Numerical series¶
Part 6 · Series · Chapter 1 · Lecture notes, volume 6 (PDF) · Exercises, volume 6 (PDF) · Exercises: Numerical series · Slides (PDF) · App: Geometric series
Prerequisites
-
From this module, «Sequences and limits of sequences»: limits of sequences and geometric progressions.
-
From this module, «Summations, geometric and arithmetic progressions»: summations and geometric progressions.
Objectives
At the end of this chapter you will be able to:
-
define a series as the limit of its partial sums;
-
use the properties of series and of their tails;
-
study the harmonic, geometric, arithmetic and telescoping series.
1. Numerical series¶
- We now introduce numerical series, which extend the operation of addition to an infinite number of terms.
The sum of infinitely many terms, even if they are all positive, can give a finite result.
-
Imagine measuring a square of area 2 using the following procedure: we divide the square in half along the diagonal and measure the first right triangle: we obtain 1; then we divide the second right triangle in half and measure the first of the two resulting right triangles: we obtain \(\frac{1}{2}\); the remaining right triangle is again divided in half... and so on indefinitely. We obtain the infinite sum:
\[ 1 + \frac{1}{2} + \frac{1}{4} + \frac{1}{8} + \frac{1}{16} + \frac{1}{32} + {\rm \dots} + \frac{1}{2^k} + {\rm \dots} = \sum_{k=0}^{\infty} \frac{1}{2^k} \]which, by the way it was constructed, must give 2 as its result.
Definition 1: of numerical series
Given a sequence \(\{a_k\}_{k \in \N}\), we call numerical series of the terms \(a_k\) the expression:
- It is read “series (or also sum) for \(k\) from 0 to \(\ip\) of \(a_k\)”. The values \(a_k\) are called the general terms of the series.
Definition 2: of sequence of partial sums
The numbers
are called the \(n\)-th partial sums of the series and they define the sequence \(\{s_n\}\) of partial sums.
The behavior (character) of the series is determined by the limit of the sequence \(\{ s_n\}\) as \(n\) tends to infinity. We say that a series is convergent, divergent, irregular, if the sequence \(\{s_n\}\) of partial sums is convergent, divergent or irregular, respectively.
Definition 3: of sum of the series
If the sequence \(\{s_n\}\) of partial sums is convergent, that is, if
we say that \(s\) is the sum of the series, and we write \(\sum_{k=0}^{\infty} a_k = s\).
Hence, if the sequence \(\{s_n\}\) is convergent, we have:
The series makes precise the idea of a sum of infinitely many terms, that is, we compute the limit as \(n \rr \ip\) of the finite sum of the first \(n\) terms.
-
If, instead of summing starting from \(0\), we start from an index \(n_0 >0\), we write \(\sum_{k=n_0}^{\infty} a_k\)
-
To denote a generic numerical series without specifying the starting index \(n_0\) we will use the symbol \(\sum a_k\)
A generic numerical series \(\sum a_k\) always involves two different sequences:
-
the sequence \(\{a_k\}\) of the terms of the series
-
the sequence \(\{ s_n\}\) of partial sums
Example 1: behavior of a series
Let us determine the behavior of the series:
We have:
Hence the sequence of partial sums \(\{s_n\}\) is divergent by the comparison theorem for sequences and, consequently, the series is divergent.
1.1 Main properties of numerical series¶
Remark 1
If a sequence \(\{a_k\}\) has non-negative terms, that is, \(a_k \ge 0,\forall k\), then the sequence of partial sums \(\{s_n\}\) is increasing and regular.
Proof
Given a sequence \(\{a_k\}\) with non-negative terms, the sequence of partial sums \(\{s_n\}\) is increasing since:
by the monotonicity theorem for sequences. Consequently the sequence \(\{s_n\}\) cannot be irregular. It is therefore regular, that is, it either converges or diverges. □
Given a sequence \(\{a_k\}\) with non-negative terms:
-
If \(\{s_n\}\) is increasing and bounded, then it has a finite limit and \(\sum a_k\) converges.
-
If \(\{s_n\}\) is increasing and unbounded, then \(\sum a_k\) diverges to \(\ip\).
- This remark also holds for sequences with eventually non-negative terms, positive terms, or eventually positive terms.
Theorem 1
If a series \(\sum a_k\) is convergent, then \(\lim_{k \rr \ip} a_k=0\)
Proof
By definition of convergent series, the sequence of partial sums \(\{s_n\}\) converges to a real number \(s\), i.e., \(s_n \rr s \in \R\) as \(n \rr \ip\). Without loss of generality we consider \(n_0=0\).
We observe that the sequence \(\{s_n\}\) can be defined recursively:
Consequently we have:
□
The convergence of a series therefore implies that \(a_k \rr 0\) as \(k \rr \ip\), that is:
but the converse is not true (Example 1 is a counterexample):
Hence the fact that \(a_k \rr 0\) as \(k \rr \ip\) is a necessary but not sufficient condition for the convergence of a series.
-
From the contrapositive of \(\eqref{BBB}\) we have:
\[ \lim_{k \rr \ip} a_k \neq 0 ~~\Rightarrow~~ \sum a_k {\rm ~is~divergent~or ~irregular~~} \]that is, if the limit is not equal to zero, then the series is not convergent.
1.2 Tails of series¶
If we modify a finite number of terms of a series, the value of the sum may change, but the behavior of the series remains unchanged.
- If we modify the value of one term, for example \(a_{n_0}\), for \(n \ge n_0\) the sequences of partial sums, the original and the modified one, differ only by that term: hence both converge, both diverge, or both are irregular. The same holds if we modify a finite number of terms.
The results on the behavior of series therefore also hold if the hypotheses are satisfied “eventually”, i.e., from a certain index \(n_0\) onward.
Definition 4: of tail of a series
Given a series \(\sum_{k=0}^{\infty} a_k\) and a value \(m \in \N\), the series \(\sum_{k=m+1}^{\infty} a_k\) is called a tail of the series.
-
Given a series \(\sum_{k=0}^{\infty} a_k\) and \(m \in \N\) we have:
\[\begin{equation} \sum_{k=m+1}^{n} a_k = \left(\sum_{k=0}^{n} a_k\right) - \left(\sum_{k=0}^{m} a_k\right) = s_n -s_m, \qquad \forall n > m \label{MMMMM} \end{equation}\]hence for every tail (that is, for every \(m\)) the associated sequence of partial sums is equal to that of the original series minus a constant. Consequently we have the following remark:
Every tail of a series has the same behavior as the original series.
-
Given a convergent series, taking the limit as \(n \rr \ip\) in \(\eqref{MMMMM}\) we obtain the relation:
\[\begin{equation} \sum_{k=m+1}^{\infty} a_k = \left(\sum_{k=0}^{\infty} a_k\right) - \left(\sum_{k=0}^{m} a_k\right) = s -s_m~~~ {\rm ~~as~~} n \rr \ip \label{NNNNN} \end{equation}\]Consequently we have the following remark:
Given \(m \in \N\), every tail \(\sum_{k=m+1}^{\infty} a_k\) of a convergent series can be interpreted as the error that we make when approximating the sum \(s\) with the partial sum \(s_m\).
Theorem 2
If a series \(\sum a_k\) is convergent, then
Proof
For every \(n > m\), the \(n\)-th partial sum of the series is:
Taking the limit as \(n \rr \ip\), we have:
Now taking the limit as \(m \rr \ip\), we have:
that is, the claim of the theorem. □
This theorem tells us that, if a series is convergent, then the value \(s - s_m\), that is, the error that we make when approximating the sum \(s\) with the partial sum \(s_m\), tends to zero as \(m \rr \ip\). In other words, the tail of a convergent series tends to zero as \(m\) tends to infinity.
1.3 Harmonic series¶
Definition 5: of harmonic series
The harmonic series is the series \(\sum_{k=1}^{\infty} \frac{1}{k}\)
Theorem 3: behavior of the harmonic series
The harmonic series diverges to \(\ip\)
- Graphically we have:
The harmonic series \(\sum_{k=1}^{\infty} \frac{1}{k}\) is a counterexample that proves \(\eqref{CCC}\), since:
-
Before proving the theorem, we observe that for certain values of \(n\) the partial sums are:
\[ \underbrace{s_1}_{\displaystyle =s_{2^{\red 0}}}=1+\frac{\red 0}{2},~~~~\underbrace{s_2}_{\displaystyle =s_{2^{\red 1}}}= s_1 +\frac{1}{2}=1+\frac{0}{2}+\frac{1}{2} = 1+\frac{\red 1}{2},~~~~\underbrace{s_4}_{\displaystyle =s_{2^{\red 2}}}=s_2 + \underbrace{\left(\frac{1}{3}+\frac{1}{4}\right)}_{\ge \frac{1}{4}+\frac{1}{4}=\frac{1}{2}}\ge 1+\frac{1}{2} + \frac{1}{2}=1 + \frac{\red 2}{2} \]\[ \underbrace{s_8}_{\displaystyle =s_{2^{\red 3}}}=s_4 + \underbrace{\left(\frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}\right)}_{\ge \frac{1}{8}+\frac{1}{8}+\frac{1}{8}+\frac{1}{8}=\frac{1}{2}} \ge 1 + \frac{2}{2} + \frac{1}{2} = 1 + \frac{\red 3}{2} \]\[ \underbrace{s_{16}}_{\displaystyle =s_{2^{\red 4}}}=s_8 + \underbrace{\left(\frac{1}{9}+\frac{1}{10}+\frac{1}{11}+\frac{1}{12}+ \frac{1}{13}+\frac{1}{14}+\frac{1}{15}+\frac{1}{16} \right)}_{\ge \frac{1}{16}+\frac{1}{16}+\frac{1}{16}+\frac{1}{16}+\frac{1}{16}+\frac{1}{16}+\frac{1}{16}+\frac{1}{16}=\frac{1}{2}} \ge 1 + \frac{3}{2} + \frac{1}{2} = 1 + \frac{\red 4}{2} \]
Proof
We prove by induction on \(2^n\) that
Base case. Let \(n = 0\). Then the statement becomes \(s_{2^0} \ge 1+\frac{0}{2}\), i.e., \(1 \ge 1\), which is clearly true.
Inductive step. Assume that it is true for \(2^{n-1}\) and let us prove it for \(2^{n}\). By the inductive hypothesis, we have:
Moreover, we have:
which is exactly the desired statement for \(2^{n}\).
We therefore have:
and \(\{s_n\}\) is unbounded above and also monotonically increasing, since \(a_k=1/k>0, \forall k >1\). Consequently \(\{s_n\}\) diverges to \(\ip\) by the monotonicity theorem for sequences, and the harmonic series diverges to \(\ip\). □
The sequence \(a_k=1/k, \forall k \ge 1\), of the harmonic series has positive terms, hence the sequence of partial sums \(\{s_n\}\) is increasing, and it is divergent since it is unbounded.
1.4 Geometric series¶
Definition 6: of geometric series
Given \(q \in \R\), the geometric series with ratio \(q\) is the series \(\sum_{k=0}^{\infty} q^k\)
Theorem 4: behavior and sum of the geometric series
Given the ratio \(q \in \R\),
If the geometric series is convergent, its sum \(s\) is \(\frac{1}{1-q}\)
Proof
Given \(q \in \R\) and \(n \in \N\), the \(n\)-th partial sum of the geometric series is:
since it equals the sum of the first \(n+1\) terms of the geometric progression. Moreover, we have:
Hence, if \(q \neq 1\):
and if \(q=1\) we have:
□
-
Given the ratio \(q \in \R\) and a first term \(a \in \R\), the result extends to the series \(\sum_{k=0}^{\infty} a \: q^k\) of the terms of the geometric progression:
\[\begin{equation} \label{GEOMA} {\rm the~series~} \sum_{k=0}^{\infty} a \: q^k {\rm ~~~~is~~~~} \begin{cases} {\rm convergent,~with~sum~} 0 & {\rm if~} a = 0\\[1ex] {\rm convergent,~with~sum~} \frac{a}{1-q} & {\rm if~} a \neq 0 {\rm ~and~} |q| < 1\\[1ex] {\rm divergent~to~} \ip & {\rm if~} a > 0 {\rm ~and~} q \ge 1\\[1ex] {\rm divergent~to~} \im & {\rm if~} a < 0 {\rm ~and~} q \ge 1\\[1ex] {\rm irregular~} & {\rm if~} a \neq 0 {\rm ~and~} q \le -1 \end{cases} \end{equation}\]Indeed, if \(a=0\) all the terms of the series are zero and therefore \(s_n = 0\) for every \(n\). If instead \(a \neq 0\), by the product by a constant of summations the \(n\)-th partial sum is:
\[ s_n = \sum_{k=0}^{n} a \: q^k = a \: \sum_{k=0}^{n} q^k \]and therefore, by the theorem on the limit of the product of a constant and a sequence, \(\{s_n\}\) has the same behavior as the sequence of the partial sums of \(\sum_{k=0}^{\infty} q^k\), with the limit multiplied by \(a\) (in the divergent cases the sign of \(a\) decides between \(\ip\) and \(\im\)).
Example 2: geometric series
Let us determine the behavior of the series:
It is a geometric series with ratio \(q=1/2\), hence it is convergent and its sum \(s\) is \(1/(1-1/2) =2\).
Let us determine the behavior of the series:
It is a geometric series with ratio \(q=13/12\), hence it is divergent.
Try it — the interactive graph below shows what you have just read: move the sliders.
1.5 Arithmetic series¶
Definition 7: of arithmetic series
Given the first term \(a \in \R\) and the common difference \(d \in \R\), the arithmetic series is the series \(\sum_{k=0}^{\infty} (a + k \: d)\)
- The terms of the arithmetic series are the terms of the arithmetic progression with first term \(a\) and common difference \(d\), that is, \(a, ~a+d, ~a+2\:d, ~\dots\) As for the geometric series, the first term is in position \(k=0\).
Theorem 5: behavior and sum of the arithmetic series
Given the first term \(a \in \R\) and the common difference \(d \in \R\),
If the arithmetic series is convergent, its sum \(s\) is \(0\)
Proof
Given \(a, d \in \R\) and \(n \in \N\), the \(n\)-th partial sum of the arithmetic series is:
since it equals the sum of the first \(n+1\) terms of the arithmetic progression. Therefore, if \(d \neq 0\), collecting the term of highest degree we have:
and if \(d=0\) we have \(s_n = a \: (n+1)\) and therefore:
□
The arithmetic series is therefore convergent only in the trivial case \(a = d = 0\), in which all its terms are zero, and it is never irregular (unlike the geometric series, which is irregular for \(q \le -1\)).
-
The fact that the arithmetic series cannot converge, except in the trivial case, also follows from the necessary condition for convergence \(\eqref{BBB}\). The general term is \(a_k = a + k \: d\) and we have:
\[ \lim_{k \rr \ip} (a + k \: d) = \begin{cases} +\infty & {\rm if~} d > 0\\[1ex] -\infty & {\rm if~} d < 0\\[1ex] a & {\rm if~} d = 0 \end{cases} \]that is, \(\lim_{k \rr \ip} a_k = 0\) only if \(d = 0\) and \(a = 0\). In all the other cases the contrapositive of \(\eqref{BBB}\) guarantees that the series is not convergent. The necessary condition does not say, however, which the behavior is: it is the computation of the partial sums carried out in the proof that establishes that the series is divergent, and with which sign.
Example 3: arithmetic series
Let us determine the behavior of the series:
It is an arithmetic series with first term \(a=1\) and common difference \(d=1/2\). Since \(d>0\), the series is divergent to \(\ip\).
1.6 Telescoping series¶
Definition 8: of telescoping series
A telescoping series is a series of the form:
Theorem 6: behavior and sum of the telescoping series
A telescoping series converges, diverges or is irregular according to whether the sequence \(\{b_k\}\) converges, diverges or is irregular, respectively.
Moreover, if \(\ell \in \R\), then the telescoping series is convergent and its sum \(s\) is \(b_{n_0} - \ell\)
Proof
We have
hence:
□
Remark 2
The series \(\sum_{k=1}^{\infty} \frac{1}{k \; (k+1)}\), called Mengoli's series, is convergent and its sum \(s\) is equal to \(1\).
Proof
We have:
Mengoli's series therefore has the form:
hence it is a convergent telescoping series (\(n_0=1\)) and moreover \(s= b_1 = 1\). □
- Graphically we have:
Proof
Alternative proof
For every \(n\ge 1\), the value of the partial sum of Mengoli's series is:
Hence we have
Consequently Mengoli's series is convergent and its sum \(s\) is 1. □
Example 4: telescoping series
Let us determine the behavior of the series:
We have:
consequently:
The series therefore has the form:
that is, a telescoping series (\(n_0=1\)), and we have:
Hence it is a convergent telescoping series and its sum \(s\) is \(\frac{1}{15}\).
2. In summary¶
In summary
-
The sum of a series is the limit of its partial sums: it converges, diverges or is irregular.
-
If the series converges, its terms tend to \(0\); the converse is false (harmonic series).
-
\(\sum q^n\) converges to \(\frac{1}{1-q}\) for \(|q| < 1\); telescoping series are summed directly.
Exercises on this chapter: Numerical series