A function F∈C1([a,b]) is called an antiderivative of f if:
F′(x)=f(x)∀x∈[a,b].
Definition 1.2
A partition of the compact interval I=[a,b] with a<b is a finite subset P⊂I with P={a=x0,x1,...,xn=b} where x0<...<xn. We denote the set of all partitions of I by
P(I):={P⊂I∣a,b∈P,P is finite}.
The mesh of the partition is defined by
δ(P):=1≤i≤nmax{∣xi−xi−1∣}with n(P):=n=∣P∣−1.
Remark: δ(P) is the length of the largest interval Ii=[xi−1,xi] with 1≤i≤n.
Let ξi∈Ii be intermediate points (sample points) with xi−1≤ξi≤xi. Every sum of the form
S(f,P,ξ):=i=1∑nf(ξi)(xi−xi−1)
is called a Riemann sum of the partition P and ξ∈{ξ1,...,ξn}.
SatzStruwe 6.2.1
f monotone⇒f is integrable.
SatzStruwe 6.2.2
f continuous⇒f is integrable.
Definition 1.3upper sum
S(f,P):=i=1∑nξ∈[xi−1,xi]supf(ξ)(xi−xi−1)
Definition 1.4lower sum
S(f,P):=i=1∑nξ∈[xi−1,xi]inff(ξ)(xi−xi−1)
Definition 1.5upper Riemann integral
∫abf(x)dx:=inf{S(f,P)∣P∈P(I)}.
Definition 1.6lower Riemann integral
∫abf(x)dx:=sup{S(f,P)∣P∈P(I)}.
Corollary 1.1Struwe 6.2.3
Let f:[a,b]→R be a bounded and integrable function. Let {P(n)} be a sequence of partitions of the compact interval [a,b] with δ(P(n))n→∞0 and {ξ(n)} a fixed choice of intermediate points for the partition P(n). Then
∫abf(x)dx=n→∞limS(f,P(n),ξ(n)).
Remark
If corollary 6.2.3 is satisfied, f is called Riemann integrable on [a,b].
Remark
Uniform partitions P(n)={a=x0,x1,...,xn} with xi=a+ni(b−a) and xi−xi−1=nb−a form a sequence of partitions with δ(P(n))=nb−an→∞0.
Example 1.1
Compute the integral 0∫1(x3−2x)dx explicitly using Riemann sums.
Show solutionHide solution
We split the interval of integration [0,1] into n subintervals of equal length Δx=n1 with xk=nk,k=1,...,n. The mesh of this partition is δ(P)=n1. We have n→∞limδ(P)=0. Hence the function f(x)=x3−2x is Riemann integrable and, by the definition of the Riemann integral (we choose ξk=xk), we get:
Hence (f∘g)(x)=f(g(x)) is an antiderivative of f′(g(x))g′(x).
Example 1.2
Find the antiderivative ∫1+x2arctan5(x)dx.
Show solutionHide solution∫1+x2arctan5(x)dx=∫f(g(x))arctan5(x)⋅g′(x)1+x21dx=61arctan6(x)+C,∀C∈R.
Example 1.3
Find the antiderivative ∫x+13x+1dx.
Show solutionHide solution∫x+13x+1dx=∫x+1elog(3)x+1dx=2∫2x+1elog(3)x+1dx=2∫f(g(x))elog(3)x+1⋅g′(x)2x+11dx=log(3)2⋅elog(3)x+1+C,∀C∈R=log(3)2⋅3x+1+C,∀C∈R.
Satzfundamental theorem of calculus (version A)
Let f∈C0([a,b]). Set F(x)=a∫xf(t)dt,x∈[a,b].
Then F∈C1((a,b)) with F′(x)=f(x).
Satzfundamental theorem of calculus (version B)
Let f∈C0([a,b]) and F an antiderivative of f.
Then:
∫abf(x)dx=F(b)−F(a).
Example 1.4
Let A(x):=−1∫cos(x)earccos(t)dt. Compute A′(x).
Show solutionHide solutionF(x)Φ(x)⇒A(x)⇒A′(x):=∫−1xearccos(t)dt:=cos(x)=(F∘Φ)(x)=F′(Φ(x))⋅Φ′(x)=earccos(cos(x))⋅(−sin(x))=−sin(x)⋅exfor x∈[0,π]
Integration (II)
Satzmonotonicity of the R-integral; Struwe 6.3.1
Let f,g:[a,b]→R be R-integrable with f≤g. Then:
∫abf(x)dx≤∫abg(x)dx.
Satzlinearity of the R-integral; Struwe 6.3.2
Let α,β∈R,I=[a,b] and f,g:I→R two R-integrable functions. Then the function αf(x)+βg(x) is integrable over [a,b] and:
∫abαf(x)+βg(x)dx=α∫abf(x)dx+β∫abg(x)dx.
Corollary 2.1standard estimate; Struwe 6.3.1
Let f be integrable over [a,b]. Then ∣f∣ is R-integrable and:
Let f:[a,b]→R be integrable over [a,b]. Let c∈[a,b]. Then f[a,c] and f[c,b] are integrable and:
∫abf(x)dx=∫acf(x)dx+∫cbf(x)dx.
Remark
Convention:
∫abf(x)dx=−∫baf(x)dx
∫aaf(x)dx=0
Satzintegration by parts
∫abf′(x)⋅g(x)dx=[f(x)g(x)]ab−∫abf(x)⋅g′(x)dx.
Derivation:
Show solutionHide solutionuv⇔∫u′vdx=∫(uv)′dx=product rule∫(u′v+uv′)dx=linearity∫u′vdx+∫uv′dx=uv−∫uv′dx
Example 2.1
Compute ∫log2(x)dx.
Show solutionHide solution∫log2(x)dx=∫log2(x)⋅1dx=IBPlog2(x)⋅x−∫2log(x)⋅x1⋅xdx=log2(x)⋅x−2∫log(x)dx=log2(x)⋅x−2(∫log(x)⋅1dx)=IBPlog2(x)⋅x−2(log(x)⋅x−∫x1⋅xdx)=log2(x)⋅x−2(log(x)⋅x−x)+C,C∈R=xlog2(x)−2xlog(x)+2x+C,C∈R
Example 2.2
Compute ∫e2xcos(x)dx.
Show solutionHide solution∫e2xcos(x)dx⇔5∫e2xcos(x)dx⇔∫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=e2xsin(x)+2e2xcos(x)+C,C∈R=51e2xsin(x)+52e2xcos(x)+C,C∈R
Example 2.3
Compute 0∫1xsinh(4x)dx.
Show solutionHide solution∫01xsinh(4x)dx=[x⋅4cosh(4x)]01−∫011⋅4cosh(4x)dx=[x⋅4cosh(4x)]01−41∫01cosh(4x)dx=[1⋅4cosh(4)−0⋅4cosh(0)]−41[4sinh(4x)]01=4cosh(4)−41[4sinh(4)−4sinh(0)]=sinh(0)=04cosh(4)−41[4sinh(4)−0]=4cosh(4)−16sinh(4)
Remark
What about ∫x1dx?
Show solutionHide solutionf1:(0,∞)→R,x↦x1⇒∫f1(x)dx=log(x)+C,C∈Rf2:R∖{0}→R,x↦x1⇒∫f2(x)dx=log∣x∣+C,C∈R
x2+bx+c in the numerator, use sin2(x)+cos2(x)=1 or cosh2(x)−sinh2(x)=1 ∫1−x2dx: substitute x=sin(t) or x=cos(t) ∫x2−1dx: substitute x=cosh(t) ∫x2+1dx: substitute x=sinh(t)
a2+b2x2
Substitution: x=ba⋅tan(t),dx=bcos2(t)a⋅dt or x=ba⋅sinh(t),dx=ba⋅cosh(t)⋅dt
b2x2−a2
Substitution: x=bcos(t)a, dx=ba⋅cos2(t)sin(t)⋅dt or x=ba⋅cosh(t),dx=ba⋅sinh(t)dt
Partial Fraction Decomposition
Remark
When we add two fractions, we have to find a common denominator:
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)=q(x)p(x)
Wanted: partial fraction decomposition of f(x)
If deg(p)≥deg(q), carry out polynomial long division to get the following form:
f(x)=q(x)p(x)=s(x)+remainderq(x)p(x)
If deg(p)<deg(q), set p(x)=p(x).
Find the roots of q(x) and factorise 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,−2, etc.) and then carry out a polynomial division.
Use the ansatz for partial fraction decomposition, with the factorised q(x) from step (ii):
−3x3−2x2+3x+213x2+8x−1=−(x−1)(x+1)(3x+2)13x2+8x−1=!(x−1)(x+1)(3x+2)(3A+3B+C)x2+(5A−B)x+(2A−2B−C)power of xx2:x1:x0:ansatz3A+3B+C5A−B2A−2B−Cgiven numerator polynomial=−13=−8=+1
Solve this system of equations; you can use the methods from linear algebra here. This gives you the following solution:
ABC=−2=−2=−1
So in the end we get the following partial fraction decomposition:
Let f:(a,b)→R be Riemann integrable over every compact interval [c,d]⊂(a,b). f is called improperly Riemann integrable over (a,b) if
∫abfdx:=c→a+,d→b−lim∫cdfdx
exists.
Definition 4.4Michaels: improper integrals of type a
Let f(x) be continuous on [a,∞). Then we set
∫a∞f(x)dx:=R→∞lim∫aRf(x)dx,
if the limit exists. This value is called an improper integral.
Let f(x) be continuous on (−∞,b]. Then we set
∫−∞bf(x)dx:=R→−∞lim∫Rbf(x)dx,
if the limit exists. This value is called an improper integral.
Definition 4.5Michaels: improper integrals of type b
Let f(x) be continuous on (a,b]. Then we set
∫abf(x)dx:=ε→0+lim∫a+εbf(x)dx,
if the limit exists. This value is called an improper integral.
Let f(x) be continuous on [a,b). Then we set
∫abf(x)dx:=ε→0+lim∫ab−εf(x)dx,
if the limit exists. This value is called an improper integral.
SatzStruwe 6.4.1
Let f:[1,∞)→R+ be monotonically decreasing. Then the series k=1∑∞f(k) converges if and only if 1∫∞fdx converges, and in that case:
0≤k=1∑∞f(k)−∫1∞fdx≤f(1).
Convergence Tests for Improper Integrals
Remarkcriterion 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.
In summary, we have derived the following important rule of thumb:
∫1∞xp1dx converges⇔p>1.
Satzcriterion 2a: comparison test
Let f,g be continuous on [a,∞) with
0≤f(x)≤g(x)∀x∈[a,∞)
If a∫∞g(x)dx converges, so does a∫∞f(x)dx,
If a∫∞f(x)dx diverges, so does a∫∞g(x)dx.
Satzcriterion 2b: comparison test
Let f,g be continuous on (a,b] with
0≤f(x)≤g(x)∀x∈(a,b]
If a∫bg(x)dx converges, so does a∫bf(x)dx,
If a∫bf(x)dx diverges, so does a∫bg(x)dx.
Satzcriterion 3a: absolute convergence
∫a∞∣f(x)∣dx<∞⇒∫a∞f(x)dx<∞.
Satzcriterion 3b: absolute convergence
∫ab∣f(x)∣dx<∞⇒∫abf(x)dx<∞.
Satzcriterion 4a: limit test
Test 1
f(x),g(x) continuous on [a,∞)
x→∞limg(x)f(x)=A=∞
then a∫∞∣g(x)∣dx<∞ converges ⇔∫a∞∣f(x)∣dx converges.
Test 2
f(x) continuous on [a,∞)
x→∞limxpf(x)=A=∞ for some p>1.
then a∫∞∣f(x)∣dx converges, and hence so does a∫∞f(x)dx (absolute convergence)
Test 3
f(x) continuous on [a,∞)
x→∞limxf(x)=A=0,∞.
then a∫∞f(x)dx diverges. (Note: test 3 fails if A=0.)
Satzcriterion 4b: limit test
Test 1
f(x),g(x) continuous on (a,b]
x→a+limg(x)f(x)=A=∞
then a∫b∣g(x)∣dx<∞ converges ⇔a∫b∣f(x)∣dx converges.
Test 2
f(x) continuous on (a,b]
x→a+lim(x−a)pf(x)=A=∞ for some 0<p<1.
then a∫b∣f(x)∣dx converges, and hence so does a∫bf(x)dx (absolute convergence)
Test 3
f(x) continuous on (a,b]
x→a+lim(x−a)f(x)=A=0,∞.
then a∫bf(x)dx diverges. (Note: test 3 fails if A=0.)
Satzcriterion 5a: Leibniz criterion for improper integrals
Let f(x) be a continuous function on [a,∞). If the function f(x) is monotonically decreasing and x→∞limf(x)=0, then the improper integrals
∫a∞f(x)sin(x)dxand∫a∞f(x)cos(x)dx
converge.
Satzcriterion 5b: Leibniz criterion for improper integrals
Let f(x) be a continuous function on (a,b]. If the function f(x)(x−a)2 is monotonically increasing and x→a+limf(x)(x−a)2=0, then the improper integrals
∫abf(x)sin(x−a1)dxand∫abf(x)cos(x−a1)dx
converge.
Satzcriterion 6b: bounded function on a bounded interval
Let f be continuous on (a,b]. If
x→a+limf(x)=A=∞,
then the integral a∫bf(x)dx converges.
The Taylor Polynomial and Taylor’s Formula
SatzTaylor formula; Struwe 5.5.1
Let f:[a,b]→R be a Cm([a,b]) function (i.e. m times differentiable on (a,b)). Then:
f(x)=Tmf(x;x0)+Rmf(x;x0)
with the Taylor polynomial of degree m at the expansion point x0:
Tmf(x;x0):=k=0∑mf(k)(x0)k!(x−x0)k
and for the remainder Rmf(x;x0):=f(x)−Tmf(x;x0) we have:
x→x0lim(x−x0)mRmf(x;x0)=0.
If f is m+1 times differentiable, the Taylor formula holds:
f(x)=Tmf(x;x0)+f(m+1)(ξ)(m+1)!(x−x0)m+1for some ξ∈(x0,x)
If f∈Cm+1[a,b], then the estimate for the remainder holds:
The Taylor series of a C∞ function f is in general not convergent. It is a power series with radius of convergence ∣x−x0∣<ρ (ρ can be 0,∞, or finite).
Remark
If Tf(x;x0) converges to f, i.e. if:
Tf(x;x0)=k=0∑∞f(k)(x0)k!(x−x0)k=!f(x)
then the function f is called real analytic (among others: trigonometric functions, logarithm, exponential function, polynomials).
Example 5.1
Find the Taylor polynomial of order m of the function f(x)=cos(x) at the point x0=0.
Show solutionHide solution
We compute the derivatives of the function f(x)=cos(x) at the point x0=0:
The order m must be chosen such that ∣Rmf(x;x0)∣<(1+x0)α−(m+1)≤101 holds. Now we can solve the last inequality for m (A) or test different values of m (B).