Processing math: 28%

3. Calculus

3.1 Integrals

3.1.1 Arithmetic rules

The primitive function F(x) of f(x) obeys the rule F(x)=f(x). With F(x) the primitive of f(x) holds for the definite integral baf(x)dx=F(b)F(a) If u=f(x) holds: bag(f(x))df(x)=f(b)f(a)g(u)du Partial integration: with F and G the primitives of f and g holds: f(x)g(x)dx=f(x)G(x)G(x)df(x)dxdx A derivative can be brought under the intergral sign (see section 1.8.3 for the required conditions): ddy[x=h(y)x=g(y)f(x,y)dx]=x=h(y)x=g(y)f(x,y)ydxf(g(y),y)dg(y)dy+f(h(y),y)dh(y)dy

3.1.2 Arc lengts, surfaces and volumes

The arc length of a curve y(x) is given by: =1+(dy(x)dx)2dx The arc length of a parameter curve F(x(t)) is: =Fds=F(x(t))|˙x(t)|dt with t=dxds=˙x(t)|˙x(t)|   ,  |t |=1 (v,t)ds=(v,˙t(t))dt=(v1dx+v2dy+v3dz) The surface A of a solid of revolution is: A=2πy1+(dy(x)dx)2dx The volume V of a solid of revolution is: V=πf2(x)dx

3.1.3 Separation of quotients

Every rational function P(x)/Q(x) where P and Q are polynomials can be written as a linear combination of functions of the type (xa)k with kZ, and of functions of the type px+q((xa)2+b2)n with b>0 and nIN. So: p(x)(xa)n=nk=1Ak(xa)k  ,   p(x)((xb)2+c2)n=nk=1Akx+B((xb)2+c2)k Recurrent relation: for n0 holds: dx(x2+1)n+1=12nx(x2+1)n+2n12ndx(x2+1)n

3.1.4 Special functions

3.1.4.1 Elliptic functions

Elliptic functions can be written as a power series as follows: 1k2sin2(x)=1n=1(2n1)!!(2n)!!(2n1)k2nsin2n(x) 11k2sin2(x)=1+n=1(2n1)!!(2n)!!k2nsin2n(x) with n!!=n(n2)!!.

3.1.4.2 The Gamma function

The gamma function Γ(y) is defined by: Γ(y)=0exxy1dx One can derive that Γ(y+1)=yΓ(y)=y!. This is a way to define faculties for non-integers. Further one can derive that Γ(n+12)=π2n(2n1)!!  and  Γ(n)(y)=0exxy1lnn(x)dx

3.1.4.3 The Beta function

The betafunction β(p,q) is defined by: β(p,q)=10xp1(1x)q1dx with p and q >0. The beta and gamma functions are related by the following equation: β(p,q)=Γ(p)Γ(q)Γ(p+q)

3.1.4.4 The Delta function

The delta function δ(x) is an infinitely thin peak function with surface 1. It can be defined by: δ(x)=lim Some properties are: \int\limits_{-\infty}^\infty\delta(x)dx=1~~,~~~ \int\limits_{-\infty}^\infty F(x)\delta(x)dx=F(0)

3.1.5 Goniometric integrals

When solving goniometric integrals it can be useful to change variables. The following holds if one defines \tan(\frac{1}{2}x):=t: dx=\frac{2dt}{1+t^2}~,~~\cos(x)=\frac{1-t^2}{1+t^2}~,~~\sin(x)=\frac{2t}{1+t^2} Each integral of the type \int R(x,\sqrt{ax^2+bx+c})dx can be converted into one of the types that were treated in section 3.1.3. After this conversion one can substitute in the integrals of the type: \begin{eqnarray*} \int R(x,\sqrt{x^2+1})dx&~:~~&x=\tan(\varphi) ~,dx=\frac{d\varphi}{\cos(\varphi)} ~~\mbox{or}~~\sqrt{x^2+1}=t+x\\ \int R(x,\sqrt{1-x^2})dx&~:~~&x=\sin(\varphi) ~,dx=\cos(\varphi)d\varphi ~~\mbox{or}~~\sqrt{1-x^2}=1-tx\\ \int R(x,\sqrt{x^2-1})dx&~:~~&x=\frac{1}{\cos(\varphi)}~,dx=\frac{\sin(\varphi)}{\cos^2(\varphi)}d\varphi~~\mbox{or}~~\sqrt{x^2-1}=x-t \end{eqnarray*} These definite integrals are easily solved: \int\limits_0^{\pi/2}\cos^n(x)\sin^m(x)dx=\frac{(n-1)!!(m-1)!!}{(m+n)!!}\cdot \left\{\begin{array}{l} \pi/2\;\;\mbox{when $m$ and $n$ are both even}\\ 1\;\;\;\;\mbox{in all other cases} \end{array}\right. Some important integrals are: \int\limits_0^\infty\frac{xdx}{{\rm e}^{ax}+1}=\frac{\pi^2}{12a^2}~~,~~ \int\limits_{-\infty}^\infty\frac{x^2dx}{({\rm e}^x+1)^2}=\frac{\pi^2}{3}~~,~~ \int\limits_0^\infty\frac{x^3dx}{{\rm e}^x-1}=\frac{\pi^4}{15}

3.2 Functions with more variables

3.2.1 Derivatives

The partial derivative with respect to x of a function f(x,y) is defined by: \left(\frac{\partial f}{\partial x}\right)_{x_0}=\lim_{h\rightarrow0}\frac{f(x_0+h,y_0)-f(x_0,y_0)}{h} The directional derivative in the direction of \alpha is defined by: \frac{\partial f}{\partial\alpha}=\lim_{r\downarrow0}\frac{f(x_0+r\cos(\alpha),y_0+r\sin(\alpha))-f(x_0,y_0)}{r}= (\vec{\nabla}f,(\sin\alpha,\cos\alpha))=\frac{\nabla f\cdot\vec{v}}{|\vec{v}|} When one changes to coordinates f(x(u,v),y(u,v)) holds: \frac{\partial f}{\partial u}=\frac{\partial f}{\partial x}\frac{\partial x}{\partial u}+\frac{\partial f}{\partial y}\frac{\partial y}{\partial u} If x(t) and y(t) depend only on one parameter t holds: \frac{\partial f}{\partial t}=\frac{\partial f}{\partial x}\frac{dx}{dt}+\frac{\partial f}{\partial y}\frac{dy}{dt} The total differential df of a function of 3 variables is given by: df=\frac{\partial f}{\partial x}dx+\frac{\partial f}{\partial y}dy+\frac{\partial f}{\partial z}dz So \frac{df}{dx}=\frac{\partial f}{\partial x}+\frac{\partial f}{\partial y}\frac{dy}{dx}+\frac{\partial f}{\partial z}\frac{dz}{dx} The tangent in point \vec{x}_0 at the surface f(x,y)=0 is given by the equation f_x(\vec{x}_0)(x-x_0)+f_y(\vec{x}_0)(y-y_0)=0.

The tangent plane in \vec{x}_0 is given by: f_x(\vec{x}_0)(x-x_0)+f_y(\vec{x}_0)(y-y_0)=z-f(\vec{x}_0).

3.2.2 Taylor series

A function of two variables can be expanded as follows in a Taylor series: f(x_0+h,y_0+k)=\sum\limits_{p=0}^n \frac{1}{p!} \left(h\frac{\partial^p}{\partial x^p}+k\frac{\partial^p}{\partial y^p}\right)f(x_0,y_0)+R(n) with R(n) the residual error and \left(h\frac{\partial^p}{\partial x^p}+k\frac{\partial^p}{\partial y^p}\right)f(a,b)=\sum\limits_{m=0}^p{p\choose m} h^mk^{p-m}\frac{\partial^pf(a,b)}{\partial x^m\partial y^{p-m}}

3.2.3 Extrema

When f is continuous on a compact boundary V there exists a global maximum and a global minumum for f on this boundary. A boundary is called compact if it is limited and closed.

Possible extrema of f(x,y) on a boundary V\in I\hspace{-1mm}R^2 are:

  1. Points on V where f(x,y) is not differentiable,
  2. Points where \vec{\nabla}f=\vec{0},
  3. If the boundary V is given by \varphi(x,y)=0, than all points where \vec{\nabla}f(x,y)+\lambda\vec{\nabla}\varphi(x,y)=0 are possible for extrema. This is the multiplicator method of Lagrange, \lambda is called a multiplicator.

The same as in I\hspace{-1mm}R^2 holds in I\hspace{-1mm}R^3 when the area to be searched is constrained by a compact V, and V is defined by \varphi_1(x,y,z)=0 and \varphi_2(x,y,z)=0 for extrema of f(x,y,z) for points (1) and (2). Point (3) is rewritten as follows: possible extrema are points where \vec{\nabla}f(x,y,z)+\lambda_1\vec{\nabla}\varphi_1(x,y,z)+\lambda_2\vec{\nabla}\varphi_2(x,y,z)=0.

3.2.4 The \nabla-operator

In cartesian coordinates (x,y,z) holds: \begin{eqnarray*} \vec{\nabla} &=&\frac{\partial}{\partial x}\vec{e}_{x}+\frac{\partial}{\partial y}\vec{e}_{y}+\frac{\partial}{\partial z}\vec{e}_{z}\\ {\rm grad}f &=&\frac{\partial f}{\partial x}\vec{e}_{x}+\frac{\partial f}{\partial y}\vec{e}_{y}+\frac{\partial f}{\partial z}\vec{e}_{z}\\ {\rm div}~\vec{a}&=&\frac{\partial a_x}{\partial x}+\frac{\partial a_y}{\partial y}+\frac{\partial a_z}{\partial z}\\ {\rm curl}~\vec{a}&=&\left(\frac{\partial a_z}{\partial y}-\frac{\partial a_y}{\partial z}\right)\vec{e}_{x}+ \left(\frac{\partial a_x}{\partial z}-\frac{\partial a_z}{\partial x}\right)\vec{e}_{y}+ \left(\frac{\partial a_y}{\partial x}-\frac{\partial a_x}{\partial y}\right)\vec{e}_{z}\\ \nabla^2 f &=&\frac{\partial^2f}{\partial x^2}+\frac{\partial^2f}{\partial y^2}+\frac{\partial^2f}{\partial z^2} \end{eqnarray*}

In cylindrical coordinates (r,\varphi,z) holds: \begin{eqnarray*} \vec{\nabla} &=&\frac{\partial}{\partial r}\vec{e}_{r}+\frac{1}{r}\frac{\partial}{\partial\varphi}\vec{e}_{\varphi}+\frac{\partial}{\partial z}\vec{e}_{z}\\ {\rm grad}f &=&\frac{\partial f}{\partial r}\vec{e}_{r}+\frac{1}{r}\frac{\partial f}{\partial\varphi}\vec{e}_{\varphi}+\frac{\partial f}{\partial z}\vec{e}_{z}\\ {\rm div}~\vec{a}&=&\frac{\partial a_r}{\partial r}+\frac{a_r}{r}+\frac{1}{r}\frac{\partial a_\varphi}{\partial\varphi}+\frac{\partial a_z}{\partial z}\\ {\rm curl}~\vec{a}&=&\left(\frac{1}{r}\frac{\partial a_z}{\partial\varphi}-\frac{\partial a_\varphi}{\partial z}\right)\vec{e}_{r}+ \left(\frac{\partial a_r}{\partial z}-\frac{\partial a_z}{\partial r}\right)\vec{e}_{\varphi}+ \left(\frac{\partial a_\varphi}{\partial r}+\frac{a_\varphi}{r}-\frac{1}{r}\frac{\partial a_r}{\partial\varphi}\right)\vec{e}_{z}\\ \nabla^2 f &=&\frac{\partial^2f}{\partial r^2}+\frac{1}{r}\frac{\partial f}{\partial r}+\frac{1}{r^2}\frac{\partial^2f}{\partial\varphi^2}+\frac{\partial^2f}{\partial z^2} \end{eqnarray*}

In spherical coordinates (r,\theta,\varphi) holds: \begin{eqnarray*} \vec{\nabla} &=&\frac{}\partial{\partial r}\vec{e}_{r}+\frac{1}{r}\frac{\partial}{\partial\theta}\vec{e}_{\theta}+\frac{1}{r\sin\theta}\frac{\partial}{\partial\varphi}\vec{e}_{\varphi}\\ {\rm grad}f &=&\frac{\partial f}{\partial r}\vec{e}_{r}+\frac{1}{r}\frac{\partial f}{\partial\theta}\vec{e}_{\theta}+\frac{1}{r\sin\theta}\frac{\partial f}{\partial\varphi}\vec{e}_{\varphi}\\ {\rm div}~\vec{a}&=&\frac{\partial a_r}{\partial r}+\frac{2a_r}{r}+\frac{1}{r}\frac{\partial a_\theta}{\partial\theta}+\frac{a_\theta}{r\tan\theta}+\frac{1}{r\sin\theta}\frac{\partial a_\varphi}{\partial\varphi}\\ {\rm curl}~\vec{a}&=&\left(\frac{1}{r}\frac{\partial a_\varphi}{\partial\theta}+\frac{a_\theta}{r\tan\theta}-\frac{1}{r\sin\theta}\frac{\partial a_\theta}{\partial\varphi}\right)\vec{e}_{r}+ \left(\frac{1}{r\sin\theta}\frac{\partial a_r}{\partial\varphi}-\frac{\partial a_\varphi}{\partial r}-\frac{a_\varphi}{r}\right)\vec{e}_{\theta}+ \left(\frac{\partial a_\theta}{\partial r}+\frac{a_\theta}{r}-\frac{1}{r}\frac{\partial a_r}{\partial\theta}\right)\vec{e}_{\varphi}\\ \nabla^2 f &=&\frac{\partial^2f}{\partial r^2}+\frac{2}{r}\frac{\partial f}{\partial r}+\frac{1}{r^2}\frac{\partial^2f}{\partial\theta^2}+\frac{1}{r^2\tan\theta}\frac{\partial f}{\partial\theta}+\frac{1}{r^2\sin^2\theta}\frac{\partial^2f}{\partial\varphi^2} \end{eqnarray*}

General orthonormal curvilinear coordinates (u,v,w) can be derived from cartesian coordinates by the transformation \vec{x}=\vec{x}(u,v,w). The unit vectors are given by: \vec{e}_{u}=\frac{1}{h_1}\frac{\partial\vec{x}}{\partial u}~,~~ \vec{e}_{v}=\frac{1}{h_2}\frac{\partial\vec{x}}{\partial v}~,~~ \vec{e}_{w}=\frac{1}{h_3}\frac{\partial\vec{x}}{\partial w} where the terms h_i give normalization to length 1. The differential operators are than given by: \begin{eqnarray*} {\rm grad}f &=&\frac{1}{h_1}\frac{\partial f}{\partial u}\vec{e}_{u}+\frac{1}{h_2}\frac{\partial f}{\partial v}\vec{e}_{v}+\frac{1}{h_3}\frac{\partial f}{\partial w}\vec{e}_{w}\\ {\rm div}~\vec{a}&=&\frac{1}{h_1h_2h_3}\left(\frac{\partial}{\partial u}(h_2h_3a_u)+\frac{\partial}{\partial v}(h_3h_1a_v)+\frac{\partial}{\partial w}(h_1h_2a_w)\right)\\ {\rm curl}~\vec{a}&=&\frac{1}{h_2h_3}\left(\frac{\partial(h_3a_w)}{\partial v}-\frac{\partial(h_2a_v)}{\partial w}\right)\vec{e}_{u}+ \frac{1}{h_3h_1}\left(\frac{\partial(h_1a_u)}{\partial w}-\frac{\partial(h_3a_w)}{\partial u}\right)\vec{e}_{v}+ \frac{1}{h_1h_2}\left(\frac{\partial(h_2a_v)}{\partial u}-\frac{\partial(h_1a_u)}{\partial v}\right)\vec{e}_{w}\\ \nabla^2 f &=&\frac{1}{h_1h_2h_3}\left[\frac{\partial}{\partial u}\left(\frac{h_2h_3}{h_1}\frac{\partial f}{\partial u}\right)+ \frac{\partial}{\partial v}\left(\frac{h_3h_1}{h_2}\frac{\partial f}{\partial v}\right)+ \frac{\partial}{\partial w}\left(\frac{h_1h_2}{h_3}\frac{\partial f}{\partial w}\right)\right] \end{eqnarray*}

Some properties of the \nabla-operator are: \begin{array}{l@{~~~~~}l@{~~~~~}l} {\rm div}(\phi\vec{v})=\phi{\rm div}\vec{v}+{\rm grad}\phi\cdot\vec{v}& {\rm curl}(\phi\vec{v})=\phi{\rm curl}\vec{v}+({\rm grad}\phi)\times\vec{v}&{\rm curl~grad}\phi=\vec{0}\\ {\rm div}(\vec{u}\times\vec{v})=\vec{v}\cdot({\rm curl}\vec{u})-\vec{u}\cdot({\rm curl}\vec{v})& {\rm curl~curl}\vec{v}={\rm grad~div}\vec{v}-\nabla^2\vec{v}&{\rm div~curl}\vec{v}=0\\ {\rm div~grad}\phi=\nabla^2\phi&\nabla^2\vec{v}\equiv(\nabla^2v_1,\nabla^2v_2,\nabla^2v_3) \end{array} Here, \vec{v} is an arbitrary vectorfield and \phi an arbitrary scalar field.

3.2.5 Integral theorems

Some important integral theorems are:

Gauss: \displaystyle\int\hspace{-2ex}\int\hspace{-2.8ex}\bigcirc (\vec{v}\cdot\vec{n})d^2A=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int({\rm div}\vec{v})d^3V
Stokes for a scalar field: \displaystyle\oint(\phi\cdot\vec{e}_t)ds=\int\hspace{-1.5ex}\int(\vec{n}\times{\rm grad}\phi)d^2A
Stokes for a vector field: \displaystyle\oint(\vec{v}\cdot\vec{e}_t)ds=\iint({\rm curl}\vec{v}\cdot\vec{n})d^2A
this gives: \displaystyle\int\hspace{-2ex}\int\hspace{-2.8ex}\bigcirc({\rm curl}\vec{v}\cdot\vec{n})d^2A=0
Ostrogradsky: \displaystyle\int\hspace{-2ex}\int\hspace{-2.8ex}\bigcirc(\vec{n}\times\vec{v})d^2A=\iiint({\rm curl}\vec{v})d^3A
\displaystyle\int\hspace{-2ex}\int\hspace{-2.8ex}\bigcirc(\phi\vec{n})d^2A=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int({\rm grad}\phi)d^3V

Here the orientable surface \int\hspace{-1mm}\int d^2A is bounded by the Jordan curve s(t).

3.2.6 Multiple integrals

Let A be a closed curve given by f(x,y)=0, than the surface A inside the curve in I\hspace{-1mm}R^2 is given by A=\int\hspace{-1.5ex}\int d^2A=\int\hspace{-1.5ex}\int dxdy Let the surface A be defined by the function z=f(x,y). The volume V bounded by A and the xy plane is than given by: V=\int\hspace{-1.5ex}\int f(x,y)dxdy The volume inside a closed surface defined by z=f(x,y) is given by: V=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int d^3V=\iint f(x,y)dxdy=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int dxdydz

3.2.7 Coordinate transformations

The expressions d^2A and d^3V transform as follows when one changes coordinates to \vec{u}=(u,v,w) through the transformation x(u,v,w): V=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int f(x,y,z)dxdydz=\int\hspace{-1.5ex}\int\hspace{-1.5ex}\int f(\vec{x}(\vec{u}))\left|\frac{\partial\vec{x}}{\partial\vec{u}}\right|dudvdw In I\hspace{-1mm}R^2 holds: \frac{\partial\vec{x}}{\partial\vec{u}}=\left|\begin{array}{cc}x_u&x_v\\ y_u&y_v\end{array}\right| Let the surface A be defined by z=F(x,y)=X(u,v). Than the volume bounded by the xy plane and F is given by: \int\hspace{-1.5ex}\int\limits_Sf(\vec{x})d^2A=\int\hspace{-1.5ex}\int\limits_Gf(\vec{x}(\vec{u})) \left|\frac{\partial X}{\partial u}\times\frac{\partial X}{\partial v}\right|dudv= \int\hspace{-1.5ex}\int\limits_Gf(x,y,F(x,y))\sqrt{1+\partial_xF^2+\partial_yF^2}dxdy

3.3 Orthogonality of functions

The inner product of two functions f(x) and g(x) on the interval [a,b] is given by: (f,g)=\int\limits_a^bf(x)g(x)dx or, when using a weight function p(x), by: (f,g)=\int\limits_a^bp(x)f(x)g(x)dx The norm \|f\| follows from: \|f\|^2=(f,f). A set functions f_i is orthonormal if (f_i,f_j)=\delta_{ij}.

Each function f(x) can be written as a sum of orthogonal functions: f(x)=\sum_{i=0}^\infty c_ig_i(x) and \sum c_i^2\leq\|f\|^2. Let the set g_i be orthogonal, than it follows: c_i=\frac{(f,g_i)}{(g_i,g_i)}

3.4 Fourier series

Each function can be written as a sum of independent base functions. When one chooses the orthogonal basis (\cos(nx),\sin(nx)) we have a Fourier series.

A periodical function f(x) with period 2L can be written as: f(x)=a_0+\sum_{n=1}^\infty\left[a_n\cos\left(\frac{n\pi x}{L}\right)+b_n\sin\left(\frac{n\pi x}{L}\right)\right] Due to the orthogonality follows for the coefficients: a_0=\frac{1}{2L}\int\limits_{-L}^Lf(t)dt~~,~~ a_n=\frac{1}{L}\int\limits_{-L}^Lf(t)\cos\left(\frac{n\pi t}{L}\right)dt~~,~~ b_n=\frac{1}{L}\int\limits_{-L}^Lf(t)\sin\left(\frac{n\pi t}{L}\right)dt A Fourier series can also be written as a sum of complex exponents: f(x)=\sum_{n=-\infty}^\infty c_n{\rm e}^{inx} with c_n=\frac{1}{2\pi}\int\limits_{-\pi}^\pi f(x){\rm e}^{-inx}dx The Fourier transform of a function f(x) gives the transformed function \hat{f}(\omega): \hat{f}(\omega)=\frac{1}{\sqrt{2\pi}}\int\limits_{-\infty}^\infty f(x){\rm e}^{-i\omega x}dx The inverse transformation is given by: \frac{1}{2}\left[f(x^+)+f(x^-)\right]=\frac{1}{\sqrt{2\pi}}\int\limits_{-\infty}^\infty\hat{f}(\omega){\rm e}^{i\omega x}d\omega where f(x^+) and f(x^-) are defined by the lower - and upper limit: f(a^-)=\lim_{x\uparrow a}f(x)~~,~~f(a^+)=\lim_{x\downarrow a}f(x) For continuous functions is \frac{1}{2}\left[f(x^+)+f(x^-)\right]=f(x).