0.0.

1 Introduction
This Calculus course is designed for students in the first year at the Electronics,
Telecommunications and Information Technologies and Civil Engineering Facul-
ties. The course is due to last 14 weeks, 2 hours every week. The first part of the
course treats functions of one variable, while the second part treats more or less
the same topics, but for functions of several real variables.

Contents:
Part I: Functions of one variable
1. Sequences: Convergent sequences, Cauchy sequences, Special limits
2. Numerical series: Convergence tests for series with positive general term, Alter-
nating series, absolutelyely convergent series
3. Limits of functions and Continuity
4. Derivatives, the Differential, Taylor Polynomials, Local extrema
5. Sequences and series of functions: Uniform convergence, Power Series, Fourier
Series

Part II: Functions of several real variables

1. Metric spaces
2. Sequences in Rp
3. Limits and continuity: Partial limits, directional limits
4. Partial derivatives, the Differential
5. Taylor Polynomial, Local Extrema, Conditional Extrema

1
0.1 Sequences (Şiruri)
I will consider this chapter as being a revision, although limits of sequences are not
studied at M2. Moreover, we shall mention a few concepts and results that are new.

A sequence of real numbers is a function x : N −→ R (or possibly x : N∗ −→ R).

Instead of using the standard notation, x(n), for the value of the function at the
point n, we use a specific notation, xn . Profiting of the fact that N is "well-ordered",
we can enumerate the terms of the sequence (we can arrange the values of the func-
tion in order): x0 , x1 , x2 , . . . , xn , . . . .
We call xn the term of rank n (termenul de rangul n).
n!
Example 1. xn = ,n≥0
2n
1 1 3
We have: x0 = 1, x1 = , x2 = , x3 = , etc.
2 2 4
Here, the terms of the sequence are explicitly given. If you need x10 , you can
10!
directly compute 10 .
2
Another way of defining a sequence is "recurrently": given the first term of the
sequence (x0 or x1 , whichever is the first term) and the way each term depends on
the previous one, we can compute the subsequent terms of the sequence.

For example, if x0 = 1 and xn+1 = xn +2, ∀ n ≥ 0, we can successively compute

x1 = 3, x2 = 5, x3 = 7, and so on. You might remember that such a sequence is
called an arithmetic sequence (progression) and that there is a formula for directly
determining the term of rank n. (In this case, xn = 2n + 1.)

Other sequences are defined by a rule that is easy to understand even if writing
a formula to it would be much more difficult. Take for example the sequence whose
first terms are:
1, 2, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 2, . . ..

Definition.
A sequence (xn )n≥0 is called increasing (crescătoare) if xn ≤ xn+1 , ∀ n, i.e. if each
term of the sequence is less then or equal to the following one.

A sequence (xn )n≥0 is called decreasing (descrescătoare) if xn ≥ xn+1 , ∀ n, i.e.

if each term of the sequence is greater then or equal to the following one.

A sequence that is either increasing or decreasing is called monotone (or mono-

tonic).
This means that the difference xn+1 − xn keeps a constant sign, + for an increasing

2
sequence, − for a decreasing one.

If all the inequalities between xn and xn+1 are strict, we say that the sequences
are strictly increasing, decreasing, and monotone.

For example, the arithmetic progression that we have seen above is a strictly
n!
increasing sequence, while the sequence xn = n is not monotone because x0 > x1
2
but x2 < x3 .

Definition. A sequence (xn ) is called bounded if there exist two constants, m

and M , such that m ≤ xn ≤ M , ∀ n.
n+1
Example. The sequence xn = is bounded. It is easy to guess and prove
n+2
(after computing the first few terms) that 0 < xn < 1, ∀ n.

Also, we can prove that the sequence is strictly increasing, therefore xn ≥ x0 =

1
. We still need to notice that xn < 1.
2
(n + 1) + 1 n+1 (n + 2)2 − (n + 3)(n + 1)
Indeed, xn+1 − xn = − = =
(n + 1) + 2 n+2 (n + 2)(n + 3)
1
> 0.
(n + 2)(n + 3)
Definition. A sequence (xn ) is called convergent if there exists a real number
ℓ such that: ∀ ε > 0, ∃n0 ∈ N: n ≥ n0 ⇒ |xn − ℓ| ≤ ε.

In this case, the number ℓ is called the limit of the sequence (xn ) and we write
lim xn = ℓ, lim xn = ℓ (for sequences, we only study the behavior as n → ∞), or
n→∞
xn → ℓ.

A sequence that is not convergent is called divergent.

n+1
Example. lim =1
n→∞ n + 2
This can be checked with the definition: the condition |xn − 1| < ε comes to
n+1 1 1
− 1 < ε, or < ε, and is satisfied as soon as n + 2 > , true for
n+2 n+2 ε
1
n > − 2.
ε
The intuition of what this concept means: the terms of the sequence (xn ) are
approaching the number ℓ. More precisely, every interval centered at ℓ, (ℓ−ε, ℓ+ε),
however small, contains all the terms of the sequence (xn ) with only a finite num-

3
ber of exceptions.

A few important theorems, whose proofs can be found in all classical textbooks:

Theorem (Weierstrass)1
Any sequence that is both monotone and bounded is a convergent sequence.

The converse of this theorem is not true: a convergent sequence does not need
to be monotone, but it needs to be bounded.

(−1)n
Example. The sequence xn = is convergent, but not monotone.
n+1
1 1 1
The limit is 0. Indeed, |xn −0| = < ε as soon as n+1 > , i.e. n > −1.
n+1 ε ε
1
We can choose n0 > −1. Then, the condition |xn −0| < ε is fulfilled for all n ≥ n0 .
ε
On the other hand, clearly (xn ) is not monotone. Actually, its values oscillate
around 0.

Theorem. Any convergent sequence is bounded.

Proof: If lim xn = ℓ, and ε > 0 is arbitrary, according to the definition of a con-
n→∞
vergent sequence, there exists an n0 such that we have xn ∈ (ℓ − ε, ℓ + ε), ∀ n0 .
Denoting m = min{ℓ − ε, x0 , x1 , . . . , xn0 −1 } and M = max{ℓ + ε, x0 , x1 , . . . , xn0 −1 },
we have m ≤ xn ≤ M , f orall n ≥ 0, i.e. the sequence is bounded.

Definition. We say that lim xn = ∞ if ∀ M , ∃n0 such that n ≥ n0 ⇒ xn > M .

n→∞
We say that lim xn = −∞ if ∀ M , ∃n0 such that n ≥ n0 ⇒ xn < M .
n→∞
S
Thus, the set of all possible limits of a sequence is R = R {−∞, ∞}, called
the extended real line.
A divergent sequence either doesn’t have any limit or has limit −∞ or +∞.

Operations with limits. Indeterminate forms

Theorem. (Operations with limits)

If (xn ) and (yn ) are two sequences such that lim xn = x ∈ R and lim yn = y ∈ R,
n→∞ n→∞

1
named after Karl Weierstrass

4
then the following limits exist
then the following limits exist
lim (xn + yn ) = x + y
n→∞
lim(x(xn n+· yynn))==xx+· y
lim
n→∞
n→∞
limxn(xn · x
yn ) = x · y
limn→∞ = (if yn ̸= 0)
n→∞ x
yn xy
lim ynn = y (if yn 6= 0)
lim
n→∞xy nn = x y (if xn > 0)
n→∞
lim xynn = xy (if xn > 0)
n→∞
if the operations on the right hand side are defined.
if the operations on the right hand side are defined.
These operations have to be understood as follows:
These operations have to be understood as follows:

How this works is explained on an easy example here.

How this works is explained on an easy example here.
However, some operations are not defined. These operations are called "inde-
However,forms".
terminate some operations
They are: are not defined. These operations are called "inde-
terminate forms". They are:
∞ 0
∞ − ∞, , , 0 · ∞, 00 , ∞0 , 1∞ .
∞
∞ 0 0 0 ∞
∞ − ∞, , , 0 · ∞, 0 , ∞ , 1 .
∞ 0 the outcome in each of these cases is not
They are called indeterminate because
determined in advance, it depends from case to case. For example, take the inde-
They are called indeterminate because the outcome in each of these cases is not
terminate form 0 · ∞. In the table below are given examples of two sequences, (xn )
determined in advance, it depends from case to case. For example, take the inde-
terminate form 0 · ∞. In the table below are given examples of two sequences, (xn )

5
and (yn ), such that lim xn = 0, lim yn = ∞, but lim (xn · yn ) has various results:
n→∞ n→∞ n→∞

xn yn xn · y n lim (xn · yn )
n→∞
1 n 1 1
n
a
n a a
n
1 n2 n ∞
n
1 n2 −n −∞
−
n
(−1)n n (−1)n ∄
n

You can see that the outcome of such indeterminate forms is unpredictable. The
limit can be any real number a, plus or minus infinity, but it is also possible that
it doesn’t exist at all.

When dealing with such forms, usually one needs to rewrite the general term
of the sequence under a different, often not very natural, way.

Worked exercises, with detailed and well explained solutions, can be found
here. The results of Exercises 3 and 7 must be known. They can be used without
proof.

 n
1
The sequence en = 1+
n
 n
1
It is known that the sequence en = 1 + is convergent and lim en = e ≈ 2, 71.
n n→∞
1
More generally, if (xn ) is a sequence such that lim xn = 0, then lim (1+xn ) xn = e.
n→∞ n→∞
We use this fact for indeterminate forms of the type 1∞ .

Indeterminate forms of the type 1∞

Further useful theorems
Theorem. (Squeeze theorem - Teorema cleştelui)If (an ), (bn ) and (cn ) are three
sequences such that an ≤ bn ≤ cn , ∀ n ≥ n0 , and lim an = lim cn = ℓ, then
n→∞ n→∞
lim bn also exists and is equal to ℓ.
n→∞

6
 2 + (−1)n
Example. We cannot determine the limit of the sequence bn =
n
simply using "operations with limits" because the numerator does not have any
1 3
limit. But we can notice that 1 ≤ 2 + (−1)n ≤ 3, therefore we have ≤ bn ≤
n n
1 3
and lim = lim = 0, therefore lim an = 0.
n→∞ n n→∞ n n→∞

Sometimes this theorem is called "the two policemen theorem": if two police-
men (an and cn ) go to the police headquarter, and a drunk, (bn ), stands between
them, then the drunk must also go to the same police headquarter.

A useful variant for the case ℓ = ∞: in bn ≥ an , ∀ n ≥ n0 , and lim an = ∞,

n→∞
then lim bn = ∞.
n→∞

Example. As n + (−1)n ≥ n − 1, ∀ n and lim (n − 1) = ∞, we have lim n +

n→∞ n→∞
(−1)n = ∞.

A useful consequence of the previous theorem:

Corollary. If the sequence (an ) is bounded and the sequence (bn ) is convergent
to 0, then their product, (an · bn ) is also convergent to 0.
1
Example. an = (−1)n is bounded, bn = has lim bn = 0, hence,
n+1 n→∞
(−1)n
lim an · bn = lim = 0.
n→∞ n→∞ n + 1

Cauchy sequences
Definition. A sequence (xn ) is called a Cauchy sequence (şir Cauchy sau şir fun-
damental) if it satisfies the following condition: ∀ ε > 0 ∃n0 ∈ N : n, m ≥ n0 ⇒
|xn − xm | < ε.

Theorem. A sequence is a Cauchy sequence if and only if it is convergent.

Why do we need this concept if it is equivalent to "convergent sequence" ?

First, because in some other frames the two concepts are no longer equivalent, but
the main reason is that sometimes it is easier to check that a sequence is (or isn’t)
a Cauchy sequence than proving that it is (isn’t) convergent. For the convergence,
one needs to know the limit.

Example. see here

7
0.2 Numerical Series (Serii numerice)
Starting from a given sequence (an )n≥n0 one can define, by summation, a new
sequence, (sn )n≥n0 , as follows:
n
X
sn = ak , n ≥ n0 .
k=n0

For example, if an = n, n ≥ 1, i.e. for the sequence 1, 2, 3, 4, ... one obtains the
following sequence: s1 = 1, s2 = 1 + 2, s3 = 1 + 2 + 3, s4 = 1 + 2 + 3 + 4, ... , sn =
n(n + 1)
1 + 2 + 3 + ... + n, i.e. the sequence: 1, 3, 6, 10, ... , in short, sn = , n ≥ 1.
2
Definition 1. (Series) Consider n0 ∈ N and (an )n≥n0 a sequence in R. The sum
Xn
sn = ak , n ≥ n0 , is called the n-th partial sum of this sequence.
k=n0
The sequence (sn )n≥n0 of the partial sums is called a series.
X∞ X
We denote a series by an or an .
n=n0 n≥n0
an is the general term of the sequence.
∞
X
Definition 2. (convergent series) The series an is called convergent,
n=n0
if the sequence (sn )n≥n0 of the partial sums is convergent, otherwise it is called
divergent.
∞
X
The limit of the sequence (sn ) is the sum of the series and it is also denoted an .
n=n0

Example 1. (the geometric series)

Let b ∈ R, b ̸= 0 and consider the geometric progression an = bn , n ≥ 0. We look
n
X
at the sequence of its partial sums (sn )n≥n0 , sn = ak . It is well known that
k=0

n+1
 1−b

if b ̸= 1
sn = 1−b

n + 1, if b = 1
X
The series bn is convergent is the sequence (sn )n≥0 is convergent. As n + 1 is
n≥0
1 − bn+1
divergent and is convergent if and only if b ∈ (−1, 1), it follows that
1−b
X
the series bn is convergent if and only if b ∈ (−1, 1). In this case we can also
n≥0

8
compute the sum of this series:
∞ n
X
n
X 1 − bn+1 1
b = lim bk = lim = .
n→∞ n→∞ 1 − b 1−b
n≥0 k=0

X1
Example 2. 2
is a convergent series. Indeed, we show that this sequence is
n≥1
n
increasing and bounded, hence convergent.
n
X 1
The sequence of the partial sums, sn = 2
is increasing because
k=1
k
   
1 1 1 1 1 1 1 1
sn+1 − sn = 2
+ 2 + ... + 2 + 2
− 2 + 2 + ... + 2 = > 0.
1 2 n (n + 1) 1 2 n (n + 1)2
1 1 1 1
This sequence is also bounded because sn ≥ s1 = 1 and sn = 2 + 2 + 2 +...+ 2 ≤
1 2 3 n
     
1 1 1 1 1 1 1 1 1 1
+ + + ... + = 1+ − + − + ... + − =
1 1·2 2·3 (n − 1)n 1 2 2 3 n−1 n
1
2 − < 2.
n
1 1 1
We have used the equality = − .
k(k + 1) k k+1
Although it has been easy to establish the fact that this series is convergent, it is
much more complicated to compute its sum. It is enough to state that its sum is
X1 π2
= , to see that discovering it cannot be a simple task.
n≥1
n2 6
The above situation is typical in the study of series: if one cannot provide a
closed formula for (sn )n≥n0 then it is rarely possible to determine the sum of the
series. This is why, when studying the convergence of series, usually we are only
interested in the, "nature" of the series: "is the series convergent or divergent?"

For the convergence of a series is the "behavior at the beginning" unim-

portant; this means:

Theorem 1. (behavior at the beginning)

X
For an arbitrary m ≥ n0 the series an is convergent if and only if the series
n≥m
X
an is convergent.
n≥n
Proof:
0 The sequence of the partial sums of the second series is sn = an0 + an0 +1 +
... + am−1 + am + ... + an , or sn = an0 + ... + am−1 + s′n where (s′n )n≥m is the
sequence of the partial sums of the first series. Then it is easy to see that (sn )n≥n0
is convergent if and only if (s′n )n≥m is convergent and that, in this case, lim sn =
n→∞

9
 ∞
X ∞
X
an0 + ... + am−1 + lim s′n , hence an = an0 + ... + am−1 + an . □
n→∞
n=n0 n=m
∞
X ∞
X
Theorem 2. Consider an and bn two convergent series and α ∈ R. The
n=n0 n=n0
∞
X ∞
X ∞
X
series (an + bn ) and αan are also convergent and we have (an + bn ) =
n=n0 n=n0 n=n0
∞
X ∞
X ∞
X ∞
X
an + bn and αan = α an .
n=n0 n=n0 n=n0 n=n0
n
X Xn
Proof: Let sn = ak , s′n = bk , n ≥ n0 and ℓ1 = lim sn , ℓ2 = lim s′n .
n→∞ n→∞
k=n0 k=n0
It is known that the sequences (sn + s′n )n≥n0 and (αsn )n≥n0 are also convergent
sequencesand that lim (sn + s′n ) = lim sn + lim s′n = ℓ1 + ℓ2 and lim (αsn ) =
n→∞ n→∞ n→∞ n→∞
α · lim sn = α · ℓ1 . □
n→∞

Theorem 3. (Necessary condition for convergence)

X
If an is a convergent series, then the sequence (an )n≥n0 is convergent and
n≥n0
lim an = 0.
n→∞
Proof: The sequence of the partial sums satisfies sn = an0 + an0 +1 + ... + an−1 +an =
| {z }
sn−1 + an . sn−1
X X∞
If an is a convergent series whose sum is S = an , then we have lim an =
n→∞
n≥n0 n=n0

lim (sn − sn−1 ) = S − S = 0. □

n→∞

This above condition ( lim an = 0) is therefore a necessary condition for the

n→∞
convergence of a series, but not a sufficient condition.
X n n
Example 3. The series is divergent because an = −→
n≥0
1000n + 1 1000n + 1
1
̸= 0.
1000
X
Example 4. The series sin n is divergent because the limit lim an does
n→∞
n≥0
not exist.
X 1 1
Example 5. The series √ √ is divergent although lim √ √ =
n≥0
n+ n+1 n→∞ n+ n+1
0.

10
 n n √ √
X 1 X k− k+1
Indeed, sn = √ √ = √ √ √ √ =
k≥0
k+ k + 1 k≥0 ( k + k + 1)( k − k + 1)
n √ √ n
X k− k+1 X √ √
= − ( k − k + 1) =
k≥0
k − (k + 1) k≥0
√ √ √ √ √ √ √ √ √ √
− [( 0 − 1) + ( 1 − 2) + ( 2 − 3) + ... + ( n − n + 1)] = −( 0 − n + 1) =
√
n + 1 −→ ∞.
In conclusion, the converse of the Theorem above is not true, in other words, from
∞
X
the fact that lim an = 0 it does not follow that the series an is convergent.
n→∞
n=n0
X
Theorem 4. (Cauchy) Let an be a series in R. The following statements
n≥n0
are equivalent:
∞
X
(a) The series an is convergent.
n=n0
n
X
(b) For all ε > 0 there is an N0 (ε) ∈ N such that ak < ε, ∀ n ≥ m ≥ N0 (ε).
k=m+1
n
X
Proof: For the sequence of the partial sums we have sn − sm = ak . Condition
k=m+1
(b) is, thus, equivalent to the fact that (sn )n≥n0 is a Cauchy-sequence. According
to last Theorem of Chapter 1, any Cauchy-sequence in R is a convergent sequence,
therefore b) is equivalent to the fact that (sn )n≥n0 is convergent. □
∞
X 1
Example 6. The series is called the harmonic series. It is divergent be-
n=1
n
cause the condition (b) is not satisfied. Indeed, writing (b) for n = 2m ≥ m and
1
ε = we get:
2
2m 2m
X 1 X 1 1 1 1 1 1 1 1
= = + +...+ ≥ + +...+ = m· =
k=m+1
k k=m+1
k m + 1 m + 2 2m 2m 2m 2m 2m
1
.
2 n
1 X
Thus, for ε = there is no N0 (ε) that satisfies the condition ak < ε for all
2 k=m+1
n ≥ m ≥ N0 (ε) .

The harmonic series is a second example of a series that satisfies the condition
lim an = 0 but is, nevertheless, divergent.
n→∞

11
SERIES WITH POSITIVE GENERAL TERMS
The case when an > 0, ∀ n ≥ n0 is especially important: first, because these
series occur very frequently, and, secondly, because these series are easier to study:
from sn = sn−1 + an it follows that sn − sn−1 = an > 0, therefore the sequence of
the partial sums is an increasing sequence.
if die series (sn )n≥n0 is also bounded, then it is convergent, meaning that in this
X
case the series an is convergent.
n≥n0
if the sequence (sn )n≥n0 is unbounded, then we have lim sn = +∞, meaning that
n→∞
X
in this case the series an is divergent.
n≥n0

Thus, for series with positive general terms, the question is:
X
”Is the series an convergent or does it have the limit +∞ ?"
n≥n0

Unfortunately, there is no general method to decide the answer to the question

above. However, there are some criteria, some "convergence tests", that one can
apply in order to try finding an answer.

1. The Comparison Test (CT)

Theorem 5. (Comparison Test)

X X
Let an and bn two series that satisfy 0 < an ≤ bn , ∀ n ≥ n0 . Then:
n≥n1 n≥n2
X X
• if the series bn is convergent, then the series an is convergent as well;
n≥n2 n≥n1
X X
• if the series an is divergent, then the series bn is divergent as well.
n≥n2 n≥n1
X
Proof: If the series bn is convergent, then the sequence (s′n )n≥n0 ,
n≥n2
n
X
s′n = bk is convergent, hence bounded. But, as ak ≤ bk , ∀ k ≥ n0 , the sequence
k=n0 n n
X X
(sn )n≥n0 of the partial sums sn = ak ≤ bk is also bounded. It si also in-
k=n0 k=n0
X
creasing, therefore it is also convergent. The fact that an is convergent means
n≥n0

12
 X
that the series an is also convergent (according to the theorem "Behavior at the
n≥n1
beginning", it is unimportant to know which is the first term of the sequence). For
X
the second part, assume by contradiction that bn is convergent. Then, from the
n≥n2
X
first part of the statement, the one we have already proven, it follows that an
n≥n1
is also convergent, which contradicts the hypothesis. □
X1
Example 7. In Example 2. we have proven that 2
is convergent. We have
n≥1
n
1 X
done this by first proving that the series is convergent, and, hence,
n≥2
n(n − 1)
1 1 X1
as 0 < an = 2 ≤ = bn , ∀ n ≥ 2, we obtained that 2
was also
n n(n − 1) n≥1
n
convergent.
X 1
Example 8. The series √ is divergent.
n≥1
n
1 1 X1
Proof (Method 1.): 0 < ≤ √ , ∀ n ≥ 1 and is divergent (Example 6.).
n n n≥1
n
X 1
From the comparison test it follows that √ is divergent.
n≥1
n
1 1 X 1
Proof (Method 2.): 0 < √ √ ≤ √ , ∀ n ≥ 1 and √ √ is
n+ n+1 2 n n≥1
n + n + 1
X1 1
divergent (Example 5.). From the comparison test it follows that · √ is
n≥1
2 n
X 1
divergent. By Theorem 2. (with α = 2) we get that √ is also divergent.
n≥1
n
X1
Remark. As by Method 1. one can prove that is divergent for all α ≤ 1.
n≥1
nα
X1
Example 9. α
is convergent for all α ≥ 2.
n≥1
n
1 1 X1
Proof: 0 < α ≤ 2 , ∀ n ≥ 1 and is convergent, hence, by the comparison
n n n≥1
n2
X1
test, the series α
is also convergent, ∀ α ≥ 2.
n≥1
n

13
 X1
Conclusion: α
is convergent if α ≥ 2 and divergent if α ≤ 1. We shall see
n≥1
n
1 1 1 1
later what happens when α ∈ (1, 2). (In this case, 0 < 2 ≤ α and 0 < α ≤ ,
n n n n
but none of these inequalities is any good when trying to apply the comparison
X1
test in order to decide whether the series α
convergent or divergent is.)
n≥1
n
This shows us well the limitations of this convergence test::
X X
if 0 < an ≤ bn , ∀ n ≥ n0 and an is convergent or if bn is divergent, nothing
n≥n0 n≥n0
follows from the comparison test. We need other convergence tests.

2. The Limit Comparison Test (LCT)

Theorem 6. (Limit Comparison Test)

X X an
Let an and bn be two series with an , bn > 0, ∀ n ≥ n0 . If the limit lim
n→∞ bn
n≥n1 n≥n2
an X X
exists and lim = ℓ ∈ (0, ∞) then the two series, an and bn , have the
n→∞ bn
n≥n n≥n
1 2

same nature, i.e. either they are both convergent, or they are both divergent.
an
Proof: Let ε > 0 be such that ℓ − ε > 0. As lim = ℓ, there exists an n′0 ≥ n0
n→∞ bn
an
such that ℓ − ε < < ℓ + ε, ∀ n ≥ n′0 . It follows that for n ≥ n′0 , we have
bn
X X
(ℓ − ε)bn < an < (ℓ + ε)bn . If an is convergent, then so is (ℓ − ε)bn (CT)),
n≥n1 n≥n2
X X X
i.e. the series bn is convergent. If an is divergent, then so is (ℓ + ε)bn
n≥n2 n≥n1 n≥n2
X
*CT), hence the series bn is also divergent. □
n≥n2
an an
What happens if lim = 0 or lim = ∞?
bn n→∞ n→∞ bn
an X X
• lim = 0 that an is "much smaller" than bn . The two series do not
n→∞ bb
n≥n1 n≥n2
X X
always have the same nature: it is possible that an is convergent and bn is
n≥n1 n≥n2
divergent. Nevertheless:
X X
- if bn is convergent, then so is an ;
n≥n2 n≥n1

14
 X X
- if an is divergent, then so is bn .
n≥n1 n≥n2
an
EIn fact, if lim = 0, then we have 0 < an ≤ bn , ∀ n ≥ n0 and we can apply the
n→∞ bn
Comparison Test.
an X X
• lim = ∞ shows that an is "much larger" than bn . Again, it is
n→∞ bb
n≥n1 n≥n2
X
not always true that the two series have the same nature: it is possible that an
n≥n1
X
is divergent and bn is convergent. Nevertheless:
n≥n2
X X
- if bn is divergent, so is an
n≥n2 n≥n1
X X
- if an is convergent, so is bn .
n≥n1 n≥n2
an
In fact, if lim = ∞, then we have 0 < bn ≤ an , ∀ n ≥ n0 and one can apply
n→∞ bn
the Comparison Test.
X 2n2 + 3n + 4
Example 10. The series is convergent.
n≥0
5n5 + 6
X 2n2 + 3n + 4 X1
We compare to (we shall see later why we have chosen
5n5 + 6
n≥0 n≥1
n 3

X1 2n2 + 3n + 4 1
precisely the series 3
). For a n = 5
and bn = 3 we compute
n≥1
n 5n + 6 n
2n2 + 3n + 4
an 5n5 + 6 2n2 + 3n + 4 n3 2n5 + 3n4 + 4n3
lim = lim = lim · = lim =
n→∞ bn n→∞ 1 n→∞ 5n5 + 6 1 n→∞ 5n5 + 6
n3
2
∈ (0, ∞).
5
X 2n2 + 3n + 4 X1
By the LCT it follows that the series and have the same
n≥0
5n5 + 6 n≥1
n3
X 
1 1
nature. As 3 with α = 3 ≥ 2 is convergent (see Example 9.), it follows
n n≥1
nα
X 2n2 + 3n + 4
that 5+6
is convergent as well.
n≥0
5n

Remarks:

15
 X 2n2 + 3n + 4 X1
1. Why did we choose to compare exactly to ?
n≥0
5n5 + 6 n≥1
n3

if n is large, the behaviors of 2n2 + 3n + 4 and 5n5 + 6 are similar to those of 2n2
2n2 2 1
and 5n5 , respectively, meaning that an behaves like 5
= · 3.
5n 5 n
1
In general: if an is the ratio of two polynomials, then we compare always with α
n
where α is the difference between the degrees of the two polynomials:
ck nk + ck−1 nk−1 + ... + c1 n + c0 1
if an = j j−1
, we compare it to bn = j−k . When we
dj n + dj−1 n + ... + d1 n + d0 n
k k−1 j−k
an ck n + ck−1 n + ... + c1 n + c0 n ck nj + ck−1 nj−1 + ... + c1
compute lim , we get lim · = lim
n→∞ bn n→∞ dj nj + dj−1 nj−1 + ... + d1 n + d0 1 n→∞ dj nj + dn−1 nj−1 + ..
ck
∈ (0, ∞).
dj
X X
2. Usually one compares an with series of the form bn whose nature is

already known. Otherwise, the LCT is only useful if one can afterwards decide the
X
nature of the series bn .

The method described at Remark 1. also works in other similar cases:

√
X n + 2 4 n9 + 1
Example 11. The series √ is convergent.
n≥0
n3 + n4 n + 3
√
4 √
As n + 2 n9 + 1 ≃ 2n9/4 and n3 + n4 n + 3 ≃ n9/2 (for n large), one can compare
√4
n + 2 n9 + 1 2n9/4 1 1
an = 3 √ ≃ = 2 · , to . Indeed:
n + n4 n + 3 n9/2 n9/4 n9/4
√4
n + 2 n9 + 1 √
√ 4
n3 +n4 n+3 n + 2 n9 + 1 n9/4 n1+9/4 + 2n9/4+9/4 + n9/4
lim = lim 3 4 √ · = lim =
n→∞ 1 n→∞ n +n n+3 1 n→∞ n3 + n9/2 + 1
n9/4  
9/2 1 1
n + 2 + 9/4
n13/4 +2n9/2 +n9/4 n5/4 n
lim 3 9/2
= lim   = 2 ∈ (0, ∞).
n→∞ n +n +1 n→∞ 1 1
n9/2 3/2 + 1 + 9/2
n n
X 1  X1 
9 X
As 9/4
is convergent, α
mit α = > 2 , is an also convergent.
n≥1
n n≥1
n 4

Conclusion (advantages and disadvantages of the LCT)

16
 advantages:
1. It allows to settle the nature of some types of series.
2. Has a broad applicability.
disadvantages:
X
1. One has to guess in advance the series bn to which one chooses to compare
X
the series an to.

2. One can use for comparison only those series for which the nature is already
known or easy to determine.

3. The Ratio Test (RT) (or d’Alembert’s Test)

Theorem 7. (Ratio Test)

X
Let an be a series with an > 0, ∀ n ≥ n1 .
n≥n1
an+1
If the limit lim = ℓ exists, then:
an
n→∞
X
a) if ℓ < 1 then the series an is convergent;
n≥n1
X
b) if ℓ > 1 then the series an is divergent;
n≥n1
c) if ℓ = 1 we can not decide with this convergence
test
X whether the series Jean-Baptiste le Rond
an is convergent or divergent. d’Alembert (1717-1783)
n≥n1

Proof: a) Let ε > 0 be such that ℓ − ε > 0 and ℓ + ε < 1.

an+1
Then there exists an N0 = N0 (ε) such that 0 < ℓ − ε < < ℓ + ε, ∀ n ≥ N0 .
an
Consider n ≥ N0 . We have:

17
 aN0 +1
0<ℓ−ε< <ℓ+ε
aN0
aN +2
0<ℓ−ε< 0 <ℓ+ε
aN0 +1
..
.
an
0<ℓ−ε< < ℓ + ε.
an−1
Multiplying together these n − N0 inequalities, we obtain
aN +1 aN +2 aN +3 an−1 an
0 < (ℓ − ε)n−N0 < 0 · 0 · 0 · ... · · < (ℓ + ε)n−N0 , hence
aN0 aN0 +1 aN0 +2 an−2 an−1
an
0 < (ℓ − ε)n−N0 < < (ℓ + ε)n−N0 .
aN 0
aN 0 X
It follows that an < bn = · (ℓ + ε)n , ∀ n ≥ N0 . The series bn
(ℓ + ε)N0 n≥N0
is convergent (see Example 1. with b = ℓ + ε < 1), therefore, according to the
X X
Comparison Test, the series an is convergent as well, and so is an .
n≥N0 n≥n1

b) Consider ε > 0 such that ℓ − ε > 1. As by a), there exists a rank N0 = N0 (ε)
such that an
0 < (ℓ − ε)n−N0 < < (ℓ + ε)n−N0 , ∀ n ≥ N0 .
aN 0
aN0 n
X
It follows that bn = · (ℓ − ε) < a n . But (ℓ − ε)n is divergent (see
(ℓ − ε)N0 n≥N 0
Example 1. with b = ℓ − ε > 1), therefore, according to the Comparison Test, the
X X
series an is divergent, and so is an .
n≥N
0 n≥n1
c) We shall prove the fact that it is possible for the series
X
1) an to be convergent
X
2) an to be divergent
an+1
if lim = 1, by giving an Example for each of the two cases.
n→∞ an
1
1 an+1 (n + 1)2 n2
1) Consider an = 2 . We have lim = lim = lim 2 =1
n n→∞ an n→∞ 1 n→∞ n + 2n + 1
n2
X
nd an is convergent.
1
1 an+1 n
= lim n + 1 = lim
X
2) Consider an = . We have lim = 1 and an
n n→∞ an n→∞ 1 n→∞ n + 1
is divergent. n □

18
 X 2n
Example 12. is convergent because:
n≥0
n!
2n+1
an+1 (n + 1)! 2n+1 n! 2 QK
lim = lim n = lim · n = lim = 0 < 1 =⇒
n→∞ an n→∞ 2 n→∞ (n + 1)! 2 n→∞ n + 1
n!
X 2n
- convergent.
n≥0
n!
X nn
Example 13. is divergent because:
n≥0
2n · n!
(n + 1)n+1
an+1 2n+1 · (n + 1)! (n + 1)n+1 2n · n! (n + 1)n
lim = lim = lim · = lim =
n→∞ an n→∞ nn n→∞ 2n+1 · (n + 1)! nn n→∞ 2nn
n 2n · n!
1

1 e QK
X nn
· lim 1 + = > 1 =⇒ - divergent.
2 n→∞ n 2 n≥0
2n · n!

Conclusion (advantages and disadvantages of the Ratio Test)

advantages:
1. It is self-contained and very easy to apply: imagination is not required.
2. It is specially well suited to the cases when an contains factors like n! or an .
Polynomial factors do bother, but do not help either.
disadvantages:
1. The limit ℓ will often be equal to 1. In that case one can not use the Ratio Test.
P (n)
This is always the case when an is a rational expression, i.e. an = , where
Q(n)
P, Q are Polynoials. In this case, the limit ℓ is always 1.
Nevertheless, as we shall see later, the Ratio Test has a natural sequel, namely
Raabe’s Test.

19
 √
4. The Root Test ( T) (or Cauchy’s Test)

Theorem 7. (Root Test)

X
Let an be a series with an > 0, ∀ n ≥ n1 .
n≥n1
√
If the limit lim n a = ℓ exists, then:
n
n→∞
X
a) if ℓ < 1 then the series an is convergent;
n≥n1
X
b) if ℓ > 1 then the series an is divergent;
n≥n1
c) if ℓ = 1 then the nature of the series
X
an can not be decided by this convergence test.
n≥n1
Baron Augustin-Louis
(The series could be convergent or divergent.) Cauchy (1789 − 1857)
Proof: a) Let ε > 0 be such that ℓ − ε > 0 and ℓ + ε < 1.
√
Then there exists a rank N0 = N0 (ε) such that 0 < ℓ − ε < n an < ℓ + ε, ∀ n ≥ N0 .
Take n ≥ N0 . Then we have:
an < bn = (ℓ + ε)n ∀ n ≥ N0 .
X
The series bn is convergent (see Example 1. with b = ℓ + ε < 1), hence, by
n≥N0
X X
the Comparison Test, the series an is convergent and therefore an is also
n≥N0 n≥n1
convergent.
b) Let ε > 0 be such that ℓ − ε > 1. Exactly as by a), there is a rank N0 = N0 (ε)
such that
0 < bn = (ℓ − ε)n < an , ∀ n ≥ N0 .
X
But (ℓ − ε)n is divergent (Example 1. with b = ℓ − ε > 1), hence, by the
n≥N0
X X
Comparison Test, the series an is divergent and therefore the series an is
n≥N0 n≥n1
divergent as well.
1 1 √ X1
c) For an = , and an = 2 we have lim n an = 1, but the series is conver-
n n n→∞ n
X1
gent while is divergent. □
n2
X n n2
Example 14. is convergent because
n+1

20
 s n2  n
n n n 1 1
lim = lim = lim  = < 1.
n→∞ n+1 n→∞ n+1 n→∞ 1 n e
1+
n
Remark: For the above series one could also apply the Ratio Test but com-
an+1
puting the limit lim is more difficult. Applying the Root Test is much more
n→∞ an
comfortable.

Remark (The connection between the Ratio Test and the Root Test)
an+1
If, for a series |disan , the Ratio Test works (this means ℓ = lim < 1 in
n→∞ an
case the series is convergent or ℓ > 1 in case it is divergent) then the Root Test
√
woks as well (meaning that we shall obtain ℓ′ = lim n an < 1 in case the series is
n→∞
convergent and ℓ′ > 1 im case it is divergent). In fact, we shall obtain ℓ′ = ℓ.
This is certified by the Cauchy-d’Alembert Theorem:
an+1
Theorem of Cauchy-d’Alembert: if an > 0, ∀ n and the limit lim =ℓ
n→∞ an
√
exists, then the limit lim n an exists as well and
n→∞

√ an+1
lim n
an = lim = ℓ.
n→∞ n→∞ an

X
Conclusion: If the Ratio Test fails to decide the nature of the series an
an+1 √
giving lim = 1, then the Root Test will fail as well because lim n an = 1.
an
n→∞ n→∞

an+1
Nevertheless, it is possible that the limit lim does not exist, and this is
n→∞ an
√
why we can not apply the Ratio Test, while the limit lim n an ̸= 1 does exist and
n→∞
the Root Test works. It is the case in the following
1 4 a2n+1
Example 15. If a2n = 2n
and a2n+1 = 2n+1 then we have = 2 and
2 2 a2n
a2n+2 1 an+1
= . In this case, the limit lim does not exist, so we can not apply the
a2n+1 8 n→∞ an
Ratio Test.
The Root Test, however, works because we have
√ √ 1 √ 1
lim 2n a2n = lim 2n+1 a2n+1 = and therefore lim n an = < 1.
n→∞ n→∞ 2 n→∞
X 2
By the Root Test we obtain the convergence of an .

The Root Test is stronger then the Ratio Test.

21
 Xn
Example 16. Die series is convergent.
2n
n+1
an+1 n+1 n + 1 2n
with the Ratio Test: lim = lim 2 n = lim n+1 · =
n→∞ an n→∞ n→∞ 2 n
2n X
n+1 1
lim = = ℓ. As ℓ < 1, it follows that an is convergent.
n→∞ 2n 2
√ √
√ n
n n
n 1
with the Root Test: lim an = lim √
n
n
= lim = = ℓ.
n→∞ n→∞ 2n n→∞ 2 2
X
As ℓ < 1, it follows that an is convergent.
an+1
As stated by the Cauchy-d’Alembert Theorem, in case the limit lim exists,
n→∞ an
an+1 √ 1
the limits lim and lim n an are equal. In this example both limits are .
n→∞ an n→∞ 2
Conclusion (advantages and disadvantages of the Root Test)

advantages:
1. As for the Ratio Test, the Root Test is self-contained and can be applied directly.
Imagination is not required.
2. It is very appropriate for the case when an is of the form an = (bn )n . In this case
√n a = b
√
n and the limit lim
n a
n n is easy to compute
n→∞
disadvantages:
1. As for the Ratio Test, we often get ℓ = 1, meaning that the Root Test can
not be applied. This is always the situation when an is a rational expression, i.e.
P (n)
something of the form , where P, Q are polynomials. The limit ℓ is always 1
Q(n)
for such series. √
2. Often it happens that the computation of the limit n an is not easy. If comput-
an+1
ing this limit is not obviously easier than computing the limit lim , then we
n→∞ an
recommend using the Ratio Test instead.

5. Cauchy’s Condensation Test

Theorem 8. (Condensation Test)

Ist (an )n≥1 eine monoton decreasinge Folge positiver reeller Zahlen, so konvergiert
X X
die series an dann und genau dann, if die series 2n · a2n konvergiert.
n≥1 n≥0
X X
(Mit andere W ,orte, haben die seriesn an und 2n · a2n dieselbe Natur.)
n≥1 n≥0

22
Proof: Als die Folge der Partialsummen increasing ist (weil sn+1 = sn + an > sn ),
X
ist die series an convergent if und genau if die Folge der Partialsummen (sn )n≥1

auch bounded ist. Es gielt einerseits:

Xn
s2n+1 −1 = a1 + a2 + a3 + a4 + a5 + a6 + a7 +...+a2n + a2n +1 + ... + a2n+1 −1 ≤ 2m ·
|{z} | {z } | {z } | {z }
a2m . =1·a1 ≤2·a2 ≤4·a4 ≤2n ·a2n m=0
X
if die series 2n · a2n convergent ist, dann ist die Folge seiner Partialsummen,
n≥0
Xn
(tn )n≥0 , tn = 2m · a2m bounded.
m=0

Von der obigen Berechnung folgt, dass die Folge s2n+1 −1 n≥0
auch bounded ist. Aus
0 ≤ sn ≤ s2n+1 −1 ≤ M , folgt dass die Folge der Partialsummen (sn )n≥1 bounded
X
ist. Als sie auch increasing ist, ist sie convergent, also ist die series an conver-
gent. n≥1

Umgekehrt gilt:
s2n = a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 +... + a2n−1 +1 + a2n−1 +2 + ... + a2n ≥
|{z} | {z } | {z } | {z }
=1·a2 ≥2·a4 ≥4·a8 ≥2n−1 ·a2n
n
1 X m 1
· 2 · a2m = · tn .
2 m=0 2
X
Ist die series 2m · a2m divergent, so ist die Folge seiner Partialsummen unbe-
m≥0
schr ,ankt und deshalb lim tn = ∞. Es folgt dass lim s2n = ∞, also ist im diesem
n→∞ n→∞
X
Fall die series an divergent. □
n≥1

Dieser Theorem sagt dass Falls an > 0 und (an )n≥1 decreasing ist dann ist die
X
Konvergenz der series an nur von wenige terms der Folge (an )n≥1 entschieden,
n≥1
n ,ahmlich von a2m , m ≥ 0. Aber, w ,ahrend es nur wenige terms z ,ahlen, haben
diese terms eine je gr ,ossere Gewicht (a2n wird 2n mal gez ,ahlt).

Example 17. (Harmonic series)

X1
Wir haber schon erw ,ahnt dass die harmonische seriesn α
, α > 0, sehr im-
n≥1
n
portant sind indem man sie, zusammen mit den Vergleichungskriterium anwenden
kann um die Natur einer series zu bestimmen. Jetzt sind wir in der Lage die Natur
dieser harmonischen seriesn endg ,ultig zu bestimmen:

23
 X1 X 1
Laut das Kondensationskriteriums, haben die seriesn α
und 2n n α
n≥1
n n≥0
(2 )
X 1 X
1−α n
2n


dieselbe Natur. Aber = 2 ist convergent if und genau if
n≥0
(2n )α n≥0
−1 < 21−α < 1, also if und genau if α > 1.

Est ist important sich zu merken dass:

X 1
α
ist convergent if α > 1
n≥1
n
divergent if α ≤ 1.

X 1
Example 18. Bestimme die Natur der series .
n≥2
n ln n
Man kann versuchen die Natur dieser series mit das Vergleichungskriterium bes-
1 1 1
timmen. Als es f ,ur alle α > 1 ein N0 gibt so dass α < < , ∀ n ≥ N0 ,
n n ln n n
X 1
mann kann es sehen dass die Vergleichung der series mit eine harmonis-
n≥2
n ln n
che series nichts gibt. Also ist das Vergleichungskriterium nicht anwendbar.
Hier hilft das Kondensationskriterium:
1
Als n und ln n increasinge un positive Folgen sind, ist decreasing. Dann hat
n ln n
X 1 X 1 X 1 1 X1 X1
dieselbe Natur wie 2n · n = = . Als
n≥2
n ln n n≥1
2 · ln(2n ) n≥1 n ln 2 ln 2 n≥1 n n≥1
n
X 1
divergent ist, folgt dass auch divergent ist.
n≥2
n ln n

Konklusion (advantages und disadvantages des Kondensationskri-

teriums)

advantages:
Dieses Kriterium ist sehr leicht anwendbar. Es ist besonderst geeignet if log n
(ln n, lg n) vorkommt denn log n wird durch die Substitution n 7→ 2n in n log 2
verwandelt (wie es der Fall war im obigen Example).
disadvantages:
Manchmal hat die zweite series ein sehr komplizierten Ausdruck und if wir dieses
Kriterium anwenden, machen wir keinen Fortschritt.

24
 6. Kriterium von Raabe (Raabesche Kriterium)

Theorem 6. (Kriterium von Raabe)

X
Sei an eine series mit an > 0, ∀ n ≥ n1 .
n≥n1
 
an
if es den limit lim n −1 =ℓ
n→∞ an+1
gibt, dann:
X
a) ist ℓ < 1 so ist die series an divergent;
n≥n1
X
b) ist ℓ > 1 so ist die series an convergent;
n≥n1
c) ist ℓ = 1 so kann man die Natur der series
X Joseph Ludwig Raabe
an mit dieses Kriterium nicht entscheiden. (1801 − 1859)
n≥n1
(Die series kann convergent oder divergent sein.)
 
an 1+ℓ
Proof: a) Falls ℓ < 1, es gibt n0 ∈ N so dass n −1 ≤ < 1, ∀ n ≥ n0 .
an+1 2
 
an ℓ−1
Dann: n −1 −1 ≤ < 0, also nan − (n + 1)an+1 < 0, ∀ n ≥ n0 . Es
an+1 2
folgt dass (nan )n≥n0 eine increasinge Folge ist, also nan ≥ n0 an0 , ∀ n ≥ n0 . Von
1 X
an ≥ n0 an0 · und das Vergleichungskriterium folgt, dass an divergent ist (weil
n n≥n
X1 0

divergent ist).
n
b) Falls ℓ > 1, es gibt n0 ∈ N so dass
 
an 1+ℓ
n −1 ≥ > 1, ∀ n ≥ n0 (∗).
an+1 2
Dann folgt, wie im obigen Fall, dass nan − (n + 1)an+1 > 0, ∀ n ≥ n0 , also
dass (nan )n≥n0 eine decreasinge Folge ist. Als sie auch bounded ist, so folgt, dass
(nan )n≥n0 eine convergente Folge ist.
ℓ−1
Aus (∗) folgt, dass · an+1 ≤ nan −(n+1)an+1 = bn . Die ("teleskopische") series
X 2
bn ist convergent weil die Folge seiner Partialsummen ist: sn = (1 · a1 − 2 · a2 ) +


(2 · a2 − 3 · a3 ) + ... + (n − 1)an−1 − nan = a1 − nan , also eine convergente Folge.
X
Laut das Vergleichungskriterium, als an ≤ bn , folgt dass an eine convergente
series ist.

25
 X1
c) Zum Example, f ,ur die divergente series ist
    n
an n+1
lim n − 1 = lim n − 1 = lim 1 = 1.
n→∞ an+1 n→∞ n n→∞
X 1
F ,ur die convergente series (die Konvergenz Prooft man mit dem Kon-
n ln2 n
densationskriterium)
  sind  
an n+1
lim n − 1 = lim n − 1 = lim 1 = 1.
n→∞ an+1 n→∞ n n→∞
□

Remarken:

1. Dieser Kriterium ist eine Erg ,anzung des Quotientenkriteriums:

an+1
if dies keinen Antwort liefert, also if lim = 1 (das bedeutet auchdass
n→∞ an 
an 1 an
lim = = 1), ist der limit den man berechnen muss, lim n − 1 , ein
n→∞ an+1 1 n→∞ an+1
unbestimmter Ausdruck der Art ∞ · 0.
Falls das QK wirkt, dann wirkt auchdas Raabesche
 Kriterium: 
an+1 an 1 an
• if lim = L > 1, dann ist lim − 1 = − 1 < 0, also lim n −
n→∞ an n→∞ an+1 L n→∞ an+1
  
1
1 =∞· − 1 = −∞ < 1, also ist die series divergent.
L   
an+1 an 1 an
• if lim = L < 1, dann ist lim − 1 = − 1 > 0, also lim n −
n→∞ an n→∞ an+1 L n→∞ an+1
  
1
1 =∞· − 1 = ∞ > 1, also ist die series convergent.
L
2. Dieses Kriterium gibt eine nat ,urliche Fortsetzung des Quotientenkriteri-
ums: die Schritte sind:
an+1
I. erstens berechnet man
an
an+1
II. dann berechnet man L = lim ; if L ̸= 1 gibt das QK eine Antwort.
n→∞ an
 Falls L 
III. = 1, ben ,utzt man die Berechnungen vom ersten Schritt um
an
n − 1 , und danach seinen limit zu berechnen.
an+1
X 1 · 3 · 5 · ... · (2n − 1)
Example 19. Bestimme die Natur der series .
n≥1
2 · 4 · 6 · ... · 2n
1 · 3 · 5 · ... · (2n − 1) 1 · 3 · 5 · ... · (2n − 1)·(2n + 1)
Es sei an = . Dann ist an+1 = .
2 · 4 · 6 · ... · 2n 2 · 4 · 6 · ... · 2n · (2n + 2)

26
Die Schritte:
an+1 2n + 1
I. =
an 2n + 2
an+1 2
II. lim = = 1, also bringt das QK keine Entscheidung;
n→∞ an 2
   
an 2n + 2 1 n 1
III. n −1 = n −1 = n· = −→ < 1 also ist
an+1 2n + 1 2n + 1 2n + 1 2
die series divergent.

Remark 3. Alles is umgekehrt bei das Raabesche Kriterium in Bezug auf das
QK:
an+1 an
anstatt haben wir hier und die series ist nicht convergent if ℓ < 1 wie
an an+1
es bei dem QK der Fall war, sondern im Gegenteil, if ℓ > 1.
Man kann eine andere, ,aquivalente Form, des Raabe Kriterium geben:
 
a n+1
if es den limit ℓ′ = lim n − 1 gibt, dann:
n→∞ an
• if ℓ′ < −1 dann ist die series convergent
• if ℓ′ > −1 dann ist die series divergent
• der Fall ℓ′ = −1 ist unentschieden.
 
an
Remark 4. Falls es den limit lim n −1 nicht gibt, da kann man dieses
n→∞ an+1
Kriterium nicht anwenden.
X1
Example 20. Man kann die Natur von auch mit das Kriterium von
n2
Raabe entscheiden:
1 1 an+1 n2 n2
sei an = 2 ; dann ist an+1 = und = = −→ 1.
n (n + 1)2 an (n + 1)2 n2 + 2n + 1
Das QK liefert keinen Antwort, aber man setzt mit das Raabe Kriterium fort:
n2 +2n+1 2n2 +n
   
an an 2n+1 an
= , − 1 = also lim n − 1 = lim =
an+1 n2 an+1 n2 n→∞ an+1 n→∞ n2
2 > 1 also ist die series convergent.
X 4n
Example 21. Bestimme die Natur der series n
.
C2n

27

2
4n 4n 4n · (n!)2 4n+1 · (n + 1)!
Sei an = n = = . Dann ist an+1 =
und
C2n (2n)! (2n)! 2(n + 1) !
n! · n!

2
an+1 4n+1 · (n + 1)! (2n)! 4(n + 1)2 2n + 2
= · n = = −→ 1, also gibt
an (2n + 2)! 4 · (n!)2 (2n + 1)(2n + 2) 2n + 1
das QK keine Antwort.
   
an 2n + 1 −1 −n 1
Man berechnet n −1 = n −1 = n· = −→ − < 1
an+1 2n + 2 2n + 2 2n + 2 2
also ist die series divergent.
an+1 2n + 2
Dies k ,onnte man schon fr ,uer festellen: = > 1, also ist (an )n eine
an 2n + 1
increasinge Folge. Als an > 0, es ist unm ,oglich dass lim an = 0. Es folgt dass die
n→∞
series divergent ist.

seriesN MIT NICHT NUR POSITIVE terms

Wir haben gesehen, dass, um die Natur einer positive series zu entscheiden,
uns eine Menge von Kriterien zur Verf ,ugung stehen.
X
Was k ,onnen wir machen um die Natur einer series an zu studieren if nicht alle

an positiv sind.
• Der einfachste Fall ist if es nur endlich viele an negativ oder Null sind, also if
X
an > 0, ∀ n ≥ n0 . Dann kann man f ,ur die series an alle Kriterien f ,ur positive
n≥n0
X
seriesn ben ,utzen um die Natur von an zu bestimmen.

• Sehr einfach ist auch der Fall if (fast) alle terms negativ sind, denn in diesem
X X X
Fall k ,onnen wir an = − | an | schreiben und die Konvergenz von | an |

mit die bekannten Kriterien studieren.

• Es bleiben zu studieren die seriesn die infinitely many positive und infinitely
many negative terms haben. Ein importanter Fall von solche seriesn sind die:

Alternierende seriesn

Definition 3. (Alternierende series)

X
Eine series an heisst alternierende series, if die Folgenterms an abwechselnd pos-

itives und negatives Vorzeichen haben.

28
 X (−1)n X (−1)n+1 X cos(nπ)
Examplee: ; ; √
n≥1
n n≥3
2n + ln n n≥0
1+ n
X X
Alle diese Folgen haben die Form (−1)n bn (oder (−1)n+1 bn ), wo bn > 0,
n≥n0 n≥n0
∀ n. F ,ur das dritte Example ist dies vielleicht nicht so leicht bemerkbar, aber
cos(nπ) = (−1)n .
F ,ur solche seriesn ist das Leibniz-Kriterium sehr n ,utzlich:

29
 Theorem 7. (Leibniz’s Test)
Sei (bn )n≥n0 eine monoton decreasinge Folge mit lim bn = 0.
X n→∞
Dann konvergiert die series (−1)n bn .
n≥n0

Proof: Sei zun ,achst n0 = 0. Der general Fall, mit

n0 ∈ N, kann durch das Umnumerierung der terms
der series und, if n ,otig, Multiplikation durch −1
auf dem Fall n0 = 0 zur ,uckgef ,urt werden. Wir
betrachten die Teilfolgen (s2n )n≥0 und (s2n+1 )n≥0 der
Folge der Partialsummen.
Als b2k+1 ≤ b2k ≤ b2k−1 , ∀ k ∈ N∗ , folgt, dass
s2k = s2k−2 + (−1)2k−1 b2k−1 + (−1)2k b2k =
s2k−2 − (b2k−1 − b2k ) ≤ s2k−2 und
2k
s2k+1 = s2k−1 + (−1) b2k + (−1) 2k+1
b2k+1 = Gottfried Wilhelm
Leibniz (1646 − 1716)
s2k−1 + (b2k − b2k+1 ) ≥ s2k−1 .
Also sind (s2n )n≥0 decreasing und (s2n+1 )n≥0 increasing.
Es gilt s1 ≤ s2k−1 = s2k − b2k ≤ s2k ≤ s0 , also sind die zwei Teilfolgen (s2n )n≥0 und
(s2n+1 )n≥0 bounded, also convergent. Es seien ℓ1 = lim s2n und ℓ2 = lim s2n+1 ,
n→∞ n→∞
ℓ1 , ℓ2 ∈ R. Als s2n = s2n−1 + a2n und lim a2n = 0, folgt, dass ℓ1 = ℓ2 + 0, also
n→∞
lim s2n = lim s2n+1 . Dann gibt es lim sn = ℓ1 = ℓ2 , also ist die series convergent.□
n→∞ n→∞ n→∞
X (−1)n+1
Example 22. Die alternierende harmonische series, ist, nach dem
n≥1
n
Leibniz-Kriterium, convergent. Ihr limit ist, wie sich sp ,ater heraustellen wird, ln 2.

absolutely convergente seriesn2

X
Definition 4. Eine series an heisst absolutely convergent falls die series
X n≥n0
| an | convergent ist.
n≥n0

Also eine series ist absolutely convergent falls die series der absolute values
convergent ist.
X
Example 23. Der Example 1. zeigt, dass die geometrische series bn abso-
lutely convergent ist f ,ur b ∈ (−1, 1). n≥0

Theorem 8. Jede absolutely convergente series ist convergent und es gilt

2
Dieser Begriff, sowie der zugeh ,origer Theorem, wirkt auch f ,ur seriesn mit terms an ∈ C.

30
 X X
an ≤ | an | .
n≥n0 n≥n0
X
Proof: Sei an eine absolutely convergente series. Dann gibt es, nach Theorem
n≥n0
n
X
4., zu jedem ε > 0 ein N0 ∈ N so dass | ak | < ε f ,ur alle n > m ≥ N0 .
k=m
n
X n
X n
X
Dann gilt auch ak ≤ | ak | = | ak | < ε f ,ur alle n > m ≥ N0 , also
k=m k=m k=m
X
erf ,ullt an das Cauchy-Kriterium. Nach Theorem 4. ist sie dann convergent. Da
n≥n0
n
X X n X X
ak ≤ | ak |, mit Grenz ,ubergang n −→ ∞ folgt dass an ≤ | an | .□
k=n0 k=n0 n≥n0 n≥n0

Remark: Die Umkehrung von Theorem 8. gilt nicht: die alternierende harmoni-
sche series ist convergent (s. Example 22.) aber nicht absolutely convergent (die
harmonische series ist divergent). Solche seriesn, die convergent sind, aber nicht
absolutely convergent sind heissen semiconvergente seriesn.
X X
Remark: Der obige Theorem sagt: if | an | convergent ist, dann ist an
auch convergent.
Deshalb, um die Konvergenz einer series zu prove ist es gen ,ugend seine absolutely
Konvergenz zu prove. Der Vorteil ist, dass, in Gegentum mit die gegebene series,
f ,ur die series der absolutelywerte man alle Kriterien anwenden kann, denn die
series der absolutelywerte ist eine series mit positive terms.
F ,ur alternierende series ist das Leibniz test besser geeignet als der obige Theo-
X (−1)n
rem. Zum Example Leibniz wirkt sehr einfach f ,ur , ∀ α > 0 (es sagt
n≥1
nα
diese seriesn sind alle convergent). Aber diese seriesn sind absolutely convergent
(und deshalb, nach dem obigen Theorem auch convergent) nur f ,ur α > 1. F ,ur
α ∈ (0, 1) beantwortet der obige Theorem die Frage der Konvergenz der series nicht
(nur das Leibniz test tut es).
X cos n
Example 24. Die series ist convergent.
n2
Weil der Vorzeichen von cos n schwer zu kontrolieren ist (es ,andert sich sehr of-
ten), ist die beste idee zu prove dass diese series absolutely convergent ist. Indeed,
cos n 1 X1
weil ≤ (weil −1 ≤ cos x ≤ 1, ∀ x) und convergent ist, so folgt
n2 n2 n2
X cos n X cos n
nach das Majorantenkriterium, dass auch convergent ist also
n2 n2
X cos n
absolutely convergent ist. Laut des Theoremes 8., ist dann auch conver-
n2

31
gent.

32