n

The limit equal to 0 if p > 0. And the limit does not exist if p < 0. So if p > 0, then limn→∞ xn = 0
10
xn nq a
lim = lim p
n→∞ xn−1 n→∞ n (n − 1)q

The limit equal to 0 if p > 0. And the limit does not exist if p < 0. So if p > 0, then limn→∞ xn = 0
11
xn np a
lim = lim
n→∞ xn−1 n→∞ (2n) (2n − 1)q
q

If p < 2q, then the limit is always 0. If p = 2q, then the limit is 0 if |a| < 4q . So if p < 2q or (p = 2q
and |a| < 4q), then limn→∞ xn = 0

39
12
xn n5 np a
lim = lim
n→∞ xn−1 n→∞ (n − 1)5 (2n)q (2n − 1)q

If p < 2q, then the limit is always 0. If p = 2q, then the limit is 0 if |a| < 4q . So if p < 2q or (p = 2q
and |a| < 4q), then limn→∞ xn = 0
√ √ √ √
1.1.55 1 The sequence n! + 1 − n! − 1 is a subsequence of n + 1 − n − 1. And
√ √
(n + 1) − (n − 1)
limn − 1 = lim √
n+1− √ =0
n→∞ n→∞ n+1+ n−1
√ √
So the subsequence has the same limit. That is limn→∞ n! + 1 − n! − 1 = 0.
1 1 1 1
2 The sequence (n!) n! is a subsequence of n n . And limn→∞ n n = 1. So limN →∞ (n!) n! = 1.
1
3 The sequence ((n + 1)!) n! satisfies
1 1 1 1
(n!) n! ≤ ((n + 1)!) n! ≤ (n!) n! · (n + n) n
1 1
Because limn→∞ (n!) n! = 1 and limn→∞ (2n) n = 1, by the Arithmetic rule and the Sandwich rule,
1
limn→∞ ((n + 1)!) n! = 1.
1
4 The sequence ((n + (−1)n )!) n satisfies
1 1
1 ≤ ((n + (−1)n )!) n ≤ ((n + 1)!) n!
1
By limn→∞ ((n + 1)!) n! = 1 and the Sandwich rule, we have that
1
lim ((n + (−1)n )!) n = 1
n→∞

1
5 The sequence (n!) (n+1)! satisfies
1 1
1 ≤ (n!) (n+1)! ≤ (n!) n!
1
By limn→∞ (n!) n! = 1, and the Sandwich rule, we have that
1
lim (n!) (n+1)! = 1
n→∞

2 2 1 1
−1
6 The sequence (2n + 3n ) n2 is a subsequence of (2n−1 + 3n ) n . And
  n  n1
1 1 2 1
3 ≤ (2n−1 + 3n ) n = 3 · +1 ≤ 3 · 2n
2 3
1
Then by limn→∞ 3 · 2 n and the Sandwich rule, we have that
2 2 1
−1
lim (2n + 3n ) n2 = 3
n→∞

1
1.1.56 1 The even subsequence is a2n = 2 and the odd sequence is a2n+1 = 2. The two subsequences
converge to different limit so the original sequence does not converge.
1 1
2 The even subsequence is a2n = (2n) 2n and the odd subsequence is a2n+1 = (2n + 1)− 2n+1 =
1
1 . Both subsequences converge to 1 and the union of {a2n } and {a2n+1 } is the original
(2n+1) 2n+1
sequence {an }. So the original sequence {an } converges to 1, too.

40
1
3 The even subsequence is a2n = 2n and the odd sequence is a2n+1 = 2n+1 . So limn→∞ a2n goes to
infinity while limn→∞ a2n+1 = 0. The two limits do not equal to each other so the original sequence
does not converge.
2n+3
4 The even subsequence is a2n = 2n−2 . Its limit is 1. The odd subsequence is a2n+1 = −2n+22n+3 . Its
limit is −1. The two subsequences have different limits. So the original sequence does not converges.
2 2
(2n) 1 (2n+1)
5 The even subsequence is a2n = (2n) 3 −1 . Its limit is 2 . The odd subsequence is a2n+1 = − (2n+1)3 −1 .

Its limit is − 21 . The two subsequences have different limits. So the original sequence does not
converge.
√ √ √
6 The even subsequence is a2n = 2n( 2n + 1 − n). Its limit is 12 . The odd subsequence is
√ √ √
a2n+1 = 2n + 1( 2n − 2n + 1). Its limit is − 12 . The two subsequences have different limits. So
the original sequence does not converge.
1
7 The even subsequence is a2n = (22n + 32n ) 2n . Its limit is 3. The odd subsequence is a2n+1 =
1
(22n+1 + 32n+1 ) 2n+1 . Its limit is 3. The two subsequences have the same limit. And the union of
the odd subsequence and the even subsequence is the original sequence. So the limit of the original
sequence exists and it equals to 3.
1
8 The even subsequence is a2n = (22n + 32n ) 2n . Its limit is 3. The odd subsequence is a2n+1 =
1
1
(22n+1 + 32n+1 ) 2n+1 . Its limit is 2. The two subsequences have different limits. So the original
sequence does not converge.

9 Consider 3 subsequences: a3n = tan 3nπ 3 = tan nπ = 0, a3n+1 = tan (3n+1)π3 = tan π3 = 3.

a3n+2 = tan (3n+2)π
3 = tan 2π
3 = − 3. The subsequences have different limits implies that the
original sequence does not converge.

(6n+1)π
10 Consider two subsequences: a6n = sin 6nπ
3 = 0, and a6n+1 = − sin 3 = − sin π3 = − 23 . These
two subsequences converge to different limits. So the original sequence {an } does not converge.
11 Consider two subsequences: a12n = sin 12nπ
2 cos 3
12nπ
= 0 and a12n+1 = sin (12n+1)π
2 cos (12n+1)π
3 =
π π 1
sin 2 cos 3 = 2 . These two subsequences converge to different limits. So the original sequence {an }
does not converge.
(12n+1)π
12n sin 12nπ (12n+1) sin
12 Consider two subsequences: a12n = 3
12n cos 12nπ
= 0 and a12n+1 = 3
(12n+1)π =
2 +2 (12n+1) cos 2 +2

3
(12n+1)·
2 . The sequence {a12n+1 } diverges. These two subsequences don’t converge to the same
2

limits. So the original sequence {an } does not converge.


13 Using arithmetic rule, and the sandwich rule we can get the limit of this sequence is 1.
1.1.57 Assume 0 < a < π, For any positive integer k, the interval [kπ − π2 , kπ + π2 ] of length π contains the
following interval of length a
h a ai
[ak , bk ] = kπ − , kπ +
2 2
For even k, we have cos x ≥ cos a2 > 0 on [ak , bk ]. For odd k, we have cos x ≤ − cos a2 < 0 on [ak , bk ].
Since the arithmetic sequence a, 2a, 3a, · · · has increment a, which is the length of [ak , bk ], we must
have nk a ∈ [ak , bk ] for some natural number nk . Then cos(n2k a) is a subsequence of cos na satisfying
cos n2k a ≥ cos a2 > 0 and cos n2k+1 a is a subsequence satisfying cos n2k ≤ − cos a2 < 0. Therefore the
two subsequences cannot converge to the same limit. As a result, the sequence cos na diverges.
For general a, we have a = 2N π ± b for an integer N and b satisfying 0 < b < π. Then we have
cos na = ± cos nb. We have shown that cos nb diverges, so that cos na diverges.
So cos na converges if and only if a = 0.
1.1.58 If yn is bounded, then we may assume that m < yn < M . The sequence xn is the union of two
subsequences x0k and x00k satisfying all x0k ≥ 1 and all x00k ≤ 1. By Proposition 1.1.6, the assumption
limn→∞ xn = 1 implies that limk→∞ x0k = limk→∞ x00k = 1.

41
m y M m y M
x0k ≥ 1 implies x0 k ≤ x0 kn ≤ x0 k . And x00k ≤ 1 implies x00 k ≥ x00 kn ≥ x00 k . By Example 1.1.21
m
lim x0 k = 1
k→∞
M
lim x0 k = 1
k→∞
m
lim x00 k = 1
k→∞
M
lim x00 k = 1
k→∞

y y
and the Sandwich rule, limn→∞ x0 kn = limn→∞ x00 kn = 1. Since the sequence xynn is the union of two
y y
subsequences x0 kn and x00 kn , by Proposition 1.1.6 again, we get limn→∞ xynn = 1.
p
1.1.59 If limn→∞ xn = l > 0, then by Arithmetic rule, limn→∞ xln = 1. Then by Example 1.1.21, limn→∞ xln =
1. Use Arithmetic rule again, we have that limn→∞ xpn = lp .
1.2.1
2 2
n − 1 − n2 − 1

n − 1 2 2
n2 + 1 − 1 = n2 + 1 ≤ n
=
n2 + 1

for sufficiently big n.


Then for any  > 0, take N = 2 . Then for any n > N , we have
2
n − 1 2
n2 + 1 − 1 ≤ n < 

n2 −1
Then by the definition of limit, limn→∞ n2 +1 = 1.
1.2.2 The range of the height of the wall is [2.7, 3.3] and the range of the width of the wall is [5.4, 6.6]. So the
area of the wall is in the range of [14.58, 21.78]. So the cost is in the range of [196.83, 294.03].
1.2.5 5x − 3y + 4z = 5(x + 2) − 3(y − 3) + 4(z − 5) + 1. If we want 5x − 3y + 4z to be within ± of 1, we may
set |x + 2| ≤ 3 , |y − 3| ≤ 3 , |z − 5| ≤ 3 , respectively.
1.2.6 xy − 4 = xy − 2y + 2y − 4 = (x − 2)(y − 2). So we could take the tolerance of x, y near 2 and 2 to be
within  and 1 respectively. Then xy could be within ± of 4.
1.2.7 x2 − 4 = (x + 2)(x − 2). We could take the tolerance of x near 2 to be min{ 21 , 6 }, then x2 is within ±
of 4.

1.2.8 x1 − 0.5 = |x−2|


1
|2x| . So we can make sure x is within ±0.1 of 0.5, by taking the tolerance to be 0.2. Then
the percentage tolerance is 0.22 = 0.1 = 10%.

1.2.9 (3) For any  > 0, take N = max{2|b|, 2|a−b|


 }, then for any n > N , we have that

n + a |a − b| |a − b| |a − b| 2|a − b|
n + b − 1 = |n + b| ≤ n − |b| ≤ n − 1 n = ≤

n 2

n+a
So limn→∞ n+b = 1.
1
(4) For any  > 0, take N = max{12, 8 }, then for any n > N , we have that 9n2 − 12n + 3 > 8n2
(which happens when n > 12 — by solving quadratic inequality)
2
2n − 3n + 2 2 n−4 n 1
3n2 − 4n + 1 − 3 = 9n2 − 12n + 3 < 8n2 = 8n < .

2n2 −3n+2
So limn→∞ 3n2 −4n+1 = 32 .

42
(5) For any  > 0, take N = max{36, 100
2 }, then for any n > N , we have that

n+2
= √ 5 5 10

√ − 1 ≤√ √ = √ ≤
n−3 n−3 n − 12 n n

So limn→∞ √n+2 = 1.
n−3
2
(12) For any  > 0, take N = max{2|a|, 2(a−b)
2 }, then for any n > N , we have that

√ √

a − b |a − b| |a − b| 2|a − b|
| n + a − n + b − 0| = √ √ ≤ √ ≤q = √ ≤
n + a + n + b n+a 1
n − 2n n

  p1 1
2a2
1.2.10(1) For any  > 0, take N = max{ 2 , (2|b|) p }, then for any n > N , we have that

√ a
|a| 2|a|
np + b − 0 ≤ q p 1 p = √np ≤ 

n − 2n

So limn→∞ √ a = 0.
np +b

1.2.11 (1) For any  > 0, take N = max{ 14 , 2}, then n > N implies
√ n−1 n−1 1
| n n − 1| = n−1 n−2 ≤ n n−1 ≤ 1 <
n n +n n + ··· + 1 2 · (2n 2n ) n4

 1
 1.6
25 ·39
(4) For any  > 0, take N = max{12,  }, then n > N implies
5.4 n
n5.4 3n−6

n 3 6

n! − 0 ≤
n(n − 1)(n − 2)(n − 3)(n − 4)(n − 5) · (n − 6)! · 3
n5.4 3 3 3 3 3
≤ · · · · · · · · 36
n(n − 12 n)5 n − 6 n − 7 3 2 1
25 39 25 · 39
≤ · ≤
n0.6 2(n − 6) n0.6 n
5 9
2 ·3
≤ 1.6 ≤ 
n

1.2.12 If limn→∞ xn = l, then for any 0 > 0, there exists N0 such that n > N0 implies |xn − l| < 0 .
Then for any  > 0, choose 0 = , then take N = N0 , we have that n > N implies n > N0 , also
|xn | − |l| ≤ |xn − l| and |l| − |xn | ≤ |xn − l|.
That is ||xn | − |l|| ≤ |xn − l| < 0 = .
Then we have limn→∞ |xn | = |l|.
1.2.13 • If limn→∞ |xn −l| = 0, then for any 0 > 0, there exists N0 such that n > N0 implies ||xn −l|−0| < 0 .
That is |xn − l| < .
Then for any  > 0, choose 0 = , and take N = N0 , then n > N implies n > N0 , that is
|xn − l| < 0 = . So limn→∞ xn = l.
• If limn→∞ xn = l, then for any 0 > 0, there exists N0 such that n > N0 implies |xn − l| < 0 .
Then for any  > 0, choose 0 = , and take N = N0 , then n > N implies n > N0 , that is
||xn − l| − 0| = |xn − l| < 0 = . So limn→∞ |xn − l| = 0.
In conclusion, limn→∞ |xn − l| = 0 if and only if limn→∞ xn = l.

43
1.2.16 Given that limn→∞ xn = 0 and xn ≥ 0 for sufficiently big n, for any 0 > 0, there exists N0 such that
n > N0 implies |xn − 0| < 0 . That is 0 ≤ xn < 0 .
1
Then for any  > 0, choose 0 =  p , then take N = N0 . Then n > N implies 0 < xn < 0 . So
0 < xpn < p0 = . That is |xpn − 0| < . So we can conclude that limn→∞ xpn = 0.
1
1.3.3 (1) xn+1 − xn = (n+1)3 > 0. So the sequence xn is an increasing sequence. And

1 1 1 1
xn ≤ 1 + + + ··· + =2− <2
2·3 3·4 n(n + 1) n+1

So xn is an increasing bounded sequence. So it must converge.


1
(3) xn+1 − xn = (2n+1)(2n+3) > 0. So the sequence xn is an increasing sequence. And
   
1 1 1 1 1 1 1 1 1
xn = 1− + − + ··· + − = 1− <
2 3 3 5 2n − 1 2n + 1 2 2n + 1 2

So xn is an increasing bounded sequence. So it must converge.


x2n +1
1.3.6 [Remark: It seems the problem has changed. Please ignore.] Because xn+1 − xn = 2 − xn =
(xn +1)2
2 ≥ 0. So the sequence xn is an increasing sequence.

(5) If x1 = 2, we claim xn > n. We prove this claim by using Mathematical Induction.


• When n = 1, x1 = 2 > 1.
x21 +1 5
• When n = 2, x2 = 2 = 2 > 2.
x22 +1 29
• When n = 3, x3 = 2 = 8 > 3.
x23 +1 10
• When n = 4, x4 = 2 > 2 > 4.
x2k +1 k2 +1
• Assume that n = k ≥ 4, xk > k. Then when n = k + 1, xk+1 = 2 > 2 > 2k > k + 1.
So by MI, we can conclude that xn > n for all positive integer n. Then limn→∞ xn > limn→∞ n.
So the limit goes to infinity. Or we say the limit does not exist.
(6) If x1 = 0, we claim that 0 ≥ xn < 1. By x1 = 0 and xn is increasing, we know xn ≥ 0 for all n.
Then
• When n = 1, x1 = 0 < 1.
• Assume that when n = k, xk < 1, then when n = k + 1,

x2n + 1 1+1
xk+1 = < =1
2 2
So xk+1 < 1.
Then by MI, we can conclude that xn < 1 for all positive integer n. That is the sequence xn is an
increasing bounded sequence. So it must converge.
Assume limn→∞ xn = l. Then take the limit on both sides of the induction equation, we have that

l2 + 1
l=
2
Solving this equation gives the solution l = 1. So limn→∞ xn = 1.
1.3.7(2) • By previous question, we already know that if |x1 | < 1, then the sequence will be an increasing
sequence bounded by [0, 1]. So it must converge. And assume the limit is l. Taking the limit of
2
both sides of the inductive equation gives us l = l 2+1 . That is l = 1. So limn→∞ xn = 1.
• If |x1 | = 1, then by the inductive equation, we have xn = 1 for all n. Then limn→∞ xn = 1.

44
x21 +1 (1+a)2 +1 2+2a+a2
• If |x1 | > 1, assume that |x1 | = 1 + a, where a > 0. Then x2 = 2 > 2 = 2 > 1 + a.
x2k +1 2
If xk > 1 + a, then xk+1 = 2 > (1+a) 2
+1
> 1 + a. So by MI, xn > 1 + a for all natural number
n. While if the limit of xn exists, we must have limn→∞ xn = 1 by the calculation in the previous
part. This contradicts with the fact that xn > 1 + a. So the limit does not exist, when |x1 | > 1.
1.3.8(3) • If the limit of the sequence exists, assume the limit is l, then we must have

√ −1 ± 1 + 4a
l = a − l ⇒ l1,2 =
2

Because the sequence is a non-negative sequence, the limit can only be −1+ 21+4a .
√ √
• Consider
√ the situation where xn+1 = a − xn and xn = xn+2 = a − xn+1 . That is xn+1 =

a − xn and xn = a − xn+1 .
That’s equivalent to the system
( √
y = a−x

x= a−y

Solution to the system is also the solution to the equation x4 − 2ax2 + x + a2 − a = 0.


Clearly l1 and l2 are two solutions to the equation. So the left hand side of the equation could be
factorized as

(x − l1 )(x − l2 )(x2 + bx + c) = 0 ⇔ (x2 + x − a)(x2 + bx + c) = 0

So we must have

x4 − 2ax2 + x + a2 − a = x4 + (1 + b)x3 + (−a + b + c)x2 + (−ab + 1)x − ac

Equating corresponding coefficients, we have that b = −1 and c = 1 − a. Then the quartic equation

is (x2 + x − a)(x2 − x + 1 − a) = 0. Solving this equation, we get the solutions are l1,2 = −1± 2 1+4a

or m1,2 = 1± 24a−3 .
– In case m1 , m2 > 0:
That is 34 < a < 1. We have that l2 < 0 < m2 < l1 < m1 .
If xn > m1 , then

s
√ 1 + 4a − 3
xn+1 = a − xn < a −
2

s
4a − 2 − 2 4a − 3
=
4

s
1 − 2 4a − 3 + 4a − 3
=
4

s
(1 − 4a − 3)2
=
4

1 − 4a − 3
= = m2
2

45
That is xn+1 < m2 . Then

s
p 1− 4a − 3
xn+2 = a − xn+1 > a −
2

s
4a − 2 + 2 4a − 3
=
4

s
1 + 2 4a − 3 + 4a − 3
=
4

s
(1 + 4a − 3)2
=
4

1 + 4a − 3
= = m1
2
That is xn+2 > m1 .
In this case, consider the even and odd subsequences. One of them must be bigger than m1 .
So the limit (if exists) must also be bigger than or equal to m1 . This contradicts with the limit
of xn (if exists) must be equal to l1 . So the limit of xn does not exist.
Similarly, if l1 < xn < m1 , then m2 < xn+1 < l1 , and l1 < xn+3 < m1 .
So if l1 < x1 < m1 , then l1 < x3 < m1 . And

q
x3 − x1 = a − a − x1 − x1

a − a − x1 − x21
=p √
a − a − x1 + x1
x41 − 2ax1 + a2 − a + x1
= p √ √
( a − a − x1 + x1 )(a − x21 + a − x1 )
(x1 − l2 )(x1 − m2 )(x1 − l1 )(x1 − m1 )
= p √ √
( a − a − x1 + x1 )(a − x21 + a − x1 )
Because l1 < x1 < m1 , the numerator is less than 0. And the denominator is always positive,
so x3 − x1 < 0. Use MI, we can obtain that the odd subsequence is a decreasing sequence
bounded p below by l1 . So the limit exists. By taking the limit of the recursive equation

x2n+1 = a − a − x2n−1 , we get the limit must be l1 .
Also if l1 < x1 < m1 , then m2 < x2 < l1 . This leads to m2 < x4 < l1 . And

q
x4 − x2 = a − a − x2 − x2

a − a − x2 − x22
=p √
a − a − x2 + x2
x42 − 2ax2 + a2 − a + x2
= p √ √
( a − a − x2 + x2 )(a − x22 + a − x2 )
(x2 − l2 )(x2 − m2 )(x2 − l1 )(x2 − m1 )
= p √ √
( a − a − x2 + x2 )(a − x22 + a − x2 )
Because m2 < x2 < l1 , both the numerator and the denominator are positive, so x4 > x2 . By
MI, the even subsequence is an increasing sequence bounded by m2 < x2n <p l1 . So√it is also
a convergent sequence. By taking the limit of the recursive equation x2n+2 = a − a − x2n ,
we get the limit must by l1 , too. √
So the limit of the original sequence is l1 = −1+ 21+4a .
If m2 < x1 < l1 , using similar argument, we will be able to show that the odd subsequence is
an increasing bounded sequence with limit being l1 , and the even subsequence is a decreasing
bounded sequence with limit being l1 , too.

46

So the limit of the original sequence is l1 = −1+ 21+4a .
– In case m2 < 0: we will always have m2 < 0 < x2 < l1 for some n. Then the subsequence
{xn , xn+2 , xn+4 , . . .} will be an increasing sequence bounded by 0 and l1 . By previous argument
the limit is l1 . At the same time the subsequence {xn+1 , xn+3 , . . .} is a decreasing subsequence
bounded √by l1 and m1 . And its limit is also l1 . So the limit of the original sequence is
l1 = −1+ 21+4a .
– In case 0 < a < 34 :
In this case, x4 − 2ax2 + x + a2 − a = 0 only has two real solutions:l1 and l2 . Then if x1 > l1

(x1 − l2 )(x1 − l1 )(x2 − x + 1 − a)


x3 − x1 = p √ √ >0
( a − a − x1 + x1 )(a − x21 + a − x1 )

Using MI, we will be able to conclude that the odd subsequence is increasing with lower bound
l1 . So even if it has a limit, the limit will be strictly bigger than l1 , which contradicts with the
result obtained from taking limit of both sides of the recursive relation. So the limit of the odd
subsequence does not exist.
If 0 < x1 < l1

(x1 − l2 )(x1 − l1 )(x2 − x + 1 − a)


x3 − x1 = p √ √ <0
( a − a − x1 + x1 )(a − x21 + a − x1 )

Using MI, we will be able to conclude that the odd subsequence is decreasing with upper bound
l1 . So even if it has a limit, the limit will be strictly smaller than l1 , which contradicts with
the result obtained from taking limit of both sides of the recursive relation. So the limit of the
odd subsequence does not exist.
In conclusion, the limit

of the sequence exists when a > 1. And starting from x1 ∈ (m2 , m1 ), the
limit will be l1 = −1+ 21+4a . When x1 is not in that range, the sequence diverges.
1.3.10 Without loss of generality, we may assume that a < b. Denote ∆ = b − a, we claim that xn =
2 n−2
a + ∆ + − 21 ∆ + − 12 ∆ + · · · − 12

∆ for n ≥ 2.

• When n = 2, x2 = a + ∆ = a + b − a = b.
• Assume that when n ≤ k the claim is true, then when n = k + 1,
xk + xk−1
xk+1 =
2
2 k−2 2 k−3
a + ∆ + − 12 ∆ + − 21 ∆ + · · · − 12 ∆ + a + ∆ + − 12 ∆ + − 12 ∆ + · · · − 12
 

=
2
   2  k−3  k−2
1 1 1 1 1
=a + ∆ + − ∆+ − ∆ + ··· − ∆+ − ∆
2 2 2 2 2
   2  k−3    k−2
1 1 1 1 1
=a + ∆ + − ∆+ − ∆ + ··· − ∆+ 1− · − ∆
2 2 2 2 2
   2  k−3  k−2  k−1
1 1 1 1 1
=a + ∆ + − ∆+ − ∆ + ··· − ∆+ − ∆+ − ∆
2 2 2 2 2
2 n−2
So by MI, we have that xn = a + ∆ + − 12 ∆ + − 12 ∆ + · · · − 12

∆ for n ≥ 2. So
2 n−2 n
1 − − 21
   
1 1 1
xn = a + ∆ + − ∆+ − ∆ + ··· − ∆=a+∆·
2 2 2 1 + 12
n
Because limn→∞ − 12 = 0, limn→∞ xn = a + 32 ∆ = a + 23 (b − a) = 31 a + 23 b.

47
xn+1 xn xn+1 xn + xn−1
1.3.12 1 By the definition of {yn }, yn+1 = and yn = . Then yn+1 = = =
xn xn−1 xn xn
xn−1 1
1+ =1+ .
xn yn
2 If {yn } converges, then assume the limit is lim yn = l and take limit on both sides of the recursive
n→∞
formula, we have that
1
lim yn+1 = 1 +
n→∞ limn→∞ yn
1
l =1+
l

1± 5
The solution to the equation is l = . Clearly the sequence {yn } is a positive sequence, then
√ 2
1+ 5
we have that lim yn = .
n→∞ 2
1 1 1 yn
3 Because yn+1 = 1 + , then yn+2 = 1 + =1+ =1+ .
yn yn+1 1 + y1n yn + 1
√ √
x1 1+ 5 x2 1+ 5
Given that y0 = =1< , y1 = =2> .
x0 2 x1 2
y2k 1 1
For the even subsequence {y2k }, we have that y2k+2 = 1 + = 2− < 2− √ =
y2k + 1 y2k + 1 1+ 5
+ 1
√ √ 2
1+ 5 1+ 5
. So l = is an upper bound for the even subsequence {y2k }.
2 2
y2k+1 1
For the odd subsequence {y2k+1 }, we have that y2k+3 = 1 + = 2− > 2−
√ √ y2k+1 + 1 y2k+1 + 1
1 1+ 5 1+ 5
√ = . So, l = is a lower bound for the odd subsequence {y2k+1 }.
1+ 5
+1 2 2
2
1 1 yn (1 + yn−2 ) − yn−2 (1 + yn ) yn − yn−2
4 By yn+2 − yn = 1 + −1− = = , and
yn+1 yn−1 (1 + yn )(1 + yn−2 ) (1 + yn )(1 + yn−2 )
x1 x2 x3 3 x4 5
y0 = = 1, y1 = = 2, y2 = = , and y3 = = . By Mathematical induction, we have
x0 x1 x2 2 x3 3
that the even sequence {y2n } is increasing and the odd sequence {y2n+1 } is decreasing.
5 By the even subsequence is an increasing sequence bounded above, it has a limit. By taking limit
on both sides of the recursive relation, we have that lim y2k = l. And the odd subsequence is a
k→∞
decreasing sequence bounded below, so it has a limit as well. Then take limit on both sides of the
odd recursive relation, we have that lim y2k+1 = l. So the whole sequence {yn } has a limit the
k→∞
same as the even and the odd subsequence, that is l.
1.3.13(1) Consider xn+1 − xn we have that
a − x2n
 
1 a
xn+1 − xn = xn + − xn =
2 xn 2xn
   
Given that x1 > 0, we have x2 = 21 x1 + xa1 > 0. Also, if xk > 0, then xk+1 = 1
2 xk + a
xk > 0. So in
  √
general xn+1 = 12 xn + xan ≥ a.
a−x2n
For sufficiently big n, we have xn+1 − xn = 2xn ≤ 0. So the sequence is a decreasing sequence.
Being a decreasing bounded sequence, xn must converge.
Assume limn→∞ xn = l, then take limit on both sides of the inductive equation, we have that
1 a
l= l+ .
2 l
√ √
Solving this equation, we have that l = ± a. Because xn is a positive sequence, so limn→∞ xn = a.

48
n n
1.3.14 First assume 0 < a < 1. limn→∞ n+1 = 1. So by the Order Rule, for sufficiently big n, n+1 > a. That
n n n n+1
is, for sufficiently big n, xn = na = (n + 1)a · n+1 > (n + 1)a = xn+1 . So the sequence nan is a
decreasing sequence for sufficiently big n. Since xn is always positive, we have the sequence is a bounded
decreasing sequence, it is convergent. Assume limn→∞ nan = l, then limn→∞ nan+1 = l, too. Then by
arithmetic rule,
n 1 n 1
l = lim nan = lim (n + 1)an · = lim (n + 1)an+1 · lim = ·l
n→∞ n→∞ n+1 a n→∞ n→∞ n + 1 a
Since a 6= 1, we get l = 0.

1.3.16 (2)
 n !n
1 1 1 1 n−1 1
lim 1− = lim 1 = lim 1 n
= lim 1 n−1
· =
n→∞ n n→∞ 1 + n−1 n→∞ (1 + n−1 )
n→∞ (1 + n−1 )
n e

(4)
 n  n
2n + 1 1
lim = lim 1+ 1
n→∞ 2n − 1 n→∞ n− 2

Because
 n  n  n
1 1 1
1+ < 1+ < 1 +
n n − 21 n−1

and
 n
1
lim 1+ =e
n→∞ n
n
  n−1  
1 1 1
lim 1+
= lim 1 + · 1+ =e
n→∞ n−1 n→∞ n−1 n−1
 n
1
by the Sandwich rule, lim 1 + = e.
n→∞ n − 12
1.3.18 By the binomial expansion
 n
1 1 n(n − 1) 1 n(n − 1) · · · 3 · 2 · 1 1
1+ =1 + n · + · 2 + ··· + · n
n n 2! n n! n
      
1 1 1 1 1 2 n−1
=1 + + 1− + ··· + 1− 1− ··· 1 −
1! 2! n n! n n n
      
1 1 1 1 1 2 k
≥1 + + 1− + ··· + 1− 1− ··· 1 −
1! 2! n k! n n n

Then because of the Order Rule,


 n       
1 1 1 1 1 1 2 k
lim 1 + ≥ lim 1 + + 1− + ··· + 1− 1− ··· 1 −
n→∞ n n→∞ 1! 2! n k! n n n

that is
1 1 1
e≥1+ + + ···
1! 2! k!

49
On the other hand,
 n
1 1 n(n − 1) 1 n(n − 1) · · · 3 · 2 · 1 1
1+ =1 + n · + · 2 + ··· + · n
n n 2! n n! n
      
1 1 1 1 1 2 n−1
=1 + + 1− + ··· + 1− 1− ··· 1 −
1! 2! n n! n n n
1 1 1
<1 + + + ··· +
1! 2! n!
So
 k
1 1 1 1
1 + + + ··· + > 1+
1! 2! k! k

Combine these two inequalities, we have


 k
1 1 1 1
e≥1+ + + ··· + > 1+
1! 2! k! k
 k
1
Then because lim 1+ = e, by the Sandwich Rule,
k→∞ k

1 1 1
lim 1 + + + ··· + =e
k→∞ 1! 2! k!
That is
1 1 1
lim 1 + + + ··· + =e
n→∞ 1! 2! n!

1.3.20 We prove {xn } is convergent by proving it is a Cauchy sequence.


1
For any  > 0, take N = + 1, then m, n > N (assume m > n) implies

|xm − xn | = |xm − xm−1 + xm−1 − xm−2 + xm−2 − · · · − xn+1 + xn+1 − xn |
≤ |xm − xm−1 | + |xm−1 − xm−2 | + · · · + |xn+1 − xn |
1 1 1
≤ + + ··· + 2
(m − 1)2 (m − 2)2 n
1 1 1
≤ + + ··· +
(m − 2)(m − 1) (m − 2)(m − 3) n(n − 1)
1 1 1 1 1 1
≤− + − + − ··· − +
m−1 m−2 m−2 m−3 n n−1
1 1
= −
n−1 m−1
1
≤ <
n−1
Then {xn } is a Cauchy sequence, so it converges.

1.3.21 Because cn is bounded, we may assume that |cn | < M for some M > 0. Then for any  > 0, take

50
(1−|r|)
log
N= M
log |r| , then m, n > N (assume that m > n) implies that

|xm − xn | =|cn+1 rn+1 + cn+2 rn+1 + · · · + cm rm |


=|cn+1 ||r|n+1 + |cn+2 ||r|n+1 + · · · + |cm ||r|m
≤M (|r|n+1 + |r|n+2 + · · · + |r|m )
1 − |r|m−n
=M |r|n+1 ·
1 − |r|
M
≤ |r|n+1
1 − |r|
<

1 1
1.3.22 (2) Let xn = 1 − 23 + − · · · + (−1)n+1 n13 . For any integers m > n, we estimate:
33

(−1)m+1 (−1)n+2

|xm − xn | = + · · · +
m3 (n + 1)3
1 1
≤ 3 + ··· +
m (n + 1)3
1 1
≤ 2 + ··· +
m (n + 1)2
1 1
≤ + ··· +
m(m − 1) (n + 1)n
     
1 1 1 1 1 1
= − + − + ··· + −
m−1 m m−2 m−1 n n+1
1 1 1
= − < .
n m n
Hence when n > 1ε , we have |xm − xn | < ε.
Now given any ε > 0, choose N = 1ε , then whenever m > n > N , according to the above estimation,
we get |xm − xn | < n1 < ε. This shows {xn } is a Cauchy sequence, and hence it converges to a
certain real number by Cauchy Criterion.
[Remark: The problem requires the use of Cauchy Criterion. The following solu-
tion (typed by a former TA) did not, but it provides an alternative approach using
monotone sequences.]
1 1 (−1)2n+1 1 1
The even subsequence is 1− 3 + 3 −· · ·+ 3
. Because 3
− > 0, the sequence
2 3 (2n) (2k − 1) (2k)3
1 1
is an increasing sequence. Because − 3
+ < 0, the sequence is bounded above by 1.
(2k) (2k + 1)3
So the even subsequence is convergent.
1 1 (−1)2n+1 (−1)2n+2 1 1
The odd subsequence is 1− 3 + 3 − · · ·+ + . Because − + < 0,
2 3 (2n)3 (2n + 1)3 (2k)3 (2k + 1)3
1 1
the subsequence is decreasing. Because 3
− > 0, the sequence is bounded below by
(2k + 1) (2k)3
0. So the odd subsequence is convergent.
1 1 (−1)2n+1 1 1 (−1)2n+2
Assume lim 1 − 3 + 3 − · · · + 3
= le . lim 1 − 3 + 3 − · · · + = lo . And
n→∞ 2 3 (2n) n→∞ 2 3 (2n + 1)3
1
lim x2n+1 − x2n = lim = 0. We have lo − le = 0. So the odd subsequence and the even
n→∞ n→∞ (2n + 1)3
subsequence have the same limit. That is the original sequence is convergent.

1.4.2 • If limn→∞ x1n = 0, then for any M > 0, there exists N > 0, such that n > N implies x1n < M 1
.

That is |xn | > M . So limn→∞ xn = ∞.

51

• If limn→∞ xn = ∞, then for any  > 0, there exists N > 0, such that |xn | > 1 . That is x1n < .

1
So limn→∞ xn = 0.

1.4.3 (2) For any M > 0, take N = (M + 1)2 , then n > N implies


n
√ n n−1
n + 1 = √n + 1 ≥ √n + 1 = n − 1 > M

n
So lim √ = ∞.
n→∞ n+1
44 M
(5) For any M > 0, take N = 3! , then n > N implies

n!
= n! ≥ 3! · n > M
4n 4n 43 4
n!
So lim = ∞.
n→∞ 4n
1.4.4(2) Claim: If limn→∞ xn = l, limn→∞ yn = +∞, then limn→∞ xn + yn = +∞.
Because limn→∞ xn = l, for any 1 > 0, there is N1 such that n > N1 implies |xn − l| < 1 .
Because limn→∞ yn = +∞, for any M2 > 0, there is N2 such that n > N2 implies yn > M2 .
Then for any M > 0, consider 1 = 1, M2 = M − l + 1, take N = max{N1 , N2 }, then n > N implies
l − 1 < xn < l + 1 , yn > M2 . So xn + yn > M − l + 1 + l − 1 = M . That is limn→∞ xn + yn = +∞.
1.4.5 (1) Consider xn = n and yn = n. Then limn→∞ xn = ∞ and limn→∞ yn = ∞. And limn→∞ xn + yn =
limn→∞ 2n = ∞.
(2) Consider xn = n+1, yn = −n+1. Then limn→∞ xn = ∞ and limn→∞ yn = −∞. And limn→∞ xn +
yn = limn→∞ 2 = 2.
(3) Consider xn = n + (−1)n , yn = −n. Then limn→∞ xn = ∞ and limn→∞ yn = −∞. And
limn→∞ xn + yn = limn→∞ (−1)n . This limit does not exist.
(4) Consider xn = n2 + n, yn = −n2 + (−1)n n. Then limn→∞ xn = ∞ and limn→∞ yn = −∞. And
limn→∞ xn + yn = limn→∞ n + (−1)n n. The sequence is {0, 2, 0, 4, 0, 6, . . .}. It diverges and does
not diverge to infinity.
1.4.7 If limn→∞ xn = +∞, then for any M1 > 0, there exists N1 such that n > N1 implies xn > M1 .
If xn ≤ yn for sufficiently big n, then there exists N2 such that n > N2 implies xn ≤ yn .
Then for any M > 0, choose M1 = M and take N = max{N1 , N2 }, then n > N implies yn ≥ xn >
M1 = M . So limn→∞ yn = +∞.

xn
xn
xn 1 + l
1.4.10 If lim = l, then limn→∞ xn−1 = |l|. So there exists N1 such that n > N1 implies > .
n→∞ xn−1 xn−1 2
So
n−N1
1 + l n

xn xn−1
· · · xN1 +1 · |xN1 | ≥ 1 + l

|xn | = · · |x N1 | = 2 ·C

xn−1 xn−2 xN
1
2
−N
· |xN1 |. So for any M > 0, take N = max{ log(M/C)

where C = 1+l
2

log | 1+l |
}. Then n > N implies |xn | > M .
2
That is xn diverges to infinity.
1.4.11 (1) For any M > 0, take N = (M + 1)2 , then n > N implies
n + sin 2n n−1 √
√ ≥√ = n−1>M
n − cos n n+1
n + sin 2n
So lim √ = +∞.
n→∞ n − cos n

52
(4)
1 1 1
√ √
n
= √
n
· √
n − 2n
n
1− 2 n
n

Because 1−1√ n
2
is of the format 01− (1 < n 2, for any positive integer n), when n goes to infinity.
Then by the Arithmetic rule, limn→∞ 1−1√ n
2
= −∞. So the original limit is of the format (−∞) · 1.
1
Then lim √ √ = −∞.
n→∞ n n − n 2n
n n
(7) Because limn→∞ 1 − 32 = 1 and limn→∞ n 13 = 0, the original limit is of the format 01 and the
denominator is always positive, so the original limit goes to +∞.
 n  n2
1 1
(9) Because lim 1 + = e > 2. Then for sufficiently big n, 1 + > 2n . Because 2n goes
n→∞ n n
 n2
1
to positive infinity, 1 + also goes to positive infinity.
n
(n+1)2 (n+1)2
1.4.12 For any a > 1, because limn→∞ n2 = 1, for sufficiently big n, we have n2 < a. Then

an an (n + 1)2 an an+1
= · < · a =
n2 (n + 1)2 n2 (n + 1)2 (n + 1)2
So for sufficiently big n, the sequence xn is increasing. If the sequence converges to a finite limit, assume
n
limn→∞ an2 = l. Then

an 1 an+1 (n + 1)2 1
l = lim = lim · = ·l
n→∞ n2 a n→∞ (n + 1)2 n2 a
Since the sequence is all positive and increasing for sufficiently big n, then we have l > 0. So we must
have a = 1. This contradicts with the condition a > 1. So the sequence does not converge to any finite
an
number l. Then by Prop. 1.4.2, lim 2 = +∞ for a > 1.
n→∞ n

1.5.3 (4) By the Arithmetic Rule:


x2 − 1 1 − x12 1 − limx→∞ x12 1−0
lim 2 = lim = = = 1.
x→∞ x + x − 2 x→∞ 1 + 1 − 2 12 1 + limx→∞ x1 − 2 limx→∞ 1 1+0−0
x x x2
(5) By the Arithmetic Rule:
2
x2 − 1 (limx→0 x) − 1 0−1 1
lim 2
= 2 = =
x→0 x + x − 2 (limx→0 x) + limx→0 x − 2 0+0−2 2

(6) By the Arithmetic Rule:

x2 − 1 (x − 1)(x + 1) x+1 limx→1 x + 1 1+1 2


lim = lim = lim = = =
x→1 x2 + x − 2 x→1 (x − 1)(x + 2) x→1 x + 2 limx→1 x + 2 1+2 3

1.5.4(4)
p (x + a)(x + b) − x2 (a + b)x + ab
lim (x + a)(x + b) − x = lim p = lim p
x→+∞ x→+∞ (x + a)(x + b) + x x→+∞ (x + a)(x + b) + x

Without loss of generality, we may assume that a < b, then for sufficiently big x:
(a + b)x + ab (a + b)x + ab (a + b)x + ab (a + b)x + ab (a + b)x + ab
=p ≤p ≤p =
x+b+x (x + b)(x + b) + x (x + a)(x + b) + x (x + a)(x + b) + x x+a+x

53
By Arithmetic Rule:
ab
(a + b)x + ab (a + b) + x a+b
lim = lim =
x→+∞ x+b+x x→+∞ 2 + xb 2
ab
(a + b)x + ab (a + b) + x a+b
lim = lim =
x→+∞ x+a+x x→+∞ 2 + xa 2
p a+b
So the Sandwich Rule gives us lim (x + a)(x + b) − x = .
x→+∞ 2
In case x → −∞: −x → +∞ and | (x + a)(x + b)| ≥ |x|
p p
2 for sufficiently small x. Then x→−∞
lim (x + a)(x + b)−
x = +∞.

1.5.5 For any a > 0 and  > 0, take δ =  a, then 0 < |x − a| < δ implies
√ √ √ √
√ √ | x − a| · | x + a| |x − a| |x − a|
| x − a| = √ √ =√ √ < √ <
| x + a| x+ a a
√ √ +
Then limx→∞ x = a .
1.5.7 (2)
1 √ √ a + x − (a − x)
lim ( a + x − a − x) = lim √ √
x→0 x x→0 x( a + x + a − x)
2x
= lim √ √
x→0 x( a + x + a − x)
2
=√ √
a + limx→0 x + a − limx→0 x
1
=√
a

(4)
r ! a+x a

r
1 a+x a b b+x
lim − = lim q 
x→0 x b b+x
q
x→0 a+x a
x b + b+x

ab + ax + bx + x2 − ab
= lim q q 
x→0 a+x a
xb(b + x) b + b+x

a+b+x
= lim q q 
x→0 a+x a
b(b + x) b + b+x

a + b + limx→0 x
= q q 
a+limx→0 x a
b(b + limx→0 x) b + b+limx→0 x

a+b
= √
2b ab

1.5.8 (2) Because

x → ∞, ⇔ x + a → ∞,

use the substitution y = x + a, we have that

lim f (x + a) = lim f (y)


x→∞ y→∞

54
(4) Because
√ √
x → 1, x 6= 1 ⇔ x → 1, x 6= 1,

use the substitution y = x, we have that

lim f ( x) = lim f (y)
x→1 y→1

(6) Because

x → a, x 6= a ⇔ x2 + x → a2 + a,

use the substitution y = x2 + x, we have that

lim f (x2 + x) = lim f (y)


x→a y→a2 +a

1.5.9 (1)

lim f (x) = lim+ 2 = 2


x→0+ x→0
lim f (x) = lim− 1 = 1
x→0− x→0

Because lim+ f (x) 6= lim− f (x), the limit lim f (x) does not exist.
x→0 x→0 x→0

(2)

lim f (x) = lim+ 1 = 1


x→0+ x→0
lim− f (x) = lim− 1 = 1
x→0 x→0

Because lim+ f (x) = lim− f (x) = 1, the limit lim f (x) = 1.


x→0 x→0 x→0

(3)

lim f (x) = lim+ −x2 = 0


x→0+ x→0
lim− f (x) = lim− x = 0
x→0 x→0

Because lim+ f (x) = lim− f (x) = 0, the limit lim f (x) = 0.


x→0 x→0 x→0

1.5.10(3)

√ √ √ x((x + a) − (x + b))
lim x( x + a − x + b) = lim √ √
x→+∞ x→+∞ x+a+ x+b

(a − b) x
= lim √ √
x→+∞ x+a+ x+b
a−b
= lim p q
x→+∞
1 + xa + 1 + xb
a−b
=p q
a b
1 + limx→+∞ x + 1 + limx→+∞ x
a−b
=
2

55
1.5.11 (2) Because

x → −∞ ⇔ x + a → −∞

use the substitution y = x + a, we have that

lim f (x + a) = lim f (y)


x→−∞ y→−∞

(4) Because
√ √
x → 0, x > 0 ⇔ x → 0, x > 0

use the substitution y = x, we have that

lim+ f ( x) = lim f (y)
x→0 y→0

(6) If a > − 21 , then

x → a, x < a ⇔ x2 + x → a2 + a, x2 + x < a2 + a

use the substitution y = x2 + x, we have that

lim f (x2 + x) = lim f (y)


x→a− y→(a2 +a)−

If a ≤ − 21 , then

x → a, x < a ⇔ x2 + x → a2 + a, x2 + x > a2 + a

use the substitution y = x2 + x, we have that

lim f (x2 + x) = lim f (y)


x→a− y→(a2 +a)+

1.5.13 (1)
sin πx − sin(πx + π) − sin πy
lim = lim = lim · π = −π
x→−1 x+1 x→−1 x+1 y→0 πy

(3)
√ 1
sin x − cos x 1 2( √2 sin x − √12 cos x)
limπ = limπ ·
x→ 4 4x − π x→ 4 4 x − π4
√ √ √
2 sin(x − π4 ) 2 2
= limπ = · 1 =
x→ 4 4 x − π4 4 4

(6)
y
− sin2 sin2 y2 y
 
1 cos y − 1 2
lim x cos − 1 = lim = lim = − lim y · =1·0=0
x→∞ x y→0 y y→0 y y→0 ( )2 4
2

(11)
tan(sin x) tan(sin x) sin x
lim = lim ·
x→0 x x→0 sin x x
sin(sin x) 1 sin x
= lim · ·
x→0 sin x cos(sin x) x
1
= 1 · · 1 = 1.
1

56
(13)

tan(sin x) 1 sin(sin x) sin x x √
lim+ √ = lim+ · · · √ · x=1·1·1·1·0=0
x→0 sin x x→0 cos(sin x) sin x x sin x

1.6.3(4) For any  > 0, take M = − 2 , then x < M implies



x+1 x + 1 + x2 + 1
√ √
x2 + 1 − (−1) =

x2 + 1

(x + 1 + x2 + 1)(x + 1 − √x2 + 1)

= √ √

2 2

x + 1(x + 1 − x + 1)
(x + 1)2 − (x2 + 1)


= √ √
x2 + 1( x2 + 1 − x − 1)

2x 1
≤ √

x2 + 1 −x


2 1
= q
1 + 1 −x

x2
2
≤ <
−x
x+1
Then lim √ = −1.
x→−∞ x2 + 1

1.6.4 (2) Because a > 0, for any  > 0, take δ =  a, then 0 < |x − a| < δ implies
√ √ |x − a| 1
| x − a| = √ √ ≤ √ |x − a| < 
x+ a a
√ √
So, lim x= a.
x→a
 2
|a−b|
(5) For any  > 0, take N = max{ 43 |a|, 34 |b|,  }, then x > N implies

√ √ |(x + a) − (x + b)| |a − b| |a − b| |a − b|
| x + a − x + b| = √ √ =√ √ ≤q q = √ <
x+a+ x+b x+a+ x+b 3 3
x − 4x + x − 4x x

1.6.7 For any ε > 0, there exists δ1 > 0 and δ2 > 0 such that

0 < |x − a| < δ1 =⇒ |f (x) − l| < ε


0 < |x − a| < δ2 =⇒ |h(x) − l| < ε

Now pick δ = min{δ1 , δ2 }, then if 0 < |x − a| < δ, we have both f (x) − l > −ε and h(x) − l < ε. Using
the fact that f (x) ≤ g(x) ≤ h(x), we have

−ε < f (x) − l ≤ g(x) − l ≤ h(x) − l < ε.

In other words, |g(x) − l| < ε. This shows the desired result.


1.6.10 Claim: (−∞) + l = −∞.
Assume there are two functions f (x) and g(x), such that limx→a f (x) = −∞, and limx→a g(x) = l. Then
for any B1 < 0, there exists δ1 such that 0 < |x − a| < δ1 implies f (x) < B1 . For any 2 > 0, there exists
δ2 such that 0 < |x − a| < δ2 implies |g(x) − l| < 2 .

57
Then for any B < 0, consider B1 = B − l − 1, 2 = 1, then take δ = min{δ1 , δ2 }, 0 < |x − a| < δ implies

f (x) + g(x) < B1 + l + 1 = B − l − 1 + l + 1 = B

This means that lim f (x) + g(x) = −∞.


x→a
l
Claim: = 0.

Assume there are two functions f (x) and g(x), such that limx→a f (x) = ∞, and limx→a g(x) = l. Then
for any B1 > 0, there exists δ1 such that 0 < |x − a| < δ1 implies |f (x)| > B1 . For any 2 > 0, there
exists δ2 such that 0 < |x − a| < δ2 implies |g(x) − l| < 2 .
Then for any  > 0, consider B = 1+|l|
 , 2 = 1, then take δ = min{δ1 , δ2 }, 0 < |x − a| < δ implies

g(x) |g(x)| |g(x) − l| + |l| 2 + |l|

f (x) − 0 |f (x)| ≤
= < =
B1 B1

g(x)
That is lim = 0.
x→a f (x)
1.6.12 Because limx→a f (x) = l, then for any 0 > 0, there exists δ0 such that 0 < |x − a| < δ0 implies
|f (x) − l| < 0 . Then for any  > 0, consider 0 =  then take δ = δ0 , then 0 < |x − a| < δ implies
0 < |x − a| < δ0 , and

| max{f (x), l} − l| = | max{f (x) − l, 0}| ≤ max{|f (x) − l|, 0} ≤ |f (x) − l| < 0 = 

So lim max{f (x), l} = l.


x→a
The sequence version is: lim xn = l implies lim max{xn , l} = l.
n→∞ n→∞
If limn→∞ xn = l, then for any 0 > 0, there exists N0 such that n > N0 implies |xn − l| < 0 . Then for
any  > 0, consider 0 = , then take N = N0 , then n > N implies n > N0 , and

| max{xn , l} − l| = | max{xn − l, 0}| ≤ max{|xn − l|, 0} ≤ |xn − l| < 0 = 

So lim max{xn , l} = l.
n→∞

1 1
1.6.17 For any 0 < x < 1, we have ≤ x ≤ for some natural number n. Then if a > 1, then
n+1 n
1 1
a n+1 ≤ ax ≤ a n
√ √
Because limn→∞ n
a = 1, we also have limn→∞ n+1 a = 1. Thus for any  > 0, there is N , such that
√ √
n < N ⇒ | n a − 1| < , | n+1 a − 1| < 

Then
1 1 1 1
0<x< ⇒ ≤ x ≤ for some natural number n > − 1 > N
1+N n+1 n x
1 1
⇒ −  < a n+1 − 1 ≤ ax − 1 ≤ a n − 1 < 
⇒|ax − 1| < 

So lim ax = 1, for a > 1.


x→0
1
In case 0 < a < 1, then b = 1
a > 1 and limx→0 bx = 1. So lim ax = lim = 1.
x→0 x→0 bx
When a = 1, limx→0 ax = limx→0 1 = 1.
So lim ax = 1 for all a > 0.
x→0

58
 x
1
1.6.22 (4) Because lim 1+ = e > 2. We have that for sufficiently big x, (1 + x1 )x > 2. So
x→+∞ x
  x2
1
lim 1+ ≥ lim 2x = +∞
x→+∞ x x→+∞

  x2
1
So lim 1+ = +∞.
x→+∞ x
(6)
x x
(1 + xa )x limx→∞ (1 + xa ) a ·a ea

x+a
lim = lim = x = = ea−b
x→∞ x+b x→∞ (1 + xb )x limx→∞ (1 + xb ) b ·b eb

1.6.23 (2)
n 2n· 21

 
1 1 1
lim 1+ = lim 1+ = e2 = e
n→∞ 2n n→∞ 2n

(8) Because
 n−1  n2
 n+1  n
1 1 1
1+ ≤ 1+ ≤ 1+
n n n
 n   n2
1 1 1 n+1
Because lim 1+ = e and lim 1+ = 1, and the Sandwich rule, we have that lim 1 + =
n→∞ n n→∞ n n→∞ n
e.
(14) Because
 n+2  1
n+1+ n−1  n2
 n−1   n2  n
1 1 1 1 n+(−1)n 1
1− < 1+ = 1− < 1+ < 1−
n n n n n
 n+2  −n !−1  2  n
1 1 1 1 1 1
and lim 1 − = lim 1− · 1− = , lim 1 − = , by the
n→∞ n n→∞ n n e n→∞ n e
Sandwich rule, we have that
 n2
 n+(−1)n
1 1
lim 1+ =
n→∞ n e
2
−1
1.7.1 (2) The function f (x) = xx−1 is undefined at x = 1, so it is discontinuous at x = 1. Because
x2 − 1
lim = lim x + 1 = 2, if we define f (1) = 2, then lim f (x) = f (1), then the function could
x→1 x − 1 x→1 x→1
become continuous on the whole real line after the extension.
(4) The function f (x) = x sin x1 is undefined at x = 0, so it is discontinuous at x = 0. Because
1
−|x| ≤ x sin x1 ≤ |x|, and limx→0 |x| = limx→0 −|x| = 0, by the Sandwich theorem, lim x sin = 0.
x→0 x
So if we define f (0) = 0, then lim f (x) = f (0), then the function could become continuous on the
x→0
whole real line after the extension.
1.7.2 f (x) = (x − 1)(x − 2)(x − 3)D(x) where the D(x) is the Dirichlet function.
1.7.5 (a) If f (0) = 0, then c = 0.
(b) If f (0) 6= 0, then f (0) > 0 by the codomain is [0, 1].

59
i. If f (1) = 1, then c = 1.
ii. If f (1) 6= 1, then f (1) < 1 by the codomain is [0, 1].
Consider the function g(x) = f (x) − x, then
g(x) is continuous because y = f (x) and y = x are both continuous. So we have

g(0) =f (0) > 0


g(1) =f (1) − 1 < 0

So g(c) = 0 for some c ∈ [0, 1] by the Intermediate Value Theorem.


That is f (c) = c.
1.7.6 (2) Let y = xx , then ln y = x ln x, so y1 y 0 = ln x + 1. Then y 0 = 0 implies that x = 1e . And when
x ∈ ( 1e , +∞), y 0 > 0, and when x ∈ (0, 1e ), y 0 < 0. So the function reaches a minimum at x = 1e .
1
And when x → ∞, xx → ∞. So y = xx can take value from 1e e to +∞.
(6) f (x) = 2x is a continuous function, and f (−1) = 21 , f (0) = 1. So the function ranges from 12 to 1.
And g(x) = x2 + 3 is also a continuous function, and g(0) = 3, g(1) = 4. Because x ∈ (0, 1], we
have g(x) ∈ (3, 4]. So the function can take value in [ 12 , 1] ∪ (3, 4].

1.7.9(3)

log x − log x log(y + a) − log a log(1 + ay ) 1 1


lim = lim = lim y · =
x→a x−a y→0 y y→0
a a a

1.7.10(2) When x → 1, we have that x2 − 1 → 0 and x2 − 2x + 1 → 0, so log(x2 − 2x + 1) → −∞. Then

log(x2 − 2x + 1)
lim =∞
x→1 x2 − 1

1.7.11 (2)

1 ax + b log( ax
b + 1) log b log( cxd + 1) log d
lim log = lim + − −
x→0 x cx + d x→0 x x x x
log(1 + ax
b ) a log(1 + cx
d ) c log b log d
= lim ax · − cx · − −
x→0
b b d d x x
a c
= −
b d
only if b = d and bd 6= 0.
(5)

ax + b 1 by + a b d
lim x log = lim log = −
x→∞ cx + d y→0 y dy + c a c
only if a = c and ac 6= 0 by the result of the previous question.
1.7.12 (2)

log(x2 + 3x + 2) − log 2 log(1 + x) + log(2 + x) − log 2


lim x → 0 = lim
x x→0 x
log(1 + x) log(1 + x2 ) 1
= lim + x ·
x→0 x 2 2
1 3
=1 + =
2 2

60
(4)

log x log(1 + y) y log(1 + y) y


lim = lim · = lim ·− = −1
x→1 sin πx y→0 y sin(π(1 + y)) y→0 y sin(y)

1.7.13 (2) Let y = ax − 1, then x = loga (y + 1). And x → 0 is equivalent to y → 0.

ax − 1 y 1 1
lim = lim = lim 1 = = log a
x→0 x y→0 loga (y + 1) y→0 log (1 + y) y
a
log ae

(6)

ax − bx ax − 1 bx − 1
lim = lim − = log a − log b
x→0 x x→0 x x
By Ex 1.7.13(2).
x
1.7.14(4) Consider the substitution y = p log(x/a) and z = a − 1, then

xp − ap ep log(x/a) − 1 p log(x/a) p−1


lim = lim · x ·a
x→a x − a
a −1
x→a p log(x/a)

ey − 1 log(1 + z)
= lim · lim · pap−1
y→0 y z→0 z
=pap−1

61

You might also like