Study Guides / Calculus I / Exercise session 3
EXERCISE SESSION 3 · ETH ZURICH · FS20 · calculus

Fundamental Limits, Subsequences and Limit Superior

Fundamental limits, subsequences, accumulation points, limit superior, limit inferior, Cauchy criterion

SB
Steven Battilana Quant · ex-ETH

Useful: Fundamental Limits

  1. lim⁡n→±∞(1±xn)n=e±x\lim\limits_{n \to \pm \infty} (1 \pm \frac{x}{n})^n = e^{\pm x}
  2. lim⁡n→∞  ∧  f(n)→∞(1+1f(n))f(n)=e\lim\limits_{n \to \infty \; \wedge \; f(n) \to \infty} (1 + \frac{1}{f(n)})^{f(n)} = e
  3. lim⁡n→0(1+n)1n=e\lim\limits_{n \to 0} (1+n)^{\frac{1}{n}} = e
  4. lim⁡n→0(1+f(n))1f(n)=e\lim\limits_{n \to 0} (1+f(n))^{\frac{1}{f(n)}} = e
Example 1.1

Solve lim⁡x→∞(1−πx)2x\lim\limits_{x \to \infty} (1-\frac{\pi}{x})^{2x}.

Show solutionHide solution lim⁡x→∞(1−πx)2x=lim⁡x→∞[(1+1−xπ)−xπ]−πx⋅2x=e−πx⋅2x=e−2π\begin{aligned} \lim_{x \to \infty} (1-\frac{\pi}{x})^{2x} &= \lim_{x \to \infty} \left[ \left(1+\frac{1}{-\frac{x}{\pi}}\right)^{-\frac{x}{\pi}} \right]^{-\frac{\pi}{x} \cdot 2x}\\ &= e^{-\frac{\pi}{x} \cdot 2x}\\ &= e^{-2\pi} \end{aligned}
Example 1.2

Solve lim⁡x→01xlog⁡(1+x1−x)\lim\limits_{x \to 0} \frac{1}{x} \log(\frac{1+x}{1-x})

Show solutionHide solution lim⁡x→01xlog⁡(1+x1−x)=lim⁡x→0log⁡((1+x1−x)1x)=lim⁡x→0log⁡((1+x+x−x1−x)1x)=lim⁡x→0log⁡((1−x+x+x1−x)1x)=lim⁡x→0log⁡((1−x1−x+x+x1−x)1x)=lim⁡x→0log⁡((1+2x1−x)1x)=lim⁡x→0log⁡([(1+2x1−x)1−x2x]2x1−x⋅1x)=lim⁡x→0log⁡(e2x1−x⋅1x)=lim⁡x→0log⁡(e21−x)=log⁡(e21−0)=log⁡(e2)=2\begin{aligned} \lim_{x \to 0} \frac{1}{x} \log \left( \frac{1+x}{1-x} \right) &= \lim_{x \to 0} \log \left(\left( \frac{1+x}{1-x}\right)^{\frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left(\left( \frac{1+x + x-x}{1-x}\right)^{\frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left(\left( \frac{1-x + x+x}{1-x}\right)^{\frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left(\left( \frac{1-x}{1-x} + \frac{x+x}{1-x}\right)^{\frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left(\left( 1 + \frac{2x}{1-x}\right)^{\frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left(\left[\left( 1 + \frac{2x}{1-x}\right)^{\frac{1-x}{2x}} \right]^{\frac{2x}{1-x} \cdot \frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left( e^{\frac{2x}{1-x} \cdot \frac{1}{x}} \right)\\ &= \lim_{x \to 0} \log \left( e^{\frac{2}{1-x}} \right)\\ &= \log \left( e^{\frac{2}{1-0}} \right)\\ &= \log \left( e^2 \right)\\ &= 2 \end{aligned}

Cauchy Criterion

Definition 2.1 Cauchy sequence

A sequence (an)n∈N(a_n)_{n \in \mathbb{N}} is called a Cauchy sequence if:

∀ε>0,  ∃N=N(ε)∈N:∣an−am∣<ε,∀n,m>N.\forall \varepsilon > 0, \; \exists N=N(\varepsilon) \in \mathbb{N}: \quad |a_n - a_m| < \varepsilon, \quad \forall n,m > N.
Satz Cauchy criterion (3.5.1)

Let (an)⊂R(a_n) \subset \mathbb{R} be a sequence. The following statements are equivalent:

  1. (an)(a_n) is convergent
  2. (an)(a_n) is a Cauchy sequence.
Example 2.1

Is the sequence n−12n\frac{n-1}{2n} convergent?

Show solutionHide solution

We choose an (arbitrarily small) ε>0\varepsilon > 0. We look for NN such that ∣an−am∣<ε|a_n - a_m| < \varepsilon for all m,n>Nm,n > N. W.l.o.g. (without loss of generality) we can fix m≥nm \geq n (or n≥mn \geq m).

∣an−am∣=∣n−12n−m−12m∣=∣m(n−1)−n(m−1)2nm∣=∣nm−m−nm+n2nm∣=∣n−m2nm∣=∣12m−12n∣≤m≥n∣12n∣=∣12n∣<!ε.⇔n>12ε⇒N=⌈12ε⌉\begin{aligned} |a_n - a_m| &= \left| \frac{n-1}{2n} - \frac{m-1}{2m} \right|\\ &= \left| \frac{m(n-1) - n(m-1)}{2nm} \right|\\ &= \left| \frac{nm-m - nm+n}{2nm} \right|\\ &= \left| \frac{n-m}{2nm} \right|\\ &= \left| \frac{1}{2m} - \frac{1}{2n} \right|\\ &\stackrel{m \geq n}{\leq} \left| \frac{1}{2n} \right|\\ &= \left| \frac{1}{2n} \right|\\ &\stackrel{!}{<} \varepsilon.\\ &\Leftrightarrow \qquad n > \frac{1}{2 \varepsilon}\\ &\Rightarrow \qquad N = \left\lceil \frac{1}{2 \varepsilon} \right\rceil \end{aligned}

The sequence ana_n is therefore a Cauchy sequence, and since all Cauchy sequences in R\mathbb{R} converge, our sequence is convergent as well.

Subsequences, Accumulation Points

Definition 3.1

Let (an)n∈N⊂R(a_n)_{n \in \mathbb{N}} \subset \mathbb{R} and let ℓ:N→Λ⊂N,n↦ℓ(n)\ell: \mathbb{N} \to \Lambda \subset \mathbb{N}, n \mapsto \ell(n), be a strictly monotonically increasing sequence of natural numbers. The composition (aℓ(n))(a_{\ell(n)}) of ℓ(n)\ell(n) and ana_n is called a subsequence of ana_n.

Remark

It is important here that Λ\Lambda contains infinitely many elements. If Λ\Lambda were finite, you would not get a subsequence at all, just a list of finitely many numbers!

Example 3.1

Let (an)n∈N=(−1)n(a_n)_{n \in \mathbb{N}} = (-1)^n be a sequence in R\mathbb{R}. Then, among others, there are the following two subsequences:

{1,1,1,...}(Λ=2N={even numbers}){−1,−1,−1,...}(Λ=2N+1={odd numbers})\begin{aligned} \{1, 1, 1, ...\} \qquad &( \Lambda = 2\mathbb{N} = \{\text{even numbers}\})\\ \{-1, -1, -1, ...\} \qquad &( \Lambda = 2\mathbb{N}+1 = \{\text{odd numbers}\}) \end{aligned}
Definition 3.2

a∈Ra \in \mathbb{R} is called an accumulation point of (an)n∈N(a_n)_{n \in \mathbb{N}} if (an)n∈N(a_n)_{n \in \mathbb{N}} has a subsequence (aℓ(n))(a_{\ell(n)}) converging to aa. That is, there is a subsequence (aℓ(n))(a_{\ell(n)}) with

lim⁡ℓ(n)→∞aℓ(n)=a.\lim_{\ell(n) \to \infty} a_{\ell(n)} = a.
Satz Bolzano-Weierstrass (3.4.1)

Every bounded sequence (an)n∈N(a_n)_{n \in \mathbb{N}} in R\mathbb{R} has a convergent subsequence, and hence also an accumulation point.

Remark
an bounded⇏an convergesan bounded⟹Bolzano-Weierstrass∃ subsequence (aℓ(n)) that converges.\begin{aligned} a_n \text{ bounded} \quad &\nRightarrow \quad a_n \text{ converges}\\ a_n \text{ bounded} \quad &\stackrel{ \text{Bolzano-Weierstrass} }{\Longrightarrow} \quad \exists \text{ subsequence } (a_{\ell(n)}) \text{ that converges}. \end{aligned}

Limit Superior and Limit Inferior

Definition 4.1 limit superior and limit inferior

Let (an)n∈N(a_n)_{n \in \mathbb{N}} be bounded, i.e.

∃M∈R,  ∀n∈N:∣an∣<M.\exists M \in \mathbb{R}, \; \forall n \in \mathbb{N}: \quad |a_n| < M.

Then, for k∈Nk \in \mathbb{N}, by the theorem on sup and inf there exist

ck=inf⁡n≥kan≤sup⁡n≥kan=bkc_k = \inf_{n \geq k} a_n \leq \sup_{n \geq k} a_n = b_k

Hence

−M≤c1≤...≤ck≤ck+1≤bk+1≤bk≤...≤b1≤M,∀k∈N-M \leq c_1 \leq ... \leq c_k \leq c_{k+1} \leq b_{k+1} \leq b_k \leq ... \leq b_1 \leq M, \quad \forall k \in \mathbb{N}

The monotone convergence theorem implies the existence of

"limit superior":lim sup⁡n→∞an:=lim⁡k→∞bk=b"limit inferior":lim inf⁡n→∞an:=lim⁡k→∞ck=c\begin{aligned} \text{"limit superior"}: \qquad \limsup_{n \to \infty} a_n &:= \lim_{k \to \infty} b_k = b\\ \text{"limit inferior"}: \qquad \liminf_{n \to \infty} a_n &:= \lim_{k \to \infty} c_k = c \end{aligned}

and c≤bc \leq b holds by the theorem on the rules for limits (rule (iv)).

Limit superior and limit inferior of a bounded sequence
Satz

We have

lim⁡n→∞an=a⇔lim sup⁡n→∞an=lim inf⁡n→∞an=lim⁡n→∞an=a.\lim_{n \to \infty} a_n = a \quad \Leftrightarrow \quad \limsup_{n \to \infty} a_n = \liminf_{n \to \infty} a_n = \lim_{n \to \infty} a_n = a.
Remark

Let (an)n∈N(a_n)_{n \in \mathbb{N}} be bounded. Let a+:=lim sup⁡an,  a−:=lim inf⁡ana_+ := \limsup a_n, \; a_- := \liminf a_n (a+,a−a_+, a_- are accumulation points.)
Then:

  1. ∀ε>0\forall \varepsilon > 0 there are only finitely many n∈Z>0n \in \mathbb{Z}_{>0} with an∉(a−−ε,a++ε)a_n \notin (a_- -\varepsilon, a_+ +\varepsilon).
  2. a+a_+ is the largest accumulation point and a−a_- is the smallest accumulation point.
  3. Furthermore: if a+=a−a_+=a_-, then (an)n∈N(a_n)_{n \in \mathbb{N}} is convergent with lim⁡n→∞an=a+=a−  (=a).\lim_{n \to \infty} a_n = a_+=a_- \;(= a).
  4. Conversely, every subsequence of a sequence an→n→∞aa_n \xrightarrow{n \to \infty} a also converges to aa, and a=a+=a−a = a_+=a_-.
Example 4.1

Compute lim sup⁡an\limsup a_n and lim inf⁡an\liminf a_n of the sequence:

an=(−1)n⋅n+5n={(−1)⋅n+5n,for n oddn+5n,for n even\begin{aligned} a_n &= (-1)^n \cdot \frac{n+5}{n}\\ &= \begin{cases} (-1) \cdot \frac{n+5}{n}, & \text{for $n$ odd}\\ \frac{n+5}{n}, & \text{for $n$ even} \end{cases} \end{aligned}
Show solutionHide solution

We define ck:=(−1)⋅k+5kc_k := (-1) \cdot \frac{k+5}{k} and bk:=k+5kb_k := \frac{k+5}{k}

  • lim inf⁡n→∞an=lim⁡k→∞ck=−1lim inf⁡n→∞(−1)n⋅n+5n=lim⁡k→∞(−1)⋅k+5k=lim⁡k→∞−k+5k=lim⁡k→∞−k(1+5k)k=lim⁡k→∞−1+5k1=lim⁡k→∞−1−5k=−1−0=−1\begin{aligned} \liminf_{n \to \infty} a_n &= \lim_{k \to \infty} c_k = -1\\ \liminf_{n \to \infty} (-1)^n \cdot \frac{n+5}{n} &= \lim_{k \to \infty} (-1) \cdot \frac{k+5}{k}\\ &= \lim_{k \to \infty} -\frac{k+5}{k}\\ &= \lim_{k \to \infty} -\frac{k(1+\frac{5}{k})}{k}\\ &= \lim_{k \to \infty} -\frac{1+\frac{5}{k}}{1}\\ &= \lim_{k \to \infty} -1-\frac{5}{k}\\ &= -1-0\\ &= -1 \end{aligned}
  • lim sup⁡n→∞an=lim⁡k→∞bk=1lim sup⁡n→∞  (−1)n⋅n+5n=lim⁡k→∞k+5k=lim⁡k→∞k(1+5k)k=lim⁡k→∞1+5k1=1+0=1\begin{aligned} \limsup_{n \to \infty} a_n &= \lim_{k \to \infty} b_k = 1\\ \limsup_{n \to \infty} \; (-1)^n \cdot \frac{n+5}{n} &= \lim_{k \to \infty} \frac{k+5}{k}\\ &= \lim_{k \to \infty} \frac{k(1+\frac{5}{k})}{k}\\ &= \lim_{k \to \infty} \frac{1+\frac{5}{k}}{1}\\ &= 1+0\\ &= 1 \end{aligned}