Study Guides / Calculus I / Exercise session 13
EXERCISE SESSION 13 · ETH ZURICH · FS20 · calculus

Partial Fraction Decomposition, Improper Integrals and Taylor

Partial fraction decomposition, improper integrals, gamma function, Gaussian integral, Taylor formula

SB
Steven Battilana Quant · ex-ETH

Integration (I)

Definition 1.1 Struwe 6.1.1

A function F∈C1([a,b])F \in C^1([a,b]) is called an antiderivative of ff if:

F′(x)=f(x)∀x∈[a,b].F'(x) = f(x) \qquad \forall x \in [a,b].
Definition 1.2
  1. A partition of the compact interval I=[a,b]I=[a,b] with a<ba<b is a finite subset P⊂IP \subset I with P={a=x0,x1,...,xn=b}P = \{a=x_0, x_1, ..., x_n=b\} where x0<...<xnx_0 < ... < x_n. We denote the set of all partitions of II by

    P(I):={P⊂I    ∣    a,b∈P,  P is finite}.\mathcal{P}(I) := \{ P \subset I \;\; | \;\; a,b \in P, \; P \text{ is finite}\}.
  2. The mesh of the partition is defined by

    δ(P):=max⁡1≤i≤n{∣xi−xi−1∣}with n(P):=n=∣P∣−1.\delta (P) := \max_{1 \leq i \leq n} \{|x_i - x_{i-1}|\} \qquad \text{with } n(P):= n = |P|-1.

    Remark: δ(P)\delta (P) is the length of the largest interval Ii=[xi−1,xi]I_i = [x_{i-1},x_i] with 1≤i≤n1 \leq i \leq n.

  3. Let ξi∈Ii\xi_i \in I_i be intermediate points (sample points) with xi−1≤ξi≤xix_{i-1} \leq \xi_i \leq x_i. Every sum of the form

    S(f,P,ξ):=∑i=1nf(ξi)(xi−xi−1)S(f,P,\xi) := \sum_{i=1}^n f(\xi_i) (x_i-x_{i-1})

    is called a Riemann sum of the partition PP and ξ∈{ξ1,...,ξn}\xi \in \{\xi_1, ..., \xi_n\}.

Satz Struwe 6.2.1
f monotone⇒f is integrable.f \text{ monotone} \qquad \Rightarrow \qquad f \text{ is integrable}.
Satz Struwe 6.2.2
f continuous⇒f is integrable.f \text{ continuous} \qquad \Rightarrow \qquad f \text{ is integrable}.
Definition 1.3 upper sum
S‾(f,P):=∑i=1nsup⁡ξ∈[xi−1,xi]f(ξ)(xi−xi−1)\overline{S}(f,P) := \sum_{i=1}^n \sup_{\xi \in [x_{i-1}, x_i]} f(\xi) (x_i-x_{i-1})
Definition 1.4 lower sum
S‾(f,P):=∑i=1ninf⁡ξ∈[xi−1,xi]f(ξ)(xi−xi−1)\underline{S}(f,P) := \sum_{i=1}^n \inf_{\xi \in [x_{i-1}, x_i]} f(\xi) (x_i-x_{i-1})
Definition 1.5 upper Riemann integral
∫ab‾f(x)dx:=inf⁡{S‾(f,P)    ∣    P∈P(I)}.\overline{\int_a^b} f(x) dx := \inf \{ \overline{S}(f,P) \;\; | \;\; P \in \mathcal{P}(I) \}.
Definition 1.6 lower Riemann integral
∫ab‾f(x)dx:=sup⁡{S‾(f,P)    ∣    P∈P(I)}.\underline{\int_a^b} f(x) dx := \sup \{ \underline{S}(f,P) \;\; | \;\; P \in \mathcal{P}(I) \}.
Corollary 1.1 Struwe 6.2.3

Let f:[a,b]→Rf: [a,b] \to \mathbb{R} be a bounded and integrable function. Let {P(n)}\{P^{(n)}\} be a sequence of partitions of the compact interval [a,b][a,b] with δ(P(n))→n→∞0\delta (P^{(n)}) \xrightarrow{n \to \infty} 0 and {ξ(n)}\{ \xi^{(n)} \} a fixed choice of intermediate points for the partition P(n)P^{(n)}. Then

∫abf(x)dx=lim⁡n→∞S(f,P(n),ξ(n)).\int_a^b f(x) dx = \lim_{n \to \infty} S(f, P^{(n)}, \xi^{(n)}).
Remark

If corollary 6.2.3 is satisfied, ff is called Riemann integrable on [a,b][a,b].

Remark

Uniform partitions P(n)={a=x0,x1,...,xn}P^{(n)} = \{a=x_0, x_1, ..., x_n \} with xi=a+i(b−a)nx_i = a+ \frac{i(b-a)}{n} and xi−xi−1=b−anx_i - x_{i-1} = \frac{b-a}{n} form a sequence of partitions with δ(P(n))=b−an→n→∞0\delta(P^{(n)}) = \frac{b-a}{n} \xrightarrow{n \to \infty} 0.

Example 1.1

Compute the integral ∫01(x3−2x)dx\int\limits_0^1 (x^3 -2x) dx explicitly using Riemann sums.

Show solutionHide solution

We split the interval of integration [0,1][0,1] into nn subintervals of equal length Δx=1n\Delta x = \frac{1}{n} with xk=kn,  k=1,...,nx_k = \frac{k}{n}, \; k=1,...,n. The mesh of this partition is δ(P)=1n\delta (P) = \frac{1}{n}. We have lim⁡n→∞δ(P)=0\lim\limits_{n \to \infty} \delta (P) = 0. Hence the function f(x)=x3−2xf(x) = x^3 -2x is Riemann integrable and, by the definition of the Riemann integral (we choose ξk=xk\xi_k = x_k), we get:

∫01f(x)dx=∫01(x3−2x)dx=lim⁡n→∞∑k=1nf(ξk)(xk−xk−1)=lim⁡n→∞∑k=1nf(kn)1n=lim⁡n→∞1n∑k=1nf(kn)=lim⁡n→∞1n∑k=1n((kn)3−2⋅kn)=lim⁡n→∞1n∑k=1n(k3n3−2⋅kn)=(∗)lim⁡n→∞1n4∑k=1nk3−2n2∑k=1nk=lim⁡n→∞1n4(n(n+1)2)2−2n2n(n+1)2=lim⁡n→∞n2(n+1)24n4−n(n+1)n2=lim⁡n→∞(n+1)24n2−n(n+1)n2=lim⁡n→∞n2+2n+14n2−n2+nn2=14−1=−34.\begin{aligned} \int_0^1 f(x) dx = \int_0^1 (x^3 -2x) dx &= \lim_{n \to \infty} \sum_{k=1}^n f(\xi_k) (x_k-x_{k-1})\\ &= \lim_{n \to \infty} \sum_{k=1}^n f \left( \frac{k}{n} \right) \frac{1}{n}\\ &= \lim_{n \to \infty} \frac{1}{n} \sum_{k=1}^n f \left( \frac{k}{n} \right)\\ &= \lim_{n \to \infty} \frac{1}{n} \sum_{k=1}^n \left( \left(\frac{k}{n}\right)^3 -2 \cdot \frac{k}{n} \right)\\ &= \lim_{n \to \infty} \frac{1}{n} \sum_{k=1}^n \left( \frac{k^3}{n^3} -2 \cdot \frac{k}{n} \right)\\ &\overset{(*)}{=} \lim_{n \to \infty} \frac{1}{n^4} \sum_{k=1}^n k^3 -\frac{2}{n^2} \sum_{k=1}^n k\\ &= \lim_{n \to \infty} \frac{1}{n^4} \left( \frac{n(n+1)}{2} \right)^2 -\frac{2}{n^2} \frac{n(n+1)}{2}\\ &= \lim_{n \to \infty} \frac{n^2(n+1)^2}{4n^4} - \frac{n(n+1)}{n^2}\\ &= \lim_{n \to \infty} \frac{(n+1)^2}{4n^2} - \frac{n(n+1)}{n^2}\\ &= \lim_{n \to \infty} \frac{n^2+2n+1}{4n^2} - \frac{n^2+n}{n^2}\\ &= \frac{1}{4} - 1\\ &= -\frac{3}{4}. \end{aligned}

(∗)(*): We used the following formulas:

∑k=1nk3=(n(n+1)2)2∑k=1nk=n(n+1)2\sum_{k=1}^n k^3 = \left( \frac{n(n+1)}{2} \right)^2 \qquad\qquad \sum_{k=1}^n k = \frac{n(n+1)}{2}
Remark

The chain rule reads:

(f∘g)′(x)=f′(g(x))g′(x).(f \circ g)'(x) = f'(g(x)) g'(x).

Hence (f∘g)(x)=f(g(x))(f \circ g)(x) = f(g(x)) is an antiderivative of f′(g(x))g′(x)f'(g(x)) g'(x).

Example 1.2

Find the antiderivative ∫arctan⁡5(x)1+x2dx\int \frac{\arctan^5(x)}{1+x^2} dx.

Show solutionHide solution ∫arctan⁡5(x)1+x2dx=∫arctan⁡5(x)⏟f(g(x))⋅11+x2⏟g′(x)dx=16arctan⁡6(x)+C,∀C∈R.\begin{aligned} \int \frac{\arctan^5(x)}{1+x^2} dx &= \int \underbrace{\arctan^5(x)}_{f(g(x))} \cdot \underbrace{\frac{1}{1+x^2}}_{g'(x)} dx\\ &= \frac{1}{6} \arctan^6(x) + C, \qquad \forall C \in \mathbb{R}. \end{aligned}
Example 1.3

Find the antiderivative ∫3x+1x+1dx\int \frac{3^{\sqrt{x+1}}}{\sqrt{x+1}} dx.

Show solutionHide solution ∫3x+1x+1dx=∫elog⁡(3)x+1x+1dx=2∫elog⁡(3)x+12x+1dx=2∫elog⁡(3)x+1⏟f(g(x))⋅12x+1⏟g′(x)dx=2log⁡(3)⋅elog⁡(3)x+1+C,∀C∈R=2log⁡(3)⋅3x+1+C,∀C∈R.\begin{aligned} \int \frac{3^{\sqrt{x+1}}}{\sqrt{x+1}} dx &= \int \frac{e^{\log(3) \sqrt{x+1}}}{\sqrt{x+1}} dx\\ &= 2 \int \frac{e^{\log(3) \sqrt{x+1}}}{2\sqrt{x+1}} dx\\ &= 2 \int \underbrace{e^{\log(3) \sqrt{x+1}}}_{f(g(x))} \cdot \underbrace{\frac{1}{2\sqrt{x+1}}}_{g'(x)} dx\\ &= \frac{2}{\log(3)} \cdot e^{\log(3) \sqrt{x+1}} + C, \qquad \forall C \in \mathbb{R}\\ &= \frac{2}{\log(3)} \cdot 3^{\sqrt{x+1}} + C, \qquad \forall C \in \mathbb{R}. \end{aligned}
Satz fundamental theorem of calculus (version A)

Let f∈C0([a,b])f \in C^0([a,b]). Set F(x)=∫axf(t)dt,  x∈[a,b]F(x) = \int\limits_a^x f(t) dt, \; x \in [a,b].
Then F∈C1((a,b))F \in C^1((a,b)) with F′(x)=f(x)F'(x) = f(x).

Satz fundamental theorem of calculus (version B)

Let f∈C0([a,b])f \in C^0([a,b]) and FF an antiderivative of ff.
Then:

∫abf(x)dx=F(b)−F(a).\int_a^b f(x) dx = F(b) - F(a).
Example 1.4

Let A(x):=∫−1cos⁡(x)earccos⁡(t)dtA(x) := \int\limits_{-1}^{\cos(x)} e^{\arccos(t)} dt. Compute A′(x)A'(x).

Show solutionHide solution F(x):=∫−1xearccos⁡(t)dtΦ(x):=cos⁡(x)⇒A(x)=(F∘Φ)(x)⇒A′(x)=F′(Φ(x))⋅Φ′(x)=earccos⁡(cos⁡(x))⋅(−sin⁡(x))=−sin⁡(x)⋅exfor x∈[0,π]\begin{aligned} F(x) &:= \int_{-1}^{x} e^{\arccos(t)} dt\\ \Phi(x) &:= \cos(x)\\ \Rightarrow \quad A(x) &= (F \circ \Phi)(x)\\ \Rightarrow \quad A'(x) &= F'(\Phi(x)) \cdot \Phi'(x)\\ &= e^{\arccos(\cos(x))} \cdot (-\sin(x))\\ &= -\sin(x) \cdot e^x \qquad \text{for } x \in [0, \pi] \end{aligned}

Integration (II)

Satz monotonicity of the R-integral; Struwe 6.3.1

Let f,g:[a,b]→Rf,g: [a,b] \to \mathbb{R} be R-integrable with f≤gf \leq g. Then:

∫abf(x)dx≤∫abg(x)dx.\int_a^b f(x) dx \leq \int_a^b g(x) dx.
Satz linearity of the R-integral; Struwe 6.3.2

Let α,β∈R,  I=[a,b]\alpha, \beta \in \mathbb{R}, \; I=[a,b] and f,g:I→Rf,g: I \to \mathbb{R} two R-integrable functions. Then the function αf(x)+βg(x)\alpha f(x) + \beta g(x) is integrable over [a,b][a,b] and:

∫abαf(x)+βg(x)dx=α∫abf(x)dx+β∫abg(x)dx.\int_a^b \alpha f(x) + \beta g(x) dx = \alpha \int_a^b f(x) dx + \beta \int_a^b g(x) dx.
Corollary 2.1 standard estimate; Struwe 6.3.1

Let ff be integrable over [a,b][a,b]. Then ∣f∣|f| is R-integrable and:

∣∫abf(x)dx∣≤∫ab∣f(x)∣dx≤(sup⁡x∈[a,b]∣f(x)∣)(b−a).\left| \int_a^b f(x) dx \right| \leq \int_a^b |f(x)| dx \leq \left( \sup_{x \in [a,b]} |f(x)| \right) (b-a).
Satz additivity over intervals; Struwe 6.3.3

Let f:[a,b]→Rf: [a,b] \to \mathbb{R} be integrable over [a,b][a,b]. Let c∈[a,b]c \in [a,b]. Then f∣[a,c]f \Big|_{[a,c]} and f∣[c,b]f \Big|_{[c,b]} are integrable and:

∫abf(x)dx=∫acf(x)dx+∫cbf(x)dx.\int_a^b f(x) dx = \int_a^c f(x) dx + \int_c^b f(x) dx.
Remark
  • Convention:

    ∫abf(x)dx=−∫baf(x)dx\int_a^b f(x) dx = -\int_b^a f(x) dx
  • ∫aaf(x)dx=0\int_a^a f(x) dx = 0
Satz integration by parts
∫abf′(x)⋅g(x)dx=[f(x)g(x)]ab−∫abf(x)⋅g′(x)dx.\int_a^b f'(x) \cdot g(x) dx = \Big[ f(x) g(x) \Big]_a^b - \int_a^b f(x) \cdot g'(x) dx.

Derivation:

Show solutionHide solution uv=∫(uv)′dx=product rule∫(u′v+uv′)dx=linearity∫u′vdx+∫uv′dx⇔∫u′vdx=uv−∫uv′dx\begin{aligned} uv &= \int (uv)' dx\\ &\overset{\text{product rule}}{=} \int (u'v + uv') dx\\ &\overset{\text{linearity}}{=} \int u'v dx + \int uv' dx\\ \Leftrightarrow \quad \int u'v dx &= uv - \int uv' dx \end{aligned}
Example 2.1

Compute ∫log⁡2(x)dx\int \log^2(x) dx.

Show solutionHide solution ∫log⁡2(x)dx=∫log⁡2(x)⋅1  dx=IBPlog⁡2(x)⋅x−∫2log⁡(x)⋅1x⋅x  dx=log⁡2(x)⋅x−2∫log⁡(x)  dx=log⁡2(x)⋅x−2(∫log⁡(x)⋅1  dx)=IBPlog⁡2(x)⋅x−2(log⁡(x)⋅x−∫1x⋅x  dx)=log⁡2(x)⋅x−2(log⁡(x)⋅x−x)+C,C∈R=xlog⁡2(x)−2xlog⁡(x)+2x+C,C∈R\begin{aligned} \int \log^2(x) dx &= \int \log^2(x) \cdot 1 \;dx\\ &\overset{\text{IBP}}{=} \log^2(x) \cdot x - \int 2 \log(x) \cdot \frac{1}{x} \cdot x \;dx\\ &= \log^2(x) \cdot x - 2 \int \log(x) \;dx\\ &= \log^2(x) \cdot x - 2 \left( \int \log(x) \cdot 1 \;dx \right)\\ &\overset{\text{IBP}}{=} \log^2(x) \cdot x - 2 \left( \log(x) \cdot x - \int \frac{1}{x} \cdot x \; dx \right)\\ &= \log^2(x) \cdot x - 2 \left( \log(x) \cdot x -x\right) + C , \quad C \in \mathbb{R}\\ &= x\log^2(x) - 2x\log(x) + 2x + C, \quad C \in \mathbb{R} \end{aligned}
Example 2.2

Compute ∫e2xcos⁡(x)  dx\int e^{2x} \cos(x) \; dx.

Show solutionHide solution ∫e2xcos⁡(x)  dx=IBPe2xsin⁡(x)−2∫e2xsin⁡(x)  dx=IBPe2xsin⁡(x)−2(e2x(−cos⁡(x))−2∫e2x(−cos⁡(x))  dx)=e2xsin⁡(x)−2(e2x(−cos⁡(x))+2∫e2xcos⁡(x)  dx)=e2xsin⁡(x)+2e2xcos⁡(x)−4∫e2xcos⁡(x)  dx⇔5∫e2xcos⁡(x)  dx=e2xsin⁡(x)+2e2xcos⁡(x)+C~,C~∈R⇔∫e2xcos⁡(x)  dx=15e2xsin⁡(x)+25e2xcos⁡(x)+C,C∈R\begin{aligned} \int e^{2x} \cos(x) \; dx &\overset{\text{IBP}}{=} e^{2x} \sin(x) -2 \int e^{2x} \sin(x) \; dx\\ &\overset{\text{IBP}}{=} e^{2x} \sin(x) -2 \left( e^{2x} (-\cos(x)) - 2 \int e^{2x} (-\cos(x)) \; dx \right)\\ &= e^{2x} \sin(x) -2 \left( e^{2x} (-\cos(x)) + 2 \int e^{2x} \cos(x) \; dx \right)\\ &= e^{2x} \sin(x) +2 e^{2x} \cos(x) - 4 \int e^{2x} \cos(x) \; dx\\ \Leftrightarrow \quad 5 \int e^{2x} \cos(x) \; dx &= e^{2x} \sin(x) +2 e^{2x} \cos(x) + \widetilde{C}, \quad \widetilde{C} \in \mathbb{R}\\ \Leftrightarrow \quad \int e^{2x} \cos(x) \; dx &= \frac{1}{5} e^{2x} \sin(x) +\frac{2}{5} e^{2x} \cos(x) + C, \quad C \in \mathbb{R} \end{aligned}
Example 2.3

Compute ∫01xsinh⁡(4x)  dx\int\limits_0^1 x \sinh(4x) \; dx.

Show solutionHide solution ∫01xsinh⁡(4x)  dx=[x⋅cosh⁡(4x)4]01−∫011⋅cosh⁡(4x)4  dx=[x⋅cosh⁡(4x)4]01−14∫01cosh⁡(4x)  dx=[1⋅cosh⁡(4)4−0⋅cosh⁡(0)4]−14[sinh⁡(4x)4]01=cosh⁡(4)4−14[sinh⁡(4)4−sinh⁡(0)4]=sinh⁡(0)=0cosh⁡(4)4−14[sinh⁡(4)4−0]=cosh⁡(4)4−sinh⁡(4)16\begin{aligned} \int_0^1 x \sinh(4x) \; dx &= \left[ x \cdot \frac{\cosh(4x)}{4} \right]_0^1 - \int_0^1 1 \cdot \frac{\cosh(4x)}{4} \; dx\\ &= \left[ x \cdot \frac{\cosh(4x)}{4} \right]_0^1 - \frac{1}{4} \int_0^1 \cosh(4x) \; dx\\ &= \left[ 1 \cdot \frac{\cosh(4)}{4} - 0 \cdot \frac{\cosh(0)}{4}\right] - \frac{1}{4} \left[ \frac{\sinh(4x)}{4} \right]_0^1\\ &= \frac{\cosh(4)}{4} - \frac{1}{4} \left[ \frac{\sinh(4)}{4} - \frac{\sinh(0)}{4} \right]\\ &\overset{\sinh(0)=0}{=} \frac{\cosh(4)}{4} - \frac{1}{4} \left[ \frac{\sinh(4)}{4} - 0 \right]\\ &= \frac{\cosh(4)}{4} - \frac{\sinh(4)}{16} \end{aligned}
Remark

What about ∫1x  dx\int \frac{1}{x} \; dx?

Show solutionHide solution f1:(0,∞)→R,    x↦1x⇒∫f1(x)  dx=log⁡(x)+C,C∈Rf2:R∖{0}→R,    x↦1x⇒∫f2(x)  dx=log⁡∣x∣+C,C∈R\begin{aligned} f_1: (0, \infty) \to \mathbb{R}, \;\; x \mapsto \frac{1}{x} \quad \Rightarrow \quad \int f_1(x)\; dx = \log(x) + C, \quad C \in \mathbb{R}\\ f_2: \mathbb{R} \setminus \{0\} \to \mathbb{R}, \;\; x \mapsto \frac{1}{x} \quad \Rightarrow \quad \int f_2(x)\; dx = \log|x| + C, \quad C \in \mathbb{R} \end{aligned}

Claim: ∫f2(x)dx=∫1x=log⁡∣x∣+C,C∈R⇒(log⁡∣x∣+C)′=1x\int f_2(x) dx = \int \frac{1}{x} = \log|x| + C, \quad C \in \mathbb{R} \quad \Rightarrow \quad (\log|x| + C)' = \frac{1}{x}.

Proof:

(log⁡∣x∣+C)′=log⁡′∣x∣(∣x∣)′=1∣x∣⋅(∣x∣)′={1x⋅1,x>0−1x⋅−1,x<0=1x\begin{aligned} (\log|x| + C)' &= \log'|x| (|x|)'\\ &= \frac{1}{|x|} \cdot (|x|)'\\ &= \begin{cases} \frac{1}{x} \cdot 1, \quad x > 0\\ \frac{-1}{x} \cdot -1, \quad x < 0 \end{cases}\\ &= \frac{1}{x} \end{aligned}
Satz substitution; Struwe 6.1.5

Let f:[a,b]→Rf: [a,b] \to \mathbb{R} be continuous, FF an antiderivative of ff, g:[α,β]→Rg: [\alpha, \beta] \to \mathbb{R} of class C1([α,β])C^1([\alpha, \beta]), and t0≤t1t_0 \leq t_1 in [α,β][\alpha, \beta] such that g([t0,t1])⊂[a,b]g([t_0,t_1]) \subset [a,b]. Then:

∫t0t1F′(g(t))g′(t)dt=∫t0t1f(g(t))g′(t)dt=∫g(t0)g(t1)f(x)dx.\int_{t_0}^{t_1} F'(g(t)) g'(t) dt = \int_{t_0}^{t_1} f(g(t)) g'(t) dt = \int_{g(t_0)}^{g(t_1)} f(x) dx.
Satz substitution; Burger 5.4.6

Let a<b, ϕ:[a,b]→Ra < b,\, \phi: [a,b] \to \mathbb{R} be continuously differentiable, I⊂RI \subset \mathbb{R} an interval with ϕ([a,b])⊂I\phi([a,b]) \subset I and f:I→Rf: I \to \mathbb{R} a continuous function. Then:

∫ϕ(a)ϕ(b)f(x)dx=∫abf(ϕ(t))ϕ′(t)dt.\int_{\phi(a)}^{\phi(b)} f(x) dx = \int_a^b f(\phi(t)) \phi'(t) dt.
Example 2.4

Compute ∫0π24cos⁡(x)  dx\int\limits_0^{\frac{\pi^2}{4}} \cos(\sqrt{x})\; dx.

Show solutionHide solution

We choose the following substitution:

u=xu = \sqrt{x}

Following the substitution rule, we change the variable in the integrand; for that we also need the following:

g(x):=u=x⇒ddx(u)=12x⇔du=12x  dx⇔2x  du=dx⇔2u  du=dx\begin{aligned} g(x) := u &= \sqrt{x}\\ \Rightarrow \quad \frac{d}{dx} (u) &= \frac{1}{2 \sqrt{x}}\\ \Leftrightarrow \quad du &= \frac{1}{2 \sqrt{x}} \;dx\\ \Leftrightarrow \quad 2 \sqrt{x} \;du &= dx\\ \Leftrightarrow \quad 2u \;du &= dx \end{aligned}

Now we can substitute:

∫0π24cos⁡(x)  dx=∫g(0)g(π24)2ucos⁡(u)  du=∫0π242ucos⁡(u)  du=∫0π22ucos⁡(u)  du\begin{aligned} \int_0^{\frac{\pi^2}{4}} \cos(\sqrt{x})\; dx &= \int_{g(0)}^{g\left(\frac{\pi^2}{4}\right)} 2u\cos(u) \;du\\ &= \int_{0}^{\sqrt{\frac{\pi^2}{4}}} 2u \cos(u) \;du\\ &= \int_{0}^{\frac{\pi}{2}} 2u \cos(u) \;du \end{aligned}

Now we can solve the integral using integration by parts:

2∫0π2ucos⁡(u)  du=IBP2([usin⁡(u)]0π2−∫0π2sin⁡(u)  du)=2(π2sin⁡(π2)−(−cos⁡(u))∣0π2)=2(π2⋅1+cos⁡(u)∣0π2)=2(π2+cos⁡(π2)−cos⁡(0))=2(π2+0−1)=2⋅π2−2=π−2\begin{aligned} 2 \int_{0}^{\frac{\pi}{2}} u \cos(u)\;du &\overset{\text{IBP}}{=} 2 \left( \Big[ u \sin(u) \Big]_0^{\frac{\pi}{2}} - \int_0^{\frac{\pi}{2}} \sin(u) \; du \right)\\ &= 2 \left( \frac{\pi}{2} \sin(\frac{\pi}{2}) - (-\cos(u)) \Big|_0^{\frac{\pi}{2}} \right)\\ &= 2 \left( \frac{\pi}{2} \cdot 1 + \cos(u) \Big|_0^{\frac{\pi}{2}} \right)\\ &= 2 \left( \frac{\pi}{2} + \cos(\frac{\pi}{2}) - \cos(0) \right)\\ &= 2 \left( \frac{\pi}{2} + 0 - 1 \right)\\ &= 2 \cdot \frac{\pi}{2} - 2\\ &= \pi - 2 \end{aligned}
Remark substitution

The following list shows which substitutions can be useful for which functions:

  1. ex,sinh⁡(x),cosh⁡(x)e^x, \sinh(x), \cosh(x)
    Substitution: t=ex,    dx=dtt,    sinh⁡(x)=t2−12t,    cosh⁡(x)=t2+12tt = e^{x},\;\; dx = \frac{dt}{t},\;\; \sinh(x)= \frac{t^2-1}{2t},\;\; \cosh(x) = \frac{t^2+1}{2t}

  2. log⁡(x)\log(x)
    Substitution: t=log⁡(x),x=et,    dx=etdtt=\log(x), x=e^t, \;\;dx = e^t dt

  3. ax+b\sqrt{ax+b}, substitute the root (preferably in the denominator)
    Substitution: x→t=x,    x+1→t=x+1\sqrt{x} \to t = \sqrt{x}, \;\; \sqrt{x+1} \to t = \sqrt{x+1}; xb→t=xb\qquad \sqrt[b]{x} \to t=\sqrt[b]{x}

  4. cos⁡2(x),    sin⁡2(x),...,    tan⁡(x)\cos^2(x), \;\;\sin^2(x), ..., \;\;\tan(x)
    Substitution: t=tan⁡(x),    dx=11+t2dt,    sin⁡2(x)=t21+t2,    cos⁡2(x)=11+t2t=\tan(x), \;\;dx = \frac{1}{1+t^2} dt, \;\;\sin^2(x) = \frac{t^2}{1+t^2}, \;\;\cos^2(x) = \frac{1}{1+t^2}

  5. cos⁡(x),    sin⁡(x),    cos⁡3(x),\cos(x), \;\; \sin(x), \;\; \cos^3(x), …
    Substitution: t=tan⁡(x2),    dx=21+t2dt,    sin⁡(x)=2t1+t2,    cos⁡(x)=1−t21+t2t=\tan(\frac{x}{2}),\;\; dx = \frac{2}{1+t^2} dt, \;\; \sin(x) = \frac{2t}{1+t^2}, \;\; \cos(x) = \frac{1-t^2}{1+t^2}

  6. x2+bx+c\sqrt{x^2+bx+c} in the numerator, use sin⁡2(x)+cos⁡2(x)=1\sin^2(x)+\cos^2(x)=1 or cosh⁡2(x)−sinh⁡2(x)=1\cosh^2(x)-\sinh^2(x) = 1
    \qquad ∫1−x2dx\int \sqrt{1-x^2} dx: substitute x=sin⁡(t)x = \sin(t) or x=cos⁡(t)x = \cos(t)
    \qquad ∫x2−1dx\int \sqrt{x^2-1} dx: substitute x=cosh⁡(t)x = \cosh(t)
    \qquad ∫x2+1dx\int \sqrt{x^2+1} dx: substitute x=sinh⁡(t)x = \sinh(t)

  7. a2+b2x2\sqrt{a^2 + b^2x^2}
    Substitution: x=ab⋅tan⁡(t),    dx=abcos⁡2(t)⋅dtx = \frac{a}{b} \cdot \tan(t), \;\; dx = \frac{a}{b \cos^2(t)} \cdot dt or x=ab⋅sinh⁡(t),    dx=ab⋅cosh⁡(t)⋅dtx = \frac{a}{b} \cdot \sinh(t), \;\; dx = \frac{a}{b} \cdot \cosh(t) \cdot dt

  8. b2x2−a2\sqrt{b^2x^2-a^2}
    Substitution: x=abcos⁡(t)x = \frac{a}{b \cos(t)},     dx=ab⋅sin⁡(t)cos⁡2(t)⋅dt\;\; dx = \frac{a}{b} \cdot \frac{\sin(t)}{\cos^2(t)} \cdot dt or x=ab⋅cosh⁡(t),    dx=ab⋅sinh⁡(t)dtx = \frac{a}{b} \cdot \cosh(t), \;\; dx=\frac{a}{b} \cdot \sinh(t) dt

Partial Fraction Decomposition

Remark

When we add two fractions, we have to find a common denominator:

xx2+1+1x+1=x⋅(x+1)+(x2+1)(x2+1)(x+1)=2x2+x+1(x2+1)(x+1)\frac{x}{x^2+1} + \frac{1}{x+1} = \frac{x \cdot (x+1) + (x^2+1)}{(x^2+1)(x+1)} = \frac{2x^2+x+1}{(x^2+1)(x+1)}

Now we want to reverse the direction and go from right to left, but how does that work? With partial fraction decomposition! (compare with Satz 6.1.6 in the lecture notes)

Recipe for partial fraction decomposition (PFD)
Given: a rational function f(x)=p(x)q(x)f(x) = \frac{p(x)}{q(x)}
Wanted: partial fraction decomposition of f(x)f(x)

  1. If deg⁡(p)≥deg⁡(q)\deg(p) \geq \deg(q), carry out polynomial long division to get the following form:

    f(x)=p(x)q(x)=s(x)+p~(x)q(x)⏟remainderf(x) = \frac{p(x)}{q(x)} = s(x) + \underbrace{\frac{\widetilde{p}(x)}{q(x)}}_{\text{remainder}}

    If deg⁡(p)<deg⁡(q)\deg(p) < \deg(q), set p(x)=p~(x)p(x) = \widetilde{p}(x).

  2. Find the roots of q(x)q(x) and factorise q(x)q(x).
    Remark. If you do not see the roots directly, you can use the quadratic formula here, or a trick: guess a root (try 1,−1,2,−21,-1,2,-2, etc.) and then carry out a polynomial division.

  3. Use the ansatz for partial fraction decomposition, with the factorised q(x)q(x) from step (ii):

    q(x)=(x−x0)(x−x1)k(x2+a1x+b1)f(x)=p(x)q(x)=A0(x−x0)+...+An(x−xn)+...⏟simple roots+A1(x−x1)+...+Ak(x−x1)k+B1(x−x1)+...⏟multiple roots+A1x+B1(x2+a1x+b1)+...+Akx+Bk(x2+a1x+b1)k+...⏟complex, multiple roots\begin{aligned} q(x) &= (x-x_0)(x-x_1)^k(x^2+a_1x+b_1)\\[1.5ex] f(x) = \frac{p(x)}{q(x)} &= \underbrace{\frac{A_0}{(x-x_0)} + ... + \frac{A_n}{(x-x_n)} + ...}_{\text{simple roots}}\\[1ex] &\quad + \underbrace{\frac{A_1}{(x-x_1)} + ... + \frac{A_k}{(x-x_1)^k} + \frac{B_1}{(x-x_1)} + ...}_{\text{multiple roots}}\\[1ex] &\quad + \underbrace{\frac{A_1x+B_1}{(x^2+a_1x+b_1)} + ... + \frac{A_kx+B_k}{(x^2+a_1x+b_1)^k} + ...}_{\text{complex, multiple roots}} \end{aligned}
  4. Bring the ansatz to a common denominator, i.e. the denominator is q(x)q(x), and compare coefficients (#unknowns = deg⁡(q)\deg(q)).
Example 3.1 partial fraction decomposition "light" from the 4th PDF

Decompose f(x)=1(x+2)(x−1)f(x) = \frac{1}{(x+2)(x-1)}.

Show solutionHide solution 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)} &\overset{!}{=} \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 &\overset{!}{=} (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\\ &\overset{A = -B}{\Longrightarrow} 2B-(-B) = 1\\ &\Leftrightarrow 3B = 1\\ &\Leftrightarrow B = \frac{1}{3}\\ &\overset{(*)\; \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}
Example 3.2

Decompose: f(x)=−21x4−14x3+34x2+22x−1−3x3−2x2+3x+2f(x) = \frac{-21x^4-14x^3+34x^2+22x-1}{-3x^3 -2x^2 + 3x +2}.

Show solutionHide solution
  1. deg⁡(p)>deg⁡(q)\deg(p) > \deg(q) with:

    p(x)=−21x4−14x3+34x2+22x−1q(x)=−3x3−2x2+3x+2\begin{aligned} p(x) &= -21x^4-14x^3+34x^2+22x-1\\ q(x) &= -3x^3 -2x^2 + 3x +2 \end{aligned}

    So we have to carry out a polynomial division:

    −21x4−14x3+34x2+22x−1:−3x3−2x2+3x+2=7x+13x2+8x−1−3x3−2x2+3x+2−21x4+14x3−21x2−14x‾−21x4−14x3+13x2+8x−1\begin{array}{l} -21x^4-14x^3+34x^2+22x-1 : -3x^3-2x^2+3x+2 = 7x + \dfrac{13x^2+8x-1}{-3x^3-2x^2+3x+2}\\ \underline{\phantom{-}21x^4+14x^3-21x^2-14x}\\ \phantom{-21x^4-14x^3+{}} 13x^2+8x-1 \end{array}

    We define p~(x)=13x2+8x−1\widetilde{p}(x) = 13x^2 + 8x -1.

  2. Find the roots of q(x)q(x) and factorise:

    f(x)=s(x)+p~(x)q(x)=7x+13x2+8x−1−(x−1)(x+1)(3x+2)⏟remainderf(x) = s(x) + \frac{\widetilde{p}(x)}{q(x)} = 7x+ \underbrace{\frac{13x^2 + 8x -1}{-(x-1)(x+1)(3x+2)}}_{\text{remainder}}
  3. Use the ansatz for the remainder term:

    Ax−1+Bx+1+C3x+2\frac{A}{x-1} + \frac{B}{x+1} + \frac{C}{3x+2}
  4. Bring the ansatz to a common denominator:

    Ax−1+Bx+1+C3x+2=A(x+1)(3x+2)+B(x−1)(3x+2)+C(x−1)(x+1)(x−1)(x+1)(3x+2)=(3A+3B+C)x2+(5A−B)x+(2A−2B−C)(x−1)(x+1)(3x+2)\begin{aligned} \frac{A}{x-1} + \frac{B}{x+1} + \frac{C}{3x+2} &= \frac{A(x+1)(3x+2) + B(x-1)(3x+2) + C(x-1)(x+1)}{(x-1)(x+1)(3x+2)}\\ &= \frac{(3A+3B+C)x^2 + (5A-B)x + (2A-2B-C)}{(x-1)(x+1)(3x+2)} \end{aligned}

    Comparing coefficients:

    13x2+8x−1−3x3−2x2+3x+2=−13x2+8x−1(x−1)(x+1)(3x+2)=!(3A+3B+C)x2+(5A−B)x+(2A−2B−C)(x−1)(x+1)(3x+2)\begin{aligned} \frac{13x^2 + 8x -1}{-3x^3 -2x^2 + 3x +2} &= -\frac{13x^2 + 8x -1}{(x-1)(x+1)(3x+2)}\\ &\overset{!}{=} \frac{(3A+3B+C)x^2 + (5A-B)x + (2A-2B-C)}{(x-1)(x+1)(3x+2)} \end{aligned}power of xansatzgiven numerator polynomialx2:3A+3B+C=−13x1:5A−B=−8x0:2A−2B−C=+1\begin{array}{lll} \text{power of } x & \text{ansatz} & \text{given numerator polynomial}\\ x^2: & 3A+3B+C & =-13\\ x^1: & 5A-B & =-8\\ x^0: & 2A-2B-C & =+1 \end{array}

    Solve this system of equations; you can use the methods from linear algebra here. This gives you the following solution:

    A=−2B=−2C=−1\begin{aligned} A &= -2\\ B &= -2\\ C &= -1 \end{aligned}

    So in the end we get the following partial fraction decomposition:

    −21x4−14x3+34x2+22x−1−3x3−2x2+3x+2=7x+13x2+8x−1−3x3−2x2+3x+2=7x−2x−1−2x+1−13x+2\begin{aligned} \frac{-21x^4-14x^3+34x^2+22x-1}{-3x^3 -2x^2 + 3x +2} &= 7x + \frac{13x^2 + 8x -1}{-3x^3 -2x^2 + 3x +2}\\ &= 7x -\frac{2}{x-1} - \frac{2}{x+1} - \frac{1}{3x+2} \end{aligned}

So the overall solution is:

f(x)=7x−2x−1−2x+1−13x+2.f(x) = 7x -\frac{2}{x-1} - \frac{2}{x+1} - \frac{1}{3x+2}.

Improper Integrals

Definition 4.1 gamma function

For real α>0\alpha>0, the gamma function is defined by:

Γ(α)=∫0∞tα−1e−tdt.\Gamma(\alpha) = \int_0^\infty t^{\alpha-1} e^{-t} dt.
Definition 4.2 Gaussian integral
∫0∞1te−t  dt=Γ(12)=πor∫0∞e−x2  dx=π2.\int_0^\infty \frac{1}{\sqrt{t}} e^{-t} \; dt = \Gamma\left(\tfrac{1}{2}\right) = \sqrt{\pi} \qquad \text{or} \qquad \int_0^\infty e^{-x^2} \; dx = \frac{\sqrt{\pi}}{2}.
Definition 4.3 Struwe 6.4.1

Let f:(a,b)→Rf: (a,b) \to \mathbb{R} be Riemann integrable over every compact interval [c,d]⊂(a,b)[c,d] \subset (a,b). ff is called improperly Riemann integrable over (a,b) if

∫abf  dx:=lim⁡c→a+,  d→b−∫cdf  dx\int_a^b f \; dx := \lim_{c \to a^+, \; d \to b^-} \int_c^d f \; dx

exists.

Definition 4.4 Michaels: improper integrals of type a
  1. Let f(x)f(x) be continuous on [a,∞)[a, \infty). Then we set

    ∫a∞f(x)  dx:=lim⁡R→∞∫aRf(x)  dx,\int_a^\infty f(x) \; dx := \lim_{R \to \infty} \int_a^R f(x) \; dx,

    if the limit exists. This value is called an improper integral.

  2. Let f(x)f(x) be continuous on (−∞,b](-\infty, b]. Then we set

    ∫−∞bf(x)  dx:=lim⁡R→−∞∫Rbf(x)  dx,\int_{-\infty}^b f(x) \; dx := \lim_{R \to -\infty} \int_R^b f(x) \; dx,

    if the limit exists. This value is called an improper integral.

Definition 4.5 Michaels: improper integrals of type b
  1. Let f(x)f(x) be continuous on (a,b](a, b]. Then we set

    ∫abf(x)  dx:=lim⁡ε→0+∫a+εbf(x)  dx,\int_a^b f(x) \; dx := \lim_{\varepsilon \to 0^+} \int_{a+\varepsilon}^b f(x) \; dx,

    if the limit exists. This value is called an improper integral.

  2. Let f(x)f(x) be continuous on [a,b)[a, b). Then we set

    ∫abf(x)  dx:=lim⁡ε→0+∫ab−εf(x)  dx,\int_a^b f(x) \; dx := \lim_{\varepsilon \to 0^+} \int_a^{b-\varepsilon} f(x) \; dx,

    if the limit exists. This value is called an improper integral.

Satz Struwe 6.4.1

Let f:[1,∞)→R+f: [1,\infty) \to \mathbb{R}_+ be monotonically decreasing. Then the series ∑k=1∞f(k)\sum\limits_{k=1}^\infty f(k) converges if and only if ∫1∞f  dx\int\limits_1^\infty f \; dx converges, and in that case:

0≤∑k=1∞f(k)−∫1∞f  dx≤f(1).0 \leq \sum_{k=1}^\infty f(k) - \int_1^\infty f \; dx \leq f(1).

Convergence Tests for Improper Integrals

Remark criterion 1: direct computation & definition

Some improper integrals have the property that the (indefinite) integral can be computed explicitly. In these situations you show the convergence of the integral using definition 6.4.1.

Example 4.1

Show the convergence of ∫1∞1xpdx\int\limits_1^\infty \frac{1}{x^p} dx.

Show solutionHide solution
  • If p=1p=1:

    ∫1∞1xdx=Def.lim⁡R→∞∫1R1xdx=lim⁡R→∞[log⁡∣x∣]1R=lim⁡R→∞log⁡∣R∣−log⁡∣1∣=lim⁡R→∞log⁡∣R∣=∞\begin{aligned} \int_1^\infty \frac{1}{x} dx &\overset{\text{Def.}}{=} \lim_{R \to \infty} \int_1^R \frac{1}{x} dx\\ &= \lim_{R \to \infty} [\log|x|]_1^R\\ &= \lim_{R \to \infty} \log|R| - \log|1|\\ &= \lim_{R \to \infty} \log|R|\\ &= \infty \end{aligned}

    So the integral diverges.

  • If p>1p>1:

    ∫1∞1xpdx=Def.lim⁡R→∞∫1R1xpdx=lim⁡R→∞[−1(p−1)xp−1]1R=lim⁡R→∞−1(p−1)Rp−1−−1(p−1)1p−1=lim⁡R→∞−1(p−1)Rp−1+1p−1=lim⁡R→∞1p−1(−1Rp−1+1)=p−1>01p−1(−0+1)=1p−1<∞\begin{aligned} \int_1^\infty \frac{1}{x^p} dx &\overset{\text{Def.}}{=} \lim_{R \to \infty} \int_1^R \frac{1}{x^p} dx\\ &= \lim_{R \to \infty} \left[ \frac{-1}{(p-1)x^{p-1}} \right]_1^R\\ &= \lim_{R \to \infty} \frac{-1}{(p-1)R^{p-1}} - \frac{-1}{(p-1)1^{p-1}}\\ &= \lim_{R \to \infty} \frac{-1}{(p-1)R^{p-1}} + \frac{1}{p-1}\\ &= \lim_{R \to \infty} \frac{1}{p-1} \left( -\frac{1}{R^{p-1}} + 1 \right)\\ &\overset{p-1>0}{=} \frac{1}{p-1} (-0 + 1)\\ &= \frac{1}{p-1} < \infty \end{aligned}

    So the integral converges.

  • If p<1p<1:

    ∫1∞1xpdx=Def.lim⁡R→∞∫1R1xpdx=...=lim⁡R→∞1p−1(−1Rp−1+1)=p−1<0"1p−1⏟<0(−∞+1)"=∞\begin{aligned} \int_1^\infty \frac{1}{x^p} dx &\overset{\text{Def.}}{=} \lim_{R \to \infty} \int_1^R \frac{1}{x^p} dx\\ &=...= \lim_{R \to \infty} \frac{1}{p-1} \left( -\frac{1}{R^{p-1}} + 1 \right)\\ &\overset{p-1<0}{=} \text{"}\underbrace{\frac{1}{p-1}}_{<0} (-\infty + 1)\text{"}\\ &= \infty \end{aligned}

    So the integral diverges.

In summary, we have derived the following important rule of thumb:

∫1∞1xp  dx converges⇔p>1.\int_1^\infty \frac{1}{x^p} \; dx \text{ converges} \quad \Leftrightarrow \quad p>1.
Satz criterion 2a: comparison test

Let f,gf,g be continuous on [a,∞)[a,\infty) with

0≤f(x)≤g(x)∀x∈[a,∞)0 \leq f(x) \leq g(x) \quad \forall x \in [a, \infty)
  1. If ∫a∞g(x)  dx\int\limits_a^\infty g(x) \; dx converges, so does ∫a∞f(x)  dx\int\limits_a^\infty f(x) \; dx,
  2. If ∫a∞f(x)  dx\int\limits_a^\infty f(x) \; dx diverges, so does ∫a∞g(x)  dx\int\limits_a^\infty g(x) \; dx.
Satz criterion 2b: comparison test

Let f,gf,g be continuous on (a,b](a,b] with

0≤f(x)≤g(x)∀x∈(a,b]0 \leq f(x) \leq g(x) \quad \forall x \in (a, b]
  1. If ∫abg(x)  dx\int\limits_a^b g(x) \; dx converges, so does ∫abf(x)  dx\int\limits_a^b f(x) \; dx,
  2. If ∫abf(x)  dx\int\limits_a^b f(x) \; dx diverges, so does ∫abg(x)  dx\int\limits_a^b g(x) \; dx.
Satz criterion 3a: absolute convergence
∫a∞∣f(x)∣  dx<∞⇒∫a∞f(x)  dx<∞.\int_a^\infty |f(x)| \; dx < \infty \quad \Rightarrow \quad \int_a^\infty f(x) \; dx < \infty.
Satz criterion 3b: absolute convergence
∫ab∣f(x)∣  dx<∞⇒∫abf(x)  dx<∞.\int_a^b |f(x)| \; dx < \infty \quad \Rightarrow \quad \int_a^b f(x) \; dx < \infty.
Satz criterion 4a: limit test

Test 1

  1. f(x),g(x)f(x), g(x) continuous on [a,∞)[a, \infty)
  2. lim⁡x→∞f(x)g(x)=A≠∞\lim\limits_{x \to \infty} \frac{f(x)}{g(x)} = A \neq \infty

then ∫a∞∣g(x)∣  dx<∞\int\limits_a^\infty |g(x)| \; dx < \infty converges ⇔∫a∞∣f(x)∣  dx\quad \Leftrightarrow \quad \int_a^\infty |f(x)| \; dx converges.

Test 2

  1. f(x)f(x) continuous on [a,∞)[a, \infty)
  2. lim⁡x→∞xpf(x)=A≠∞\lim\limits_{x \to \infty} x^p f(x) = A \neq \infty for some p>1p>1.

then ∫a∞∣f(x)∣  dx\int\limits_a^\infty |f(x)| \; dx converges, and hence so does ∫a∞f(x)  dx\int\limits_a^\infty f(x) \; dx (absolute convergence)

Test 3

  1. f(x)f(x) continuous on [a,∞)[a, \infty)
  2. lim⁡x→∞xf(x)=A≠0,∞\lim\limits_{x \to \infty} x f(x) = A \neq 0,\infty.

then ∫a∞f(x)  dx\int\limits_a^\infty f(x) \; dx diverges. (Note: test 3 fails if A=0A=0.)

Satz criterion 4b: limit test

Test 1

  1. f(x),g(x)f(x), g(x) continuous on (a,b](a, b]
  2. lim⁡x→a+f(x)g(x)=A≠∞\lim\limits_{x \to a^+} \frac{f(x)}{g(x)} = A \neq \infty

then ∫ab∣g(x)∣  dx<∞\int\limits_a^b |g(x)| \; dx < \infty converges ⇔∫ab∣f(x)∣  dx\quad \Leftrightarrow \quad \int\limits_a^b |f(x)| \; dx converges.

Test 2

  1. f(x)f(x) continuous on (a,b](a, b]
  2. lim⁡x→a+(x−a)pf(x)=A≠∞\lim\limits_{x \to a^+} (x-a)^p f(x) = A \neq \infty for some 0<p<10<p<1.

then ∫ab∣f(x)∣  dx\int\limits_a^b |f(x)| \; dx converges, and hence so does ∫abf(x)  dx\int\limits_a^b f(x) \; dx (absolute convergence)

Test 3

  1. f(x)f(x) continuous on (a,b](a, b]
  2. lim⁡x→a+(x−a)f(x)=A≠0,∞\lim\limits_{x \to a^+} (x-a) f(x) = A \neq 0,\infty.

then ∫abf(x)  dx\int\limits_a^b f(x) \; dx diverges. (Note: test 3 fails if A=0A=0.)

Satz criterion 5a: Leibniz criterion for improper integrals

Let f(x)f(x) be a continuous function on [a,∞)[a,\infty). If the function f(x)f(x) is monotonically decreasing and lim⁡x→∞f(x)=0\lim\limits_{x \to \infty} f(x) = 0, then the improper integrals

∫a∞f(x)sin⁡(x)  dxand∫a∞f(x)cos⁡(x)  dx\int_a^\infty f(x) \sin(x) \; dx \quad \text{and} \quad \int_a^\infty f(x) \cos(x) \; dx

converge.

Satz criterion 5b: Leibniz criterion for improper integrals

Let f(x)f(x) be a continuous function on (a,b](a,b]. If the function f(x)(x−a)2f(x)(x-a)^2 is monotonically increasing and lim⁡x→a+f(x)(x−a)2=0\lim\limits_{x \to a^+} f(x)(x-a)^2 = 0, then the improper integrals

∫abf(x)sin⁡(1x−a)  dxand∫abf(x)cos⁡(1x−a)  dx\int_a^b f(x) \sin(\frac{1}{x-a}) \; dx \quad \text{and} \quad \int_a^b f(x) \cos(\frac{1}{x-a}) \; dx

converge.

Satz criterion 6b: bounded function on a bounded interval

Let ff be continuous on (a,b](a,b]. If

lim⁡x→a+f(x)=A≠∞,\lim_{x \to a^+} f(x) = A \neq \infty,

then the integral ∫abf(x)dx\int\limits_a^b f(x) dx converges.

The Taylor Polynomial and Taylor’s Formula

Satz Taylor formula; Struwe 5.5.1

Let f:[a,b]→Rf: [a,b] \to \mathbb{R} be a Cm([a,b])C^{m}([a,b]) function (i.e. mm times differentiable on (a,b)(a,b)). Then:

f(x)=Tmf(x;x0)+Rmf(x;x0)f(x) = T_mf(x;x_0) + R_mf(x;x_0)

with the Taylor polynomial of degree mm at the expansion point x0x_0:

Tmf(x;x0):=∑k=0mf(k)(x0)(x−x0)kk!T_mf(x;x_0) := \sum_{k=0}^m f^{(k)}(x_0) \frac{(x-x_0)^k}{k!}

and for the remainder Rmf(x;x0):=f(x)−Tmf(x;x0)R_mf(x;x_0) := f(x) - T_mf(x;x_0) we have:

lim⁡x→x0Rmf(x;x0)(x−x0)m=0.\lim_{x \to x_0} \frac{R_mf(x;x_0)}{(x-x_0)^m} = 0.

If ff is m+1m+1 times differentiable, the Taylor formula holds:

f(x)=Tmf(x;x0)+f(m+1)(ξ)(x−x0)m+1(m+1)!for some ξ∈(x0,x)f(x) = T_mf(x;x_0) + f^{(m+1)}(\xi) \frac{(x-x_0)^{m+1}}{(m+1)!} \quad\quad \text{for some } \xi \in (x_0,x)

If f∈Cm+1[a,b]f \in C^{m+1}[a,b], then the estimate for the remainder holds:

∣Rmf(x;x0)∣≤sup⁡x0<ξ<x∣f(m+1)(ξ)∣(x−x0)m+1(m+1)!.\vert R_mf(x;x_0) \vert \leq \sup_{x_0 < \xi < x} \left\vert f^{(m+1)}(\xi) \right\vert \frac{(x-x_0)^{m+1}}{(m+1)!}.
Definition 5.1 Taylor series

Let f∈C∞f \in C^\infty be a C∞C^\infty function. The Taylor series of the function f(x)f(x) with expansion point x0x_0 is the power series:

T∞f(x;x0):=Tf(x;x0):=∑k=0∞f(k)(x0)(x−x0)kk!.T_\infty f(x;x_0) := Tf(x;x_0) := \sum_{k=0}^\infty f^{(k)}(x_0) \frac{(x-x_0)^k}{k!}.
Remark

The Taylor series of a C∞C^\infty function ff is in general not convergent. It is a power series with radius of convergence ∣x−x0∣<ρ|x-x_0| < \rho (ρ\rho can be 0,  ∞0, \; \infty, or finite).

Remark

If Tf(x;x0)Tf(x;x_0) converges to ff, i.e. if:

Tf(x;x0)=∑k=0∞f(k)(x0)(x−x0)kk!=!f(x)Tf(x;x_0) = \sum_{k=0}^\infty f^{(k)}(x_0) \frac{(x-x_0)^k}{k!} \overset{!}{=} f(x)

then the function ff is called real analytic (among others: trigonometric functions, logarithm, exponential function, polynomials).

Example 5.1

Find the Taylor polynomial of order mm of the function f(x)=cos⁡(x)f(x) = \cos(x) at the point x0=0x_0=0.

Show solutionHide solution

We compute the derivatives of the function f(x)=cos⁡(x)f(x) = \cos(x) at the point x0=0x_0=0:

f(x)=cos⁡(x)⇒f(0)=1f′(x)=−sin⁡(x)⇒f′(0)=0f′′(x)=−cos⁡(x)⇒f′′(0)=−1f′′′(x)=sin⁡(x)⇒f′′′(0)=0⋮\begin{aligned} f(x) = \cos(x) \quad &\Rightarrow \quad f(0) = 1\\ f'(x) = -\sin(x) \quad &\Rightarrow \quad f'(0) = 0\\ f''(x) = -\cos(x) \quad &\Rightarrow \quad f''(0) = -1\\ f'''(x) = \sin(x) \quad &\Rightarrow \quad f'''(0) = 0\\ &\vdots \end{aligned}

So we find

f(2n)(0)=(−1)n,f(2n+1)(0)=0f^{(2n)}(0) = (-1)^n, \quad f^{(2n+1)}(0) = 0

So the Taylor polynomial we want is

Tmf(x;x0)=f(x0)+f′(x0)(x−x0)+f′′(x0)(x−x0)22+...+f(m)(x0)(x−x0)mm!Tmf(x;0)=f(0)+f′(0)(x−0)+f′′(0)(x−0)22+...+f(m)(0)(x−0)mm!=cos⁡(0)+cos⁡′(0)(x−0)+cos⁡′′(0)(x−0)22+...+cos⁡(m)(0)(x−0)mm!=cos⁡(0)−sin⁡(0)(x−0)−cos⁡(0)(x−0)22+...+cos⁡(m)(0)(x−0)mm!=1−sin⁡(0)(x−0)⏟=0−(x−0)22+...+(−1)n(x−0)2n(2n)!=1−x22+x44!+...+(−1)nx2n(2n)!,\begin{aligned} T_mf(x;x_0) &= f(x_0) + f'(x_0)(x-x_0) + f''(x_0) \frac{(x-x_0)^2}{2} + ... + f^{(m)}(x_0) \frac{(x-x_0)^m}{m!}\\ T_mf(x;0) &= f(0) + f'(0)(x-0) + f''(0) \frac{(x-0)^2}{2} + ... + f^{(m)}(0) \frac{(x-0)^m}{m!}\\ &= \cos(0) + \cos'(0)(x-0) + \cos''(0) \frac{(x-0)^2}{2} + ... + \cos^{(m)}(0) \frac{(x-0)^m}{m!}\\ &= \cos(0) - \sin(0)(x-0) - \cos(0) \frac{(x-0)^2}{2} + ... + \cos^{(m)}(0) \frac{(x-0)^m}{m!}\\ &= 1 - \underbrace{\sin(0)(x-0)}_{=0} - \frac{(x-0)^2}{2} + ... + \frac{(-1)^n(x-0)^{2n}}{(2n)!}\\ &= 1 - \frac{x^2}{2} + \frac{x^4}{4!} + ... + \frac{(-1)^n x^{2n}}{(2n)!}, \end{aligned}

where m=2nm=2n if mm is even and m=2n+1m = 2n+1 if mm is odd.

Example 5.2

Using the Taylor approximation, compute the function f(x)=(1+x)αf(x) = (1+x)^\alpha at the point x=1.2x = 1.2 for α=17\alpha = \frac{1}{7}, accurate to one decimal place.

Show solutionHide solution

The nearest point that is "easy" to compute is x0=1x_0 = 1.
We compute the derivatives of the function f(x)=(1+x)17f(x) = (1+x)^\frac{1}{7} at the point x0=1x_0=1:

f(x)=(1+x)α⇒f(1)=217f′(x)=α(1+x)α−1⇒f′(1)=17⋅2−67f′′(x)=α(α−1)(1+x)α−2⇒f′′(1)=17⋅−67⋅2−137=−649⋅2−137f′′′(x)=α(α−1)(α−2)(1+x)α−3⇒f′′′(1)=17⋅−67⋅−137⋅2−207=78343⋅2−207⋮f(m+1)(x)=α(α−1)⋅...⋅(α−m)(1+x)α−(m+1)\begin{aligned} f(x) = (1+x)^\alpha \quad &\Rightarrow \quad f(1) = 2^{\frac{1}{7}}\\ f'(x) = \alpha (1+x)^{\alpha-1} \quad &\Rightarrow \quad f'(1) = \frac{1}{7} \cdot 2^{-\frac{6}{7}}\\ f''(x) = \alpha (\alpha-1)(1+x)^{\alpha-2} \quad &\Rightarrow \quad f''(1) = \frac{1}{7} \cdot \frac{-6}{7} \cdot 2^{-\frac{13}{7}} = -\frac{6}{49} \cdot 2^{-\frac{13}{7}}\\ f'''(x) = \alpha (\alpha-1)(\alpha-2)(1+x)^{\alpha-3} \quad &\Rightarrow \quad f'''(1) = \frac{1}{7} \cdot \frac{-6}{7} \cdot \frac{-13}{7} \cdot 2^{-\frac{20}{7}} = \frac{78}{343} \cdot 2^{-\frac{20}{7}}\\ &\vdots\\ f^{(m+1)}(x) = \alpha (\alpha-1) \cdot .&.. \cdot (\alpha-m)(1+x)^{\alpha-(m+1)} \end{aligned}

Now we compute, or rather estimate, the remainder to reach the desired accuracy.

∣f(m+1)(x)∣=∣α∣∣α−1∣⋅...⋅∣α−m∣∣1+x∣α−(m+1)∣Rmf(x;x0)∣≤sup⁡x0<ξ<x∣f(m+1)(ξ)∣(x−x0)m+1(m+1)!⇒sup⁡x0<ξ<x∣f(m+1)(ξ)∣(x−x0)m+1(m+1)!=sup⁡x0<ξ<x∣α∣∣α−1∣⋅...⋅∣α−m∣∣1+ξ∣α−(m+1)(x−x0)m+1(m+1)!=sup⁡x0<ξ<xα⏟<1∣α−1∣2⏟<1⋅...⋅∣α−m∣m+1⏟<1∣1+ξ∣α−(m+1)(x−x0)m+1⏟<1<sup⁡x0<ξ<x∣1+ξ∣α−(m+1)=ξ>0sup⁡x0<ξ<x(1+ξ)α−(m+1)⏟monotonically decreasing=(1+x0)α−(m+1)(i)\begin{aligned} |f^{(m+1)}(x)| &= |\alpha| |\alpha-1| \cdot ... \cdot |\alpha-m||1+x|^{\alpha-(m+1)}\\ |R_mf(x;x_0)| &\leq \sup_{x_0 < \xi < x} |f^{(m+1)}(\xi)| \frac{(x-x_0)^{m+1}}{(m+1)!}\\ \Rightarrow \quad \sup_{x_0 < \xi < x} |f^{(m+1)}(\xi)| \frac{(x-x_0)^{m+1}}{(m+1)!} &= \sup_{x_0 < \xi < x} |\alpha| |\alpha-1| \cdot ... \cdot |\alpha-m||1+\xi|^{\alpha-(m+1)} \frac{(x-x_0)^{m+1}}{(m+1)!}\\ &= \sup_{x_0 < \xi < x} \underbrace{\alpha}_{<1} \underbrace{\frac{|\alpha-1|}{2}}_{<1} \cdot ... \cdot \underbrace{\frac{|\alpha-m|}{m+1}}_{<1} |1+\xi|^{\alpha-(m+1)} \underbrace{(x-x_0)^{m+1}}_{<1}\\ &< \sup_{x_0 < \xi < x} |1+\xi|^{\alpha-(m+1)}\\ &\overset{\xi >0}{=} \sup_{x_0 < \xi < x} \underbrace{(1+\xi)^{\alpha-(m+1)}}_{\text{monotonically decreasing}}\\ &= (1+x_0)^{\alpha-(m+1)} \qquad\qquad\qquad (i) \end{aligned}

The order m must be chosen such that ∣Rmf(x;x0)∣<(1+x0)α−(m+1)≤110|R_mf(x;x_0)| < (1+x_0)^{\alpha-(m+1)} \leq \frac{1}{10} holds. Now we can solve the last inequality for mm (A) or test different values of mm (B).

(A)

(1+1)α−(m+1)≤110⇔2α−(m+1)≤110⇔α−(m+1)≤log⁡2(110)⇔α−m−1≤−log⁡2(10)⇔α−1+log⁡2(10)≤m⇒α=1717−1+log⁡2(10)⏟≈2.46≤m⟹choose m=3.\begin{aligned} (1+1)^{\alpha-(m+1)} &\leq \frac{1}{10}\\ \Leftrightarrow \quad 2^{\alpha-(m+1)} &\leq \frac{1}{10}\\ \Leftrightarrow \quad \alpha-(m+1) &\leq \log_2\left(\frac{1}{10}\right)\\ \Leftrightarrow \quad \alpha-m-1 &\leq -\log_2(10)\\ \Leftrightarrow \quad \alpha-1 + \log_2(10) &\leq m\\ \overset{\alpha=\frac{1}{7}}{\Rightarrow} \quad \underbrace{\frac{1}{7}-1 + \log_2(10)}_{\approx 2.46} &\leq m\\ \Longrightarrow \text{choose } m=3. \end{aligned}

(B)

m=0m=0 & (i):(i):

(1+1)17−(0+1)=217−1=2−67=1267≈0.55>110\begin{aligned} (1+1)^{\frac{1}{7} -(0+1)} &= 2^{\frac{1}{7} -1}\\ &= 2^{-\frac{6}{7}}\\ &= \frac{1}{2^\frac{6}{7}} \approx 0.55 > \frac{1}{10} \end{aligned}

m=1m=1 & (i):(i):

(1+1)17−(1+1)=217−2=2−137=12137≈0.27>110\begin{aligned} (1+1)^{\frac{1}{7} -(1+1)} &= 2^{\frac{1}{7} -2}\\ &= 2^{-\frac{13}{7}}\\ &= \frac{1}{2^{\frac{13}{7}}} \approx 0.27 > \frac{1}{10} \end{aligned}

m=2m=2 & (i):(i):

(1+1)17−(2+1)=217−3=2−207=12207≈0.14>110\begin{aligned} (1+1)^{\frac{1}{7} -(2+1)} &= 2^{\frac{1}{7} -3}\\ &= 2^{-\frac{20}{7}}\\ &= \frac{1}{2^{\frac{20}{7}}} \approx 0.14 > \frac{1}{10} \end{aligned}

m=3m=3 & (i):(i):

(1+1)17−(3+1)=217−4=2−277=12277≈0.07<110\begin{aligned} (1+1)^{\frac{1}{7} -(3+1)} &= 2^{\frac{1}{7} -4}\\ &= 2^{-\frac{27}{7}}\\ &= \frac{1}{2^{\frac{27}{7}}} \approx 0.07 < \frac{1}{10} \end{aligned}

Here we test different values of mm.

So we have found that the 3rd-order Taylor approximation meets our accuracy requirement, and we now compute the approximation:

T3f(x;x0)=f(x0)+f′(x0)(x−x0)+12f′′(x0)(x−x0)2+13!f′′′(x0)(x−x0)3T3f(1.2;1)=f(1)+f′(1)(1.2−1)+12f′′(1)(1.2−1)2+13!f′′′(1)(1.2−1)3=f(1)+f′(1)⋅0.2+12f′′(1)⋅0.22+16f′′′(1)⋅0.23=217+17⋅2−67⋅0.2+12−649⋅2−137⋅0.22+1678343⋅2−207⋅0.23≈1.1192280938(exact: 1.11922531815).\begin{aligned} T_3f(x;x_0) &= f(x_0) + f'(x_0) (x-x_0) + \frac{1}{2} f''(x_0) (x-x_0)^2 + \frac{1}{3!} f'''(x_0) (x-x_0)^3\\ T_3f(1.2;1) &= f(1) + f'(1) (1.2-1) + \frac{1}{2} f''(1) (1.2-1)^2 + \frac{1}{3!} f'''(1) (1.2-1)^3\\ &= f(1) + f'(1) \cdot 0.2 + \frac{1}{2} f''(1) \cdot 0.2^2 + \frac{1}{6} f'''(1) \cdot 0.2^3\\ &= 2^{\frac{1}{7}} + \frac{1}{7} \cdot 2^{-\frac{6}{7}} \cdot 0.2 + \frac{1}{2} \frac{-6}{49} \cdot 2^{-\frac{13}{7}} \cdot 0.2^2 + \frac{1}{6} \frac{78}{343} \cdot 2^{-\frac{20}{7}} \cdot 0.2^3\\ &\approx 1.1192280938 \qquad\qquad\qquad (\text{exact: } 1.11922531815). \end{aligned}