Successions

Successions

Think that are the inheritance

Informally, a succession of real numbers is a list limited by numbers s1,s2,s3,s4,,sn,s_1, s_2, s_3, s_4, \cdots, s_n, \cdots (where n indicates the place in the number sns_n list); it's obvious that this is a real function with domain N\mathbb{N}

Succession

A succession of elements of a set is an application domain N\mathbb{N} and codomain said set. In particular, a sequence of real numbers is a real function with domain R\mathbb{R}, that is, an application sNRs|\mathbb{N}\rightarrow\mathbb{R}

Traditionally, the value that a succession takes in each nNn\in\mathbb{N} it is denoted by sns_n, instead of s(n)s(n) like any other function. We will usually refer to sns_n with the name of the nth term of a succession, but it should not be overlooked that each term carries a double information: its value and the n place it occupies

As the domain N\mathbb{N} is common to all successions, rather than using notation sNRs|\mathbb{N}\rightarrow\mathbb{R}, is more common to find notations of the type (sn)nN(s_n)_{n\in\mathbb{N}} or (sn)n=1{sn}n=1(s_n)_{n=1}^{\infty}\text{, }\left\{s_n\right\}_{n=1}^{\infty} or (sn)(s_n), if it does not lead to confusion, or similar, with greater emphasis on the terms

Although notation may lead to confusion, it should not be necessary to insist on the difference between succession and the set of values that the succession takes, which is the same as between any function and its set of values (image set or range); note, for example, that a succession always has infinite terms even if it takes a single value, as is the case with constant successions

Successions are indicated by giving a formula that defines the term nth, the most common being:

  • Constant succession:
    sn=as_n=a, where a is a prefixed real number and consists of the terms a,a,a,a,,a,a, a, a, a, \cdots, a, \cdots
  • Succession of natural numbers:
    sn=ns_n=n, consists of the terms 1,2,3,4,,n,1, 2, 3, 4, \cdots, n, \cdots
  • Succession of fractional numbers:
    sn=1ns_n=\frac{1}{n}, consists of the terms 1,12,13,14,,1n,1, \frac{1}{2}, \frac{1}{3}, \frac{1}{4}, \cdots, \frac{1}{n}, \cdots
  • Succession of 1, -1:
    sn=(1)ns_n=(-1)^n, consists of the terms 1,1,1,1,,(1)n,-1, 1, -1, 1, \cdots, (-1)^n, \cdots
  • Complex algebraic successions:
    Formulas do not have to refer only to simple algebraic operations. For example, the succession of decimal approximations of π\pi, consists of the terms 3.1,3.14,3.141,3.1415,,3.14159265,3.1, 3.14, 3.141, 3.1415, \cdots, 3.14159265, \cdots

    We can give an explicit formula for the nth term with the help of the entire part function, even if we did not know how to write all the figures of the term, for example, a million

    Specifically, for each nNn\in\mathbb{N}, sn=3+a110+a2102++ak10k++an10ns_n= 3 + \frac{a_1}{10} + \frac{a_2}{10^2} + \cdots + \frac{a_k}{10^k} + \cdots + \frac{a_n}{10^n} where ak=[10kπ]10[10k1π](1kn)a_k= [10^k \cdot \pi] - 10 \cdot [10^{k-1} \cdot \pi] (1 \leq k\leq n)

    The fact that this formula does not provide a computational algorithm for the aka_k it does not prevent that these are defined without ambiguity and without any exception

  • Recurring successions:
    This name is given successions whose terms are defined according to the above (by means of an inductive or recursive definition). An example of this type is the succession of Fibonacci:

    s1=s2=1,sn+2=sn+1+sn,nNs_1=s_2=1, s_{n+2}=s_{n+1}+s_n, n\in\mathbb{N}, consists of the terms 1,1,2,3,5,8,13,21,34,55,89,144,1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, \cdots

  • Succession of non-mathematical
    The rule that defines a succession does not have to be of a strictly mathematical character. For example, we can define a succession as follows:

    sn={1073if the name of the number n contains the letter dnotherwise\tiny s_n=\left\{\begin{matrix}\frac{10^7}{3} & \text{if the name of the number n contains the letter d} \\ \sqrt{n} & \text{otherwise}\end{matrix}\right.

    Or any other condition allowing to ensure that for each nNn\in\mathbb{N} is associated with it without exception, unequivocally a single perfectly defined real number

  • Succession, which is exactly a set numerical
    There are successions whose rank is exactly Z\mathbb{Z} or Q\mathbb{Q}; the usual construction is done through The Diagonal Process of Cantor

    s1=0,s2=1,s3=12,s4=12,s_1=0, s_2=1, s_3=\frac{1}{2}, s_4=\frac{-1}{2}, \cdots

    This construction, which supports repeated, allows to prove that the whole of the rationals is numerable; that is, its cardinal matches that of the natural

  • Limit of the succession

    A succession (sn)(s_n) it is said convergent if there is a real number such that for each ϵ>0\epsilon > 0 you can find a natural number N=N(ϵ)N=N(\epsilon) so as long as n>Nn>N check sna<ϵ\left \|s_n-a\right \|<\epsilon

    It is then said that the number is limit of the succession (sn)(s_n) and writes a=limnsna=\underset {n \to \infty} {\lim} s_n. We'll also say that sns_n converges the number to

    The expression snas_n \to a is used to indicate that the sequence of nth term sns_n is convergent and has the limit

    Remember that the inequality sna<ϵ\left \|s_n-a\right \|<\epsilon it is equivalent to the two inequalities ϵ<sna<ϵ-\epsilon < s_n - a < \epsilon that are equivalent to the inequalities aϵ<sn<a+ϵa-\epsilon < s_n < a+\epsilon

    • The constant succession sn=c(cR)s_n=c (c\in\mathbb{R}) converges to the number ϵ\epsilon
    • The succession sn=1ns_n=\frac{1}{n} converges to 0 (as a result of The Archemedian property)
    • The succession sn=(1)ns_n=(-1)^n is not convergent if I had limit 1, taking ϵ=2\epsilon=2 in defining limit, it would have to be sn1<2\left \|s_n-1\right \|<2 for all n big enough; However sn1=2\left \|s_n-1\right \|=2 for all odd n'n. And if I had a limit a1a\not=1, taking ϵ=1a\epsilon=\left \|1-a\right \|, it would have to be sna<1a\left \|s_n-a \right \| < \left \|1-a \right \| for all n big enough; However sna=1a\left \|s_n-a \right \| = \left \|1-a \right \| for all n pairs

      Conclusion: the succession has no limit

    • The succession sn=ns_n=n can't be convergent, because if I had a limit to, taking ϵ=1\epsilon=1 in the definition of convergence, for some N would have to be n<a+1n<a+1 as long as n was greater than N, which is impossible (as a result of Archimadian property)

    Proposition: uniqueness of the boundary of a succession

    It (sn)(s_n) a succession convergent and a,bRa, b \in \mathbb{R} such that a=limnsn,b=limnsna=\underset {n \to \infty} {\lim} s_n, b=\underset {n \to \infty} {\lim} s_n then a=ba=b

    Demonstration

    Since a and b are the limits of the succession sns_nDice ϵ>0\epsilon > 0 there will be N and N' such that

    {snaϵ2if n>Nsnbϵ2if n>N\left\{\begin{matrix}\left \| s_n-a \right \|\leq \frac{\epsilon}{2} & \text{if }n > N \\\left \| s_n-b \right \|\leq \frac{\epsilon}{2} & \text{if }n > N'\end{matrix}\right.


    Then ab=asn+snbsna+snb<ϵ2+ϵ2=ϵ\left \|a-b\right\|=\left\|a-s_n+s_n-b\right\|\leq \left\|s_n-a\right\|+\left\|s_n-b\right\|<\frac{\epsilon}{2}+\frac{\epsilon}{2}=\epsilon, therefore ab=0a=b\left\|a-b\right\|=0\rightarrow a=b

    Conclusion: the limit of a sequence converging is the unique real number to which the succession converges

    Succession bounded

    A succession (sn)n=1(s_n)_{n=1}^{\infty} it is said that is bounded superiorly if there is a number CRC\in\mathbb{R} such that for all nN,snCn\in\mathbb{N}, s_n \leq C

    It is said that is bounded inferiorly if there is a number KRK\in\mathbb{R} such that for all nN,Ksnn\in\mathbb{N}, K\leq s_n

    It is said that is limited if it's top and bottom. It is equivalent to saying that there is a number M0M\geq 0 such that for all nN,snMn\in\mathbb{N}, \left\|s_n\right\|\leq M

    Proposition: Convergent and bounded

    The whole succession is convergent it is bounded

    Demonstration

    It (sn)(s_n) a sequence converging to a number aRa\in\mathbb{R}

    We take ϵ=1\epsilon=1 in the definition of limit, and there will be some NNN\in\mathbb{N} such that sna<1\left\|s_n-a\right\| < 1 for all n>Nn >N

    If B=max{1,s1a,s2a,,sNa}B=\max\lbrace 1,\left\|s_1-a\right\|, \left\|s_2-a\right\|, \cdots, \left\|s_N-a\right\|\rbrace you have to snaB\left\|s_n-a\right\|\leq B, that is to say, aBsna+Ba-B\leq s_n \leq a+B for all nNn\in \mathbb{N}

    Conclusion: the succession is bounded

    Succession monotonous

    A succession (sn)(s_n) it is monotonically non decreasing if nN\forall n \in \mathbb{N} verified snsn+1s_n\leq s_{n+1}

    A succession (sn)(s_n) it is monotonically non-increasing if nN\forall n \in \mathbb{N} verified snsn+1s_n \geq s_{n+1}

    A succession (sn)(s_n) it is strictly increasing if nN\forall n \in \mathbb{N} verified sn<sn+1s_n < s_{n+1}

    A succession (sn)(s_n) it is strictly decreasing if nN\forall n \in \mathbb{N} verified sn>sn+1s_n > s_{n+1}

    A succession is monotonous if you meet any of the above cases

    Proposition: equivalence with convergence at 0

    If (sn)(s_n) it is a succession bounded and (tn)(t_n) is a convergent succession to 0, succession (sntn)(s_n \cdot t_n) converges to 0

    Demonstration

    It ϵ>0,K>0\epsilon > 0, K > 0 such that snK,nN\left\|s_n\right\| \leq K, \forall n \in\mathbb{N}

    Using the definition of convergence of tnt_n for ϵK\frac{\epsilon}{K} you have to sntnKtnKϵK=ϵ\left\|s_n \cdot t_n\right\| \leq K \cdot \left\|t_n\right\|\leq K\cdot \frac{\epsilon}{K}=\epsilon

    Conclusion: (sntn)(s_n \cdot t_n) converges to 0

    Proposition: convergent if lying above or below

    • It (sn)(s_n) a non-decreasing monotonous succession. Then (sn)(s_n) is convergent if and only if it is lying down higher, in which case limnsn=sup{snnN}\underset {n \to \infty} {\lim} s_n=\sup\lbrace s_n|n\in\mathbb{N}\rbrace
    • It (sn)(s_n) a monotonous non-increasing succession. So (sn)(s_n) is convergent if and only if it is bounded below, in which case limnsn=inf{snnN}\underset {n \to \infty} {\lim} s_n=\inf\lbrace s_n|n\in\mathbb{N}\rbrace

    Demonstration: A

    It (sn)(s_n) a non-decreasing monotonous succession. If the succession converges then it is bound (above); this demonstrates an implication of section A

    Now suppose that the succession is superiorly bound, be it to its supreme and we will see that the succession converges to the point a

    It ϵ>0\epsilon > 0 and as aϵ<aa-\epsilon < a, the number aϵa-\epsilon it may not be an upper bound of the sequence and therefore there will be some NNN\in\mathbb{N} such that aϵ<sNa-\epsilon < s_N

    As succession is non-diminishing, for every n>Nn > N you have to aϵ<sNsN+1sna-\epsilon < s_N \leq s_{N+1}\leq \cdots\leq s_n and therefore, aϵ<sn<a+ϵa-\epsilon < s_n < a + \epsilon

    Conclusion: the succession converges to the point

    Demonstration: B

    It (sn)(s_n) an unsrising monotonous succession. If the succession converges then it is bound (lower); this demonstrates an implication of paragraph B

    Now suppose the succession is lower, be b its limit and we will see that the succession converges to point b

    It ϵ>0\epsilon > 0 and as bϵ>bb-\epsilon > b, the number bϵb-\epsilon may not be a lower bound of the sequence and therefore there will be some NNN\in\mathbb{N} such that bϵ>sNb-\epsilon > s_N

    As succession is non-increasing, for every n<Nn < N you have to bϵ>sNsN+1snb-\epsilon > s_N \geq s_{N+1}\geq \cdots\geq s_n and therefore, bϵ>sn>b+ϵb-\epsilon > s_n > b + \epsilon

    Conclusion: the succession converges to point b

    Proposition: Converging succession equivalences

    Are (sn)(s_n) and (tn)(t_n) succession are convergent with limits a=limnsna = \underset {n \to \infty} {\lim} s_n, b=limntnb = \underset {n \to \infty} {\lim} t_n and cCc \in \mathbb{C}

    • (sn+tn)(s_n + t_n) is convergent and has limit a+ba + b
    • (csn)(c \cdot s_n) is convergent and has limit cac \cdot a
    • (sntn)(s_n \cdot t_n) is convergent and has limit aba \cdot b
    • If the succession (tn)(t_n) contains no terms null and b0b \not= 0 then (sntn)(\frac{s_n}{t_n}) is convergent and has limit (ab)(\frac{a}{b})

    Demonstration: A

    It ϵ>0\epsilon > 0
    a=limnsnω1a = \underset {n \to \infty} {\lim} s_n \Rightarrow \exists \omega_1 such that sna<ϵ2;n>N1\left\|s_n - a\right\| < \frac{\epsilon}{2}; \forall n > N_1 with n,N1Nn, N_1 \in \mathbb{N}
    b=limntnω2b = \underset {n \to \infty} {\lim} t_n \Rightarrow \exists \omega_2 such that tnb<ϵ2;n>N2\left\|t_n - b\right\| < \frac{\epsilon}{2}; \forall n > N_2 with n,N2Nn, N_2 \in \mathbb{N}
    Then N>max{N1,N2}N > \max\lbrace N_1, N_2 \rbrace, therefore if n>Nn > N is true that (sn+tn)(a+b)(sna)+(tnb)\left\|(s_n + t_n) - (a + b)\right\| \leq \left\|(s_n - a)\right\| + \left\|(t_n - b)\right\| < ϵ2+ϵ2=ϵ\frac{\epsilon}{2} + \frac{\epsilon}{2} = \epsilon

    Conclusion: the succession (sn+tn)(s_n + t_n) is convergent and has limit a+ba + b

    Demonstration: B

    If c=0c = 0 the result is trivial. So we'll assume that c0c \not= 0
    n,N1N\exists n, N_1 \in \mathbb{N} such that n>N1n > N_1 then sna<ϵc\left\|s_n - a\right\| < \frac{\epsilon}{\left\|c\right\|}
    Then csnca=csna<ccc=ϵ\left\|c \cdot s_n - c \cdot a\right\| = \left\|c\right\| \cdot \left\|s_n - a\right\| < \frac{\left\|c\right\| \cdot c}{\left\|c\right\|} = \epsilon

    Conclusion: the succession (csn)(c \cdot s_n) is convergent and has limit cac \cdot a

    Demonstration: C

    It ϵ>0\epsilon > 0
    a=limnsnω1a = \underset {n \to \infty} {\lim} s_n \Rightarrow \exists \omega_1 such that sna<ϵ2b+1;n>N1\left\|s_n - a\right\| < \frac{\epsilon}{2 \cdot \left\|b\right\| + 1}; \forall n > N_1 with n,N1Nn, N_1 \in \mathbb{N}
    b=limntnω2b = \underset {n \to \infty} {\lim} t_n \Rightarrow \exists \omega_2 such that tnb<ϵ2K;n>N2\left\|t_n - b\right\| < \frac{\epsilon}{2 \cdot K}; \forall n > N_2 with n,N2Nn, N_2 \in \mathbb{N}
    Then sntnab=sntnsnbabsntnb+bsnaKϵ2K+bϵ2b+1<ϵ\left\|s_n \cdot t_n - a \cdot b\right\| = \left\|s_n \cdot t_n - s_n \cdot b - a \cdot b \right\| \leq \left\|s_n\right\| \cdot \left\|t_n - b\right\| + \left\|b\right\| \cdot \left\|s_n - a\right\| \leq K \cdot \frac{\epsilon}{2 \cdot K} + \left\|b\right\| \cdot \frac{\epsilon}{2 \cdot \left\|b\right\| + 1} < \epsilon

    Conclusion: the succession (sntn)(s_n \cdot t_n) is convergent and has limit aba \cdot b

    Demonstration: D

    It ϵ>0\epsilon > 0
    sntnab=snbtnabtn=snbtnabtn=snb+sntnsntntnabtnsntnb+tnsnabtn\left\|\frac{s_n}{t_n}-\frac{a}{b}\right\| = \left\|\frac{s_n \cdot b - t_n \cdot a}{b \cdot t_n}\right\| = \left\|\frac{s_n \cdot b - t_n \cdot a}{\left\|b\right\| \cdot \left\|t_n\right\|}\right\| = \left\|\frac{s_n \cdot b + s_n \cdot t_n - s_n \cdot t_n - t_n \cdot a}{\left\|b\right\| \cdot \left\|t_n\right\|}\right\| \leq \left\|\frac{\left\|s_n\right\| \cdot \left\|t_n - b\right\| + \left\|t_n\right\| \cdot \left\|s_n - a\right\|}{\left\|b\right\| \cdot \left\|t_n\right\|}\right\|

    Then n,N1N\exists n, N_1\in \mathbb{N} such that if n>N1n > N_1 you have to tn>b2\left\|t_n\right\| > \frac{\left\|b\right\|}{2}

    By be sns_n and tnt_n inheritance of convergent K1,K2>0;nN\exists K_1, K_2 > 0; \forall n \in \mathbb{N} such that sn<K1\left\|s_n\right\| < K_1 and tn<K2\left\|t_n\right\| < K_2

    Then for the limit sns_n n,N2N\exists n, N_2\in \mathbb{N} such that if n>N2n > N_2 you have to sna<ϵb24K2\left\|s_n - a\right\| < \frac{\epsilon \cdot \left\|b\right\|^{2} }{4 \cdot K_2}

    Then for the limit tnt_n n,N3N\exists n, N_3\in \mathbb{N} such that if n>N3n > N_3 you have to tnb<ϵb24K1\left\|t_n - b\right\| < \frac{\epsilon \cdot \left\|b\right\|^{2} }{4 \cdot K_1}

    Then n>max{N1,N2,N3}n > \max\lbrace N_1, N_2, N_3 \rbrace, therefore if n>Nn > N is true that sntnabsntnb+tnsnabtn<k1ϵb24K1+k2ϵb24K2bb2=ϵ\left\|\frac{s_n}{t_n}-\frac{a}{b}\right\| \leq \left\|\frac{\left\|s_n\right\| \cdot \left\|t_n - b\right\| + \left\|t_n\right\| \cdot \left\|s_n - a\right\|}{\left\|b\right\| \cdot \left\|t_n\right\|}\right\| < \frac{k_1 \cdot \frac{\epsilon \cdot \left\|b\right\|^{2} }{4 \cdot K_1} + k_2 \cdot \frac{\epsilon \cdot \left\|b\right\|^{2} }{4 \cdot K_2} }{\frac{\left\|b\right\| \cdot \left\|b\right\|}{2} } = \epsilon

    Conclusion: the succession (sntn)(\frac{s_n}{t_n}) is convergent and has limit (ab)(\frac{a}{b})

    Proposition: existence of limited intervals in convergent successions

    Are (sn)(s_n) and (tn)(t_n) convergent successions and msntnn>m\exists m | s_n \leqslant t_n \forall n > m

    Then limnsnlimntn\underset {n \to \infty} {\lim} s_n \leqslant \underset {n \to \infty} {\lim} t_n

    Demonstration

    The succession tnsnt_n - s_n meets inequality 0tnsn,n>m0 \leqslant t_n - s_n, \forall n > m and converges to limntnlimnsn\underset {n \to \infty} {\lim} t_n - \underset {n \to \infty} {\lim} s_n

    We replace inequality 0limntnlimnsn0 \leqslant \underset {n \to \infty} {\lim} t_n - \underset {n \to \infty} {\lim} s_n, therefore it is fulfilled that limnsnlimntn\underset {n \to \infty} {\lim} s_n \leqslant \underset {n \to \infty} {\lim} t_n

    Conclusion: the bounded interval exists sntns_n \leqslant t_n for convergent successions (sn)(s_n) and (tn)(t_n)

    Cantor's Theorem of embedded intervals

    As a result of the proposition of limited intervals in convergent successions we can list the Theorem of Cantor of the fitted intervals so that the intersection is formed by a single point

    nN\forall n \in \mathbb{N}, it In=[an,bn]I_n =[ a_n, b_n] \leq \emptyset a closed interval

    Suppose that In+1InI_{n+1}\subseteq I_n that is to say anan+1bn+1bna_n\leqslant a_{n+1}\leqslant b_{n+1}\leqslant b_n and that in addition limn(bnan)=0\underset {n \to \infty} {\lim} \left ( b_n - a_n \right ) = 0

    Then NIn={x}\underset {\in \mathbb{N}} {\bigcap} I_n = \left \{ x \right \} where x=limnan=limnbnx = \underset {n \to \infty} {\lim} a_n = \underset {n \to \infty} {\lim} b_n

    Demonstration

    By hypothesis, succession (an)(a_n) is monotonous, non-decreasing and quoted superiorly (for example b1b_1), therefore converging xRx\in\mathbb{R}

    Similarly, succession (bn)(b_n) converges to rRr\in\mathbb{R} and by the previous proposition anxrbn,nNa_n\leqslant x \leqslant r \leqslant b_n, \forall n \in\mathbb{N}

    We replace the condition limn(bnan)=0\underset {n \to \infty} {\lim} (b_n - a_n) = 0 that assures us that x=rx = r and that {x}=nNIn\left \{ x \right \} = \underset {n \in \mathbb{N}} {\bigcap} I_n

    With what is proven the Cantor's Theorem of embedded intervals

    Proposition: Sandwich Rule

    Are (sn)(s_n), (tn)(t_n) and (un)(u_n) and mNsntnun,n>m\exists m \in\mathbb{N} | s_n \leqslant t_n \leqslant u_n, \forall n > m

    If (sn)(s_n) and (un)(u_n) are convergent successions and limnsn=limnun=a\underset {n \to \infty} {\lim} s_n = \underset {n \to \infty} {\lim} u_n = a

    Then (tn)(t_n) is also convergent and limntn=a\underset {n \to \infty} {\lim} t_n = a

    Demonstration

    It ϵ>0\epsilon > 0, by the definition of limit N1Nif n>N1sna<ϵ\exists N_1 \in \mathbb{N}\Rightarrow \text{if }n > N_1 \rightarrow \left \| s_n - a \right \| < \epsilon

    That is to say aϵ<sn<a+ϵa - \epsilon < s_n < a + \epsilon

    Similarly N2Nif n>N2una<ϵ\exists N_2 \in \mathbb{N}\Rightarrow \text{if }n > N_2 \rightarrow \left \| u_n - a \right \| < \epsilon

    So if n>maˊx{m,N1,N2}n > \text{m\'{a}x}\left \{ m, N_1, N_2 \right \} you have to aϵ<sntnun<a+ϵa - \epsilon < s_n \leqslant t_n \leqslant u_n < a + \epsilon or what's equivalent tna<ϵ\left \| t_n -a \right \| < \epsilon

    With what we've tried the Sandwich Rule