Study Guides / Calculus I / Exercise session 6
EXERCISE SESSION 6 · ETH ZURICH · FS20 · calculus

Partial Fractions, Power Series and Continuity

Partial fraction decomposition, power series representation, continuity, continuous extension, intermediate value theorem

SB
Steven Battilana Quant · ex-ETH

Uncategorised

Partial fraction decomposition ("light", explained by example)

1(x+2)(x−1)=!Ax+2+Bx−1=A(x−1)+B(x+2)(x+2)(x−1)=(A+B)x+(2B−A)⋅1(x+2)(x−1)\begin{aligned} \frac{1}{(x+2)(x-1)} &\stackrel{!}{=} \frac{A}{x+2} + \frac{B}{x-1}\\ &= \frac{A(x-1) + B(x+2)}{(x+2)(x-1)}\\ &= \frac{(A+B)x + (2B-A) \cdot 1}{(x+2)(x-1)} \end{aligned}

Comparing coefficients:

1=!(A+B)x+(2B−A)⋅1⇒A+B=0∧2B−A=1⇔A=−B(∗)⇒2B−A=1⟹A=−B2B−(−B)=1⇔3B=1⇔B=13⟹(∗)  ∧  B=13A=−13⟹1(x+2)(x−1)=−13x+2+13x−1=−13(x+2)+13(x−1)\begin{aligned} 1 &\stackrel{!}{=} (A+B)x + (2B-A) \cdot 1\\ &\Rightarrow A+B = 0 \quad \wedge \quad 2B-A = 1\\ &\Leftrightarrow A = -B \quad\quad (*)\\ &\Rightarrow 2B-A = 1\\ &\stackrel{A = -B}{\Longrightarrow} 2B-(-B) = 1\\ &\Leftrightarrow 3B = 1\\ &\Leftrightarrow B = \frac{1}{3}\\ &\stackrel{(*)\; \wedge \; B = \frac{1}{3}}{\Longrightarrow} A = -\frac{1}{3}\\ &\Longrightarrow \quad \frac{1}{(x+2)(x-1)} = \frac{-\frac{1}{3}}{x+2} + \frac{\frac{1}{3}}{x-1} = -\frac{1}{3(x+2)} + \frac{1}{3(x-1)} \end{aligned}

Note that we have now split one fraction into two fractions.

Power Series (Reminder)

Definition 2.1

A power series is a series of the form

∑n=0∞anxn=a0+a1x+a2x2+...+anxn+...\sum_{n=0}^\infty a_n x^n = a_0 + a_1 x + a_2 x^2 + ... + a_n x^n + ...

where xx is a real (or complex) variable and (an)n∈N(a_n)_{n \in \mathbb{N}} is a real (or complex) sequence.
Sometimes the more general notion of a power series is given with a centre x0x_0

∑n=0∞an(x−x0)n.\sum_{n=0}^\infty a_n (x-x_0)^n.
Definition 2.2

The radius of convergence is defined as the supremum of all numbers ρ≥0\rho \geq 0 for which the power series converges for all xx with ∣x−x0∣<ρ|x-x_0| < \rho:

ρ:=sup⁡{∣x−x0∣    ∣    ∑n=0∞an(x−x0)n   is convergent}\rho := \sup \left\{ |x-x_0| \;\; \left| \;\; \sum_{n=0}^\infty a_n(x-x_0)^n \; \text{ is convergent} \right. \right\}
Satz radius of convergence

For the radius of convergence ρ\rho of the power series f(x)=∑n=0∞anxnf(x) = \sum\limits_{n=0}^\infty a_n x^n the following formulas hold:

  1. ρ=lim⁡n→∞∣anan+1∣\rho = \lim\limits_{n \to \infty} \left| \frac{a_n}{a_{n+1}} \right| \quad (for !,  xn,  ...!, \; x^n, \; ...)
  2. ρ=1lim sup⁡n→∞∣an∣n\rho = \frac{1}{\limsup\limits_{n \to \infty} \sqrt[n]{|a_n|}} \quad (for (⋅)n,xn,!,...(\cdot)^n, x^n,!, ...).
Proof of (i)

We apply the ratio test to the series ∑n=0∞anxn\sum\limits_{n=0}^\infty a_n x^n. We obtain absolute convergence if

lim⁡n→∞∣an+1xn+1anxn∣=lim⁡n→∞∣an+1xan∣=lim⁡n→∞∣an+1an∣∣x∣=∣x∣⋅lim⁡n→∞∣an+1an∣<!1.⇔∣x∣<1lim⁡n→∞∣an+1an∣=lim⁡n→∞∣anan+1∣=:ρ.\begin{aligned} \lim_{n \to \infty} \left| \frac{a_{n+1} x^{n+1}}{a_n x^n} \right| &= \lim_{n \to \infty} \left| \frac{a_{n+1} x}{a_n} \right|\\ &= \lim_{n \to \infty} \left| \frac{a_{n+1}}{a_n} \right| |x|\\ &= |x| \cdot \lim_{n \to \infty} \left| \frac{a_{n+1}}{a_n} \right| \stackrel{!}{<} 1.\\ \Leftrightarrow \quad |x| &< \frac{1}{ \lim_{n \to \infty} \left| \frac{a_{n+1}}{a_n} \right| }\\ &= \lim_{n \to \infty} \left| \frac{a_{n}}{a_{n+1}} \right| =: \rho. \end{aligned}□\square
Proof of (ii)

We apply the root test to the series ∑n=0∞anxn\sum\limits_{n=0}^\infty a_n x^n. We obtain absolute convergence if

lim sup⁡n→∞∣anxn∣n=lim sup⁡n→∞∣an∣∣xn∣n=lim sup⁡n→∞∣an∣n∣xn∣n=lim sup⁡n→∞∣an∣n⋅∣x∣=∣x∣⋅lim sup⁡n→∞∣an∣n<!1.⇔∣x∣<1lim sup⁡n→∞∣an∣n=:ρ.\begin{aligned} \limsup_{n \to \infty} \sqrt[n]{|a_n x^n|} &= \limsup_{n \to \infty} \sqrt[n]{|a_n| |x^n|}\\ &= \limsup_{n \to \infty} \sqrt[n]{|a_n|} \sqrt[n]{|x^n|}\\ &= \limsup_{n \to \infty} \sqrt[n]{|a_n|} \cdot |x|\\ &= |x| \cdot \limsup_{n \to \infty} \sqrt[n]{|a_n|} \stackrel{!}{<} 1.\\ \Leftrightarrow \quad |x| &< \frac{1}{ \limsup_{n \to \infty} \sqrt[n]{|a_n|} } =: \rho. \end{aligned}□\square
Remark

Both formulas follow directly from the ratio test and the root test for (i) and (ii) respectively.

Remark

As with the ratio and root tests, absolute convergence follows from (i) and (ii).

Remark Important

On the boundary of the disc of convergence, i.e. in the case ∣x−x0∣=ρ|x-x_0| = \rho, no statement about convergence is possible. You therefore have to consider this case separately.

Example 2.1

For which x∈Rx \in \mathbb{R} does the following power series converge?

∑n=1∞1n2(n2+n−n2+1)n(x+1)n\sum_{n=1}^\infty \frac{1}{n^2} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right)^n (x+1)^n
Show solutionHide solution

We compute the radius of convergence using formula (ii):

ρ~=lim sup⁡n→∞∣an∣n=lim sup⁡n→∞∣1n2⏟>0(n2+n−n2+1⏟≥0)n∣n=lim sup⁡n→∞1n2(n2+n−n2+1)nn=lim sup⁡n→∞1n2n(n2+n−n2+1)=lim sup⁡n→∞1(nn)2(n2+n−n2+1)=lim sup⁡n→∞1(nn)2(n2+n−n2+1)⋅n2+n+n2+1n2+n+n2+1=lim sup⁡n→∞1(nn)2n2+n−(n2+1)n2+n+n2+1=lim sup⁡n→∞1(nn)2n2+n−n2−1n2+n+n2+1=lim sup⁡n→∞1(nn)2n−1n2+n+n2+1=lim sup⁡n→∞1(nn)2n(1−1n)n2(1+1n)+n2(1+1n2)=lim sup⁡n→∞1(nn)2n(1−1n)∣n∣1+1n+∣n∣1+1n2=n>0lim sup⁡n→∞1(nn)2n(1−1n)n1+1n+n1+1n2=lim sup⁡n→∞1(nn)2n(1−1n)n(1+1n+1+1n2)=lim sup⁡n→∞1(nn)21−1n1+1n+1+1n2=1121−01+0+1+0=12⇒ρ=1ρ~=2\begin{aligned} \widetilde{\rho} &= \limsup_{n \to \infty} \sqrt[n]{|a_n|}\\ &= \limsup_{n \to \infty} \sqrt[n]{\left| \underbrace{\frac{1}{n^2}}_{>0} \left( \underbrace{\sqrt{n^2+n} - \sqrt{n^2+1}}_{\geq 0} \right)^n \right|}\\ &= \limsup_{n \to \infty} \sqrt[n]{ \frac{1}{n^2} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right)^n }\\ &= \limsup_{n \to \infty} \frac{1}{\sqrt[n]{n^2}} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right)\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right)\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right) \cdot \frac{ \sqrt{n^2+n} + \sqrt{n^2+1} }{ \sqrt{n^2+n} + \sqrt{n^2+1} }\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n^2+n - (n^2+1) }{ \sqrt{n^2+n} + \sqrt{n^2+1} }\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n^2+n - n^2-1 }{ \sqrt{n^2+n} + \sqrt{n^2+1} }\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n-1 }{ \sqrt{n^2+n} + \sqrt{n^2+1} }\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n(1-\frac{1}{n}) }{ \sqrt{n^2(1+\frac{1}{n})} + \sqrt{n^2(1+\frac{1}{n^2}) }}\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n(1-\frac{1}{n}) }{ |n| \sqrt{1+\frac{1}{n}} + |n| \sqrt{1+\frac{1}{n^2} }}\\ &\stackrel{n>0}{=} \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n(1-\frac{1}{n}) }{ n \sqrt{1+\frac{1}{n}} + n \sqrt{1+\frac{1}{n^2} }}\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ n(1-\frac{1}{n}) }{ n \left( \sqrt{1+\frac{1}{n}} + \sqrt{1+\frac{1}{n^2} } \right)}\\ &= \limsup_{n \to \infty} \frac{1}{(\sqrt[n]{n})^2} \frac{ 1-\frac{1}{n} }{ \sqrt{1+\frac{1}{n}} + \sqrt{1+\frac{1}{n^2}} }\\ &= \frac{1}{1^2} \frac{ 1-0 }{ \sqrt{1+0} + \sqrt{1+0} }\\ &= \frac{1}{2}\\ \Rightarrow \quad \rho &= \frac{1}{\widetilde{\rho}} = 2 \end{aligned}

So the power series converges for ∣x+1∣<2|x+1| < 2 (∗)(*) and diverges for ∣x+1∣>2|x+1| > 2. Now we compute the region of convergence in terms of xx for (∗)(*):

  • ⇒(∗)−(x+1)=2⇔−x−1=2⇔−x=3⇔x=−3\begin{aligned} &\stackrel{(*)}{\Rightarrow} \quad -(x+1) = 2\\ &\Leftrightarrow \quad -x-1 = 2\\ &\Leftrightarrow \quad -x = 3\\ &\Leftrightarrow \quad x = -3 \end{aligned}
  • ⇒(∗)x+1=2⇔x=1\begin{aligned} &\stackrel{(*)}{\Rightarrow} \quad x+1 = 2\\ &\Leftrightarrow \quad x = 1 \end{aligned}
  • ⟹∣x+1∣<2⇔x∈(−3,1)\Longrightarrow \quad |x+1| < 2 \Leftrightarrow x \in (-3,1)

Now we only have to cover the case ∣x+1∣=2|x+1| = 2:

∑n=1∞1n2(n2+n−n2+1)n(±2)n=...derivation analogous to the one above=∑n=1∞1n2(1−1n1+1n+1+1n2)n(±2)n=∑n=1∞1n2((±2)(1−1n)1+1n+1+1n2)n≤∑n=1∞∣1n2((±2)(1−1n)1+1n+1+1n2)n∣≤∑n=1∞1n2\begin{aligned} \sum_{n=1}^\infty \frac{1}{n^2} \left( \sqrt{n^2+n} - \sqrt{n^2+1} \right)^n (\pm 2)^n &= \stackrel{\text{derivation analogous to the one above}}{...}\\ &= \sum_{n=1}^\infty \frac{1}{n^2} \left( \frac{ 1-\frac{1}{n} }{ \sqrt{1+\frac{1}{n}} + \sqrt{1+\frac{1}{n^2}} } \right)^n (\pm 2)^n\\ &= \sum_{n=1}^\infty \frac{1}{n^2} \left( \frac{ (\pm 2) (1-\frac{1}{n}) }{ \sqrt{1+\frac{1}{n}} + \sqrt{1+\frac{1}{n^2}} } \right)^n\\ &\leq \sum_{n=1}^\infty \left| \frac{1}{n^2} \left( \frac{ (\pm 2) (1-\frac{1}{n}) }{ \sqrt{1+\frac{1}{n}} + \sqrt{1+\frac{1}{n^2}} } \right)^n \right|\\ &\leq \sum_{n=1}^\infty \frac{1}{n^2} \end{aligned}

since the fraction in brackets has absolute value ≤1\leq 1 (numerator ≤2\leq 2, denominator ≥2\geq 2).

So we have a convergent majorant. Hence the power series converges for x=1x=1 and x=−3x=-3.
In summary: the power series converges absolutely for x∈[−3,1]x \in [-3,1] and diverges otherwise.

Example 2.2

Determine the region of convergence of: ∑n=0∞(−1)nn+1(x2−1)n\sum\limits_{n=0}^\infty \frac{(-1)^n}{n+1} (x^2-1)^n.

Show solutionHide solution

We first consider ∑n=0∞(−1)nn+1yn\sum\limits_{n=0}^\infty \frac{(-1)^n}{n+1} y^n (∗)(*) with y:=(x2−1)y:=(x^2-1). Now we determine the radius of convergence using formula (i):

ρ=lim⁡n→∞∣anan+1∣=lim⁡n→∞∣1n+1(n+1)+11∣=lim⁡n→∞∣n+2n+1∣=n>0lim⁡n→∞n+2n+1=lim⁡n→∞n(1+2n)n(1+1n)=lim⁡n→∞1+2n1+1n=1\begin{aligned} \rho &= \lim_{n \to \infty} \left| \frac{a_n}{a_{n+1}} \right|\\ &= \lim_{n \to \infty} \left| \frac{1}{n+1} \frac{(n+1)+1}{1} \right|\\ &= \lim_{n \to \infty} \left| \frac{n+2}{n+1} \right|\\ &\stackrel{n>0}{=} \lim_{n \to \infty} \frac{n+2}{n+1}\\ &= \lim_{n \to \infty} \frac{n(1+\frac{2}{n})}{n(1+\frac{1}{n})}\\ &= \lim_{n \to \infty} \frac{1+\frac{2}{n}}{1+\frac{1}{n}}\\ &= 1 \end{aligned}

The power series (∗)(*) therefore converges absolutely for ∣y∣<1|y| < 1 and diverges for ∣y∣>1|y| > 1. Now we consider the case y=1y = 1:

∑n=0∞(−1)nn+11n=∑n=0∞(−1)nn+1(∗∗)\sum_{n=0}^\infty \frac{(-1)^n}{n+1} 1^n = \sum_{n=0}^\infty \frac{(-1)^n}{n+1} \qquad (**)
  1. lim⁡n→∞1n+1=0\lim\limits_{n \to \infty} \frac{1}{n+1} = 0
  2. 1n+1≥0\frac{1}{n+1} \geq 0
  3. an≥an+1⇔an−an+1≥0⇔1n+1−1(n+1)+1≥0⇔1n+1−1n+2≥0⇔n+2−(n+1)(n+1)(n+2)≥0⇔n+2−n−1(n+1)(n+2)≥0⇔1(n+1)(n+2)≥0⇒so an is monotonically decreasing\begin{aligned} a_{n} \geq a_{n+1} \quad &\Leftrightarrow \quad a_n - a_{n+1} \geq 0\\ &\Leftrightarrow \quad \frac{1}{n+1} - \frac{1}{(n+1)+1} \geq 0\\ &\Leftrightarrow \quad \frac{1}{n+1} - \frac{1}{n+2} \geq 0\\ &\Leftrightarrow \quad \frac{n+2 - (n+1)}{(n+1)(n+2)} \geq 0\\ &\Leftrightarrow \quad \frac{n+2 - n-1}{(n+1)(n+2)} \geq 0\\ &\Leftrightarrow \quad \frac{1}{(n+1)(n+2)} \geq 0\\ &\Rightarrow \quad \text{so $a_n$ is monotonically decreasing} \end{aligned}

    ⟹Leibniz test\stackrel{\text{Leibniz test}}{\Longrightarrow} \quad (∗∗)(**) converges.

Now we consider the case y=−1y=-1:

∑n=0∞(−1)nn+1(−1)n=∑n=0∞(−1)2nn+1=∑n=0∞1n+1=∑n=1∞1n\begin{aligned} \sum_{n=0}^\infty \frac{(-1)^n}{n+1} (-1)^n &= \sum_{n=0}^\infty \frac{(-1)^{2n}}{n+1}\\ &= \sum_{n=0}^\infty \frac{1}{n+1}\\ &= \sum_{n=1}^\infty \frac{1}{n} \end{aligned}

We have obtained the harmonic series and therefore know that the series diverges in this case. Hence (∗)(*) converges for y∈(−1,1]y \in (-1,1]. Finally, we have to substitute back y=x2−1y = x^2-1 and get:

−1<y≤1⇔−1<x2−1≤1⇔0<x2≤2⟹B=[−2,0)∪(0,2].\begin{aligned} -1 < y \leq 1 \quad &\Leftrightarrow \quad -1 < x^2-1 \leq 1\\ &\Leftrightarrow \quad 0 < x^2 \leq 2\\ &\Longrightarrow \quad B = [-\sqrt{2},0) \cup (0, \sqrt{2}]. \end{aligned}

Power Series Representation

Definition 3.1

The exponential, sine and cosine functions have the following power series representations:

  1. Exp(xx) =∑n=0∞xnn!= \sum\limits_{n=0}^\infty \frac{x^n}{n!}
  2. Sin(xx) =∑n=0∞(−1)n⋅x2n+1(2n+1)!= \sum\limits_{n=0}^\infty (-1)^n \cdot \frac{x^{2n+1}}{(2n+1)!}
  3. Cos(xx) =∑n=0∞(−1)n⋅x2n(2n)!= \sum\limits_{n=0}^\infty (-1)^n \cdot \frac{x^{2n}}{(2n)!}
Example 3.1

Using the power series representation of Exp, compute the following limit: lim⁡x→0Exp(x)−1x\lim\limits_{x \to 0} \frac{\text{Exp}(x)-1}{x}.

Show solutionHide solution Exp(x)−1x=Exp(x)x⏟(i)−1x(1)\frac{\text{Exp}(x)-1}{x} = \underbrace{\frac{\text{Exp}(x)}{x}}_{(i)} - \frac{1}{x} \tag{1}

We now consider (i):

Exp(x)x=∑n=0∞xnn!x(2)\frac{\text{Exp}(x)}{x} = \frac{\sum_{n=0}^\infty \frac{x^n}{n!}}{x} \tag{2}Exp(x)x=∑n=0∞xn−1n!(3)\phantom{\frac{\text{Exp}(x)}{x}} = \sum_{n=0}^\infty \frac{x^{n-1}}{n!} \tag{3}

Now we have to check whether (3) converges and, if so, for which xx. We therefore apply the ratio test:

lim⁡n→∞∣an+1an∣=lim⁡n→∞∣x(n+1)−1(n+1)!n!xn−1∣=lim⁡n→∞∣xnn!(n+1)n!xn−1∣=lim⁡n→∞∣x(n+1)∣=lim⁡n→∞∣x∣(n+1)=0<1⇒the series converges absolutely.\begin{aligned} \lim_{n \to \infty} \left| \frac{a_{n+1}}{a_n} \right| &= \lim_{n \to \infty} \left| \frac{x^{(n+1)-1}}{(n+1)!} \frac{n!}{x^{n-1}} \right|\\ &= \lim_{n \to \infty} \left| \frac{x^n}{n!(n+1)} \frac{n!}{x^{n-1}} \right|\\ &= \lim_{n \to \infty} \left| \frac{x}{(n+1)}\right|\\ &= \lim_{n \to \infty} \frac{|x|}{(n+1)}\\ &= 0 < 1 \quad \Rightarrow \quad \text{the series converges absolutely.} \end{aligned}

Now we can finally compute (1)(1); to do so, we write out the first few terms of the power series representation of Exp.

lim⁡x→0Exp(x)−1x=lim⁡x→0Exp(x)x−1x=lim⁡x→0Exp(x)x⏟∑n=0∞xn−1n!=1x+1+O(x)−1x=lim⁡x→01x+1+O(x)−1x=lim⁡x→01+O(x)⏟→x→00=1.\begin{aligned} \lim_{x \to 0} \frac{\text{Exp}(x)-1}{x} &= \lim_{x \to 0} \frac{\text{Exp}(x)}{x} - \frac{1}{x}\\ &= \lim_{x \to 0} \underbrace{\frac{\text{Exp}(x)}{x}}_{\sum\limits_{n=0}^\infty \frac{x^{n-1}}{n!} = \frac{1}{x} + 1 + O(x)} - \frac{1}{x}\\ &= \lim_{x \to 0} \frac{1}{x} + 1 + O(x) - \frac{1}{x}\\ &= \lim_{x \to 0} 1 + \underbrace{O(x)}_{\xrightarrow{x \to 0} 0}\\ &= 1. \end{aligned}

Continuity

Definition 4.1 (4.1.2)

Let f:Ω⊂Rd→Rn,  x0∈Ω‾,  a∈Rnf: \Omega \subset \mathbb{R}^d \to \mathbb{R}^n, \; x_0 \in \overline{\Omega}, \; a \in \mathbb{R}^n.
ff has the limit aa at the point x0x_0 if for every sequence (xk)k∈N(x_k)_{k \in \mathbb{N}} in Ω\Omega with xk→k→∞x0x_k \xrightarrow{k \to \infty} x_0 we have f(xk)→k→∞af(x_k) \xrightarrow{k \to \infty} a.
Notation: lim⁡x→x0f(x)=f(x0)=a\lim\limits_{x \to x_0} f(x) = f(x_0)=a.

Remark

The definition of the limit via sequences above can alternatively be written as follows:

lim⁡x→x0f(x)=a⇔∀(xn)n∈N⊆Ω with lim⁡n→∞xn=x0, we have lim⁡n→∞f(xn)=a.\lim_{x \to x_0} f(x) = a \quad \Leftrightarrow \quad \forall (x_n)_{n \in \mathbb{N}} \subseteq \Omega \text{ with }\lim_{n \to \infty} x_n = x_0, \text{ we have } \lim_{n \to \infty} f(x_n) = a.
Definition 4.2 (4.1.3)

Let Ω⊂Rd,  f:Ω→Rn\Omega \subset \mathbb{R}^d, \; f: \Omega \to \mathbb{R}^n.

  1. ff is called continuous at the point x0∈Ωx_0 \in \Omega if lim⁡x→x0f(x)=f(x0)=:a\lim\limits_{x \to x_0} f(x) = f(x_0) =: a exists.

  2. ff is called continuously extendable at the point x0∈Ω‾∖Ωx_0 \in \overline{\Omega} \setminus \Omega if lim⁡x→x0f(x)=:a\lim\limits_{x \to x_0} f(x) =:a exists.
    (In this case the function ff extended by f(x0)=af(x_0) = a is continuous at the point x0x_0.)

Definition 4.3 (4.2.1)

Let f:Ω⊂Rd→Rnf: \Omega \subset \mathbb{R}^d \to \mathbb{R}^n.
ff is called continuous on Ω\Omega if ff is continuous at every point x0∈Ωx_0 \in \Omega.

Remark

To check a given function f:Ω⊂R→Rf: \Omega \subset \mathbb{R} \to \mathbb{R} for continuity at the point x0x_0, you therefore have to show the following three points:

  1. ff must be defined on Ω\Omega
  2. lim⁡x→x0f(x)\lim\limits_{x \to x_0} f(x) exists, i.e. lim⁡x→x0f(x)≠∞\lim\limits_{x \to x_0} f(x) \neq \infty and lim⁡x→x0+f(x)=lim⁡x→x0−f(x)\lim\limits_{x \to x_0^+} f(x) = \lim\limits_{x \to x_0^-} f(x)
  3. lim⁡x→x0f(x)=f(x0)\lim\limits_{x \to x_0} f(x) = f(x_0)
Satz (Rules for continuous functions)

Let f:Ω→Rf: \Omega \to \mathbb{R} and g:Ω→Rg: \Omega \to \mathbb{R} be continuous. Then

  1. f+gf+g
  2. f−gf-g
  3. fgfg
  4. fg\frac{f}{g} if g≠0g \neq 0

are continuous.
If moreover f:Ω→Rf: \Omega \to \mathbb{R} and g:f(Ω)→Rg: f(\Omega) \to \mathbb{R} are continuous, then so is g∘f:Ω→Rg \circ f: \Omega \to \mathbb{R}.

Proof of (i)

This statement follows directly from the sum rule for limits. Let a∈Ωa \in \Omega. Since ff and gg are continuous at the point aa, we have

lim⁡x→af(x)=f(a)andlim⁡x→ag(x)=g(a).\lim_{x \to a} f(x) = f(a) \quad \text{and} \quad \lim_{x \to a} g(x) = g(a).

By the rule "the limit of the sum equals the sum of the limits" it follows that

lim⁡x→a(f(x)+g(x))=lim⁡x→af(x)+lim⁡x→ag(x)=f(a)+g(a)=(f+g)(a).\lim_{x \to a} (f(x) + g(x)) = \lim_{x \to a} f(x) + \lim_{x \to a} g(x) = f(a) + g(a) = (f+g)(a).

Hence f+gf+g is continuous at the point aa. □\square
(Remark. The proof for the other rules is analogous.)

Remark

The exponential function, sine, cosine, tangent (between the zeros of the cosine), the hyperbolic functions and the logarithm are all continuous functions.

Remark Important

The composition of continuous maps is also continuous. Excluded, of course, are problem points, e.g. the denominator vanishes or the definition of the function changes.

Example 4.1

Is the following function continuous?

f(x)={x2−2,for x<−21x−3,for x≥−2f(x) = \begin{cases} x^2 -2, & \text{for } x < -2\\ \frac{1}{x-3}, & \text{for } x \geq -2 \end{cases}
Show solutionHide solution

To the left of −2-2, the function f(x)f(x) is a composition of continuous functions and is therefore also continuous. To the right of −2-2, the function f(x)f(x) is a composition of right-continuous functions and is therefore right-continuous, except at the point x=3x=3.

  1. lim⁡x→(−2)−x2−2=2\lim\limits_{x \to (-2)^-} x^2-2 = 2
  2. f(−2)=−15f(-2) = -\frac{1}{5}

⟹(i)&(ii)2≠−15⇒f(x)\stackrel{(i) \& (ii)}{\Longrightarrow} \quad 2 \neq -\frac{1}{5} \quad \Rightarrow \quad f(x) is not continuous at the point −2-2.

Example 4.2

Is the following function continuous?

f(x)={x2,for x≤1x2−4x+4,for x>1f(x) = \begin{cases} x^2, & \text{for } x \leq 1\\ x^2 -4x+4, & \text{for } x > 1 \end{cases}
Show solutionHide solution

To the left of 11, the function f(x)f(x) is a composition of left-continuous functions and is therefore also left-continuous. To the right of 11, the function f(x)f(x) is a composition of continuous functions and is therefore also continuous.

  1. f(1)=1f(1) = 1
  2. lim⁡x→1+x2−4x+4=1\lim\limits_{x \to 1^+} x^2 -4x +4 = 1

⟹(i)&(ii)(i)=(ii)⇒f(x)\stackrel{(i) \& (ii)}{\Longrightarrow} \quad (i) = (ii) \quad \Rightarrow \quad f(x) is continuous at the point 11.

Example 4.3

Consider the function

f(x)={8a+16xfor x<2a2(x+2)for x≥2f(x) = \begin{cases} 8a + 16x & \text{for } x <2\\ a^2 (x+2) & \text{for } x \geq 2 \end{cases}

Determine a∈Ra \in \mathbb{R} such that f(x)f(x) is continuous at x=2x=2.

Show solutionHide solution

The functions x↦8a+16xx \mapsto 8a + 16x and x↦a2(x+2)x \mapsto a^2(x+2) are continuous. Therefore the function ff is continuous away from the point 2, and it is right-continuous at the point 2.
For it to be continuous at the point 2 as well, the following must additionally hold:

lim⁡x→2−f(x)=!f(2)⇔lim⁡x→2−8a+16x=a2⋅4⇔8a+32=a2⋅4⇔4a2−8a−32=0⇔a1,2=8±64−4⋅4⋅(−32)2⋅4=8±5768=8±248={4−2\begin{aligned} \lim_{x \to 2^-} f(x) &\stackrel{!}{=} f(2)\\ \Leftrightarrow \quad \lim_{x \to 2^-} 8a + 16x &= a^2 \cdot 4\\ \Leftrightarrow \quad 8a + 32 &= a^2 \cdot 4\\ \Leftrightarrow \quad 4a^2 -8a -32 &= 0\\ \Leftrightarrow \quad a_{1,2} &= \frac{8 \pm \sqrt{64 - 4 \cdot 4 \cdot (-32)}}{2 \cdot 4}\\ &= \frac{8 \pm \sqrt{576}}{8}\\ &= \frac{8 \pm 24}{8} = \begin{cases} 4\\ -2 \end{cases} \end{aligned}

It follows that ff is continuous at the point x=2x=2 for a=−2a = -2 and a=4a = 4.

Satz intermediate value theorem

Let f:[a,b]→Rf: [a,b] \to \mathbb{R} be continuous with f(a)≤f(b)f(a) \leq f(b). Then for every y∈[f(a),f(b)]y \in [f(a),f(b)] there exists an x∈[a,b]x \in [a,b] with f(x)=yf(x) = y.

Example 4.4

Show: the polynomial f(x)=4x3−6x2+3x−2f(x) = 4x^3 -6x^2 +3x -2 has a zero lying between 11 and 22.

Show solutionHide solution
  • Continuity: ff is continuous on [1,2][1,2] because polynomials are compositions of continuous functions.

  • We evaluate ff at the boundary, i.e. at the points x=1x=1 and x=2x=2:

    f(1)=4−6+3−2=−1<0f(2)=32−24+6−2>0\begin{aligned} f(1) &= 4-6+3-2 = -1 < 0\\ f(2) &= 32-24+6-2 > 0 \end{aligned}

    By the intermediate value theorem, ff must therefore take all values between f(1)f(1) (negative) and f(2)f(2) (positive) on the interval [1,2][1,2]. In particular, there is an x∗∈(1,2)x^* \in (1,2) with f(x∗)=0f(x^*) = 0. So x∗x^* is the zero of the polynomial we are looking for.

Example 4.5

Does the equation sin⁡(2x)−ex=−5\sin(2x) -e^x = -5 have a solution in the interval [0,π][0,\pi]?

Show solutionHide solution

We define the function f(x)=sin⁡(2x)−ex+5f(x) = \sin(2x) -e^x +5.

sin⁡(2x)−ex=−5 has a solution in [0,π]⇔f(x)=0\sin(2x) -e^x = -5 \text{ has a solution in } [0,\pi] \quad \Leftrightarrow \quad f(x) = 0
  • Continuity: since sine and the exponential function are continuous, ff is a continuous function.

  • We evaluate ff at the boundary, i.e. at the points x=0x=0 and x=πx=\pi:

    f(0)=sin⁡(0)−e0+5=4≥0f(π)=sin⁡(2π)−eπ+5=−eπ+5≤0\begin{aligned} f(0) &= \sin(0) - e^0 +5 = 4 \geq 0\\ f(\pi) &= \sin(2\pi) - e^\pi + 5\\ &= - e^\pi + 5 \leq 0 \end{aligned}

    By the intermediate value theorem there is therefore an x∗∈[0,π]x^* \in [0,\pi] such that f(x∗)=0f(x^*) = 0. So x∗x^* is the solution of the equation in the interval [0,π][0,\pi] that we are looking for.