通过对 R\mathbb R 上的初等函数进行原点处的 Taylor 展开,即 Maclaurin 展开,可以自然地将其推广到复数域 C\mathbb C,并以幂级数的形式存在

# 指数函数

定义
对于 zCz \in \mathbb C,称

exp(z)=ez=n=0znn!\exp(z) = e^z = \sum_{n=0}^{\infty} \frac{z^n}{n!}

复指数函数

根据 D'Alembert 判别法

zn+1(n+1)!/znn!=zn+10(n)\left| \frac{z^{n+1}}{(n+1)!} \bigg / \frac{z^n}{n!} \right| = \frac{|z|}{n+1} \to 0\quad (n \to \infty)

可知级数全域绝对收敛,确保定义域

指数函数具有如下基本性质

命题 指数函数基本性质
以下 z,wC,tRz,w \in \mathbb C,\ t \in \mathbb R

  1. ez+w=ezewe^{z+w} = e^z \cdot e^w
  2. Re(eit)=cost,Im(eit)=sint\mathrm{Re}(e^{it}) = \cos t,\quad \mathrm{Im}(e^{it}) = \sin t
  3. ez=ez\overline{e^z} = e^{\overline z} \quad
  4. eit=1|e^{it}| = 1
  5. ez=eRez>0|e^z| = e^{\mathrm{Re} z} > 0
  6. ez=ewz=w+2kπi,kZe^z = e^w \iff z = w + 2k\pi i,\ k \in \mathbb Z
证明

(1)
注意,对于两个绝对收敛的级数,有

n=0anm=0bm=n=0(k=0nakbnk)\sum_{n=0}^{\infty} a_n \cdot \sum_{m=0}^{\infty} b_m = \sum_{n=0}^{\infty} \left( \sum_{k=0}^n a_k b_{n-k} \right)

所以

ez+w=n=0(z+w)nn!=n=0(k=0nzkk!wnk(nk)!)=ezewe^{z+w} = \sum_{n=0}^{\infty} \frac{(z+w)^n}{n!} = \sum_{n=0}^{\infty} \left( \sum_{k=0}^n \frac{z^k}{k!} \cdot \frac{w^{n-k}}{(n-k)!} \right) = e^z \cdot e^w

(2)

eit=n=0(it)nn!=n=0(1)nt2n(2n)!+in=0(1)nt2n+1(2n+1)!=cost+isinte^{it} = \sum_{n=0}^{\infty} \frac{(it)^n}{n!} = \sum_{n=0}^{\infty} (-1)^n \frac{t^{2n}}{(2n)!} + i \sum_{n=0}^{\infty} (-1)^n \frac{t^{2n+1}}{(2n+1)!} = \cos t + i \sin t

(3)
只需证明 eit=eit\overline{e^{it}} = e^{-it} \quad

eit=cost+isint=costisint=cos(t)+isin(t)=eit\overline{e^{it}} = \overline{\cos t + i \sin t} = \cos t - i \sin t = \cos(-t) + i \sin(-t) = e^{-it}

(4)

eit=cos2t+sin2t=1|e^{it}| = \sqrt{\cos^2 t + \sin^2 t} = 1

(5)
z=x+iyz = x + iy

ez=ex+iy=exeiy=ex1=eRez>0|e^z| = |e^{x+iy}| = |e^x| \cdot |e^{iy}| = e^x \cdot 1 = e^{\mathrm{Re} z} > 0

(6)(\Rightarrow)
ezw=ezew=ewew=1e^{z-w} = e^z e^{-w} = e^w e^{-w} = 1

1=ezw=eRe(zw)Re(zw)=01 = |e^{z-w}| = e^{\mathrm{Re}(z-w)} \implies \mathrm{Re}(z-w) = 0

zw=it,tRz-w = it,\ t \in \mathbb R,对比 ezw=1e^{z-w} = 1 的虚部得

0=Im(ezw)=Im(eit)=sintt=2kπ,kZ0 = \mathrm{Im}(e^{z-w}) = \mathrm{Im}(e^{it}) = \sin t \implies t = 2k\pi,\ k \in \mathbb Z

(\Leftarrow)
z=w+2kπi,kZz = w + 2k\pi i,\ k \in \mathbb Z,则

ez=ew+2kπi=ewe2kπi=ew1=ewe^z = e^{w + 2k\pi i} = e^w \cdot e^{2k\pi i} = e^w \cdot 1 = e^w

由此,任意复数可以转为极坐标形式

z=reiθ,r=z,θ=argzz = re^{i\theta},\quad r = |z|,\ \theta = \arg z

命题
指数函数全纯,且

ddzez=ez\frac{d}{dz} e^z = e^z

证明

绝对收敛确保逐项可微,所以

ddzez=ddzn=0znn!=n=1nzn1n!=n=0znn!=ez\frac{d}{dz} e^z = \frac{d}{dz} \sum_{n=0}^{\infty} \frac{z^n}{n!} = \sum_{n=1}^{\infty} \frac{n z^{n-1}}{n!} = \sum_{n=0}^{\infty} \frac{z^n}{n!} = e^z

# 三角函数

定义
对于 zCz \in \mathbb C,称

cosz=n=0(1)nz2n(2n)!,sinz=n=0(1)nz2n+1(2n+1)!\cos z = \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n}}{(2n)!},\quad \sin z = \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n+1}}{(2n+1)!}

复三角函数

根据 D'Alembert 判别法

(1)n+1z2n+2/(2n+2)!(1)nz2n/(2n)!=z2(2n+2)(2n+1)0(n)\frac{|(-1)^{n+1} z^{2n+2}/(2n+2)!|}{|(-1)^n z^{2n}/(2n)!|} = \frac{|z|^2}{(2n+2)(2n+1)} \to 0\quad (n \to \infty)

(1)n+1z2n+3/(2n+3)!(1)nz2n+1/(2n+1)!=z2(2n+3)(2n+2)0(n)\frac{|(-1)^{n+1} z^{2n+3}/(2n+3)!|}{|(-1)^n z^{2n+1}/(2n+1)!|} = \frac{|z|^2}{(2n+3)(2n+2)} \to 0\quad (n \to \infty)

得到两级数全域绝对收敛

指数函数与三角函数之间可以转化

命题
zCz \in \mathbb C

cosz=eiz+eiz2,sinz=eizeiz2i\cos z = \frac{e^{iz} + e^{-iz}}{2},\quad \sin z = \frac{e^{iz} - e^{-iz}}{2i}

证明

直接计算验证

eiz+eiz2=12n=0(iz)n+(iz)nn!=n=0(1)nz2n(2n)!=cosz\frac{e^{iz} + e^{-iz}}{2} = \frac{1}{2} \sum_{n=0}^{\infty} \frac{(iz)^n + (-iz)^n}{n!} = \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n}}{(2n)!} = \cos z

eizeiz2i=12in=0(iz)n(iz)nn!=n=0(1)nz2n+1(2n+1)!=sinz\frac{e^{iz} - e^{-iz}}{2i} = \frac{1}{2i} \sum_{n=0}^{\infty} \frac{(iz)^n - (-iz)^n}{n!} = \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n+1}}{(2n+1)!} = \sin z

特别地,有以下著名的 Euler 公式:eiπ+1=0e^{i\pi} + 1 = 0

定理 Euler 公式
对于 zCz \in \mathbb C

eiz=cosz+isinze^{iz} = \cos z + i \sin z

证明

由上命题直接得出

eiz=eiz+eiz2+eizeiz2ii=cosz+isinze^{iz} = \frac{e^{iz} + e^{-iz}}{2} + \frac{e^{iz} - e^{-iz}}{2i} \cdot i = \cos z + i \sin z

命题 三角函数基本性质
以下 zCz \in \mathbb C

  1. cos2z+sin2z=1\cos^2 z + \sin^2 z = 1
  2. cos(z)=cosz,sin(z)=sinz\cos(-z) = \cos z,\quad \sin(-z) = -\sin z
  3. cos(z+w)=coszcoswsinzsinw\cos(z+w) = \cos z \cos w - \sin z \sin w
  4. sin(z+w)=sinzcosw+coszsinw\sin(z+w) = \sin z \cos w + \cos z \sin w
  5. cos(z+2π)=cosz,sin(z+2π)=sinz\cos(z + 2\pi) = \cos z,\quad \sin(z + 2\pi) = \sin z
证明

(1)

cos2z+sin2z=(eiz+eiz2)2+(eizeiz2i)2=e2iz+2+e2iz4+e2iz+2e2iz4=1\cos^2 z + \sin^2 z = \left( \frac{e^{iz} + e^{-iz}}{2} \right)^2 + \left( \frac{e^{iz} - e^{-iz}}{2i} \right)^2 = \frac{e^{2iz} + 2 + e^{-2iz}}{4} + \frac{-e^{2iz} + 2 - e^{-2iz}}{4} = 1

(2)

cos(z)=eiz+eiz2=eiz+eiz2=cosz,sin(z)=eizeiz2i=eizeiz2i=sinz\cos(-z) = \frac{e^{-iz} + e^{iz}}{2} = \frac{e^{iz} + e^{-iz}}{2} = \cos z,\quad \sin(-z) = \frac{e^{-iz} - e^{iz}}{2i} = -\frac{e^{iz} - e^{-iz}}{2i} = -\sin z

(3)

cos(z+w)=ei(z+w)+ei(z+w)2=eizeiw+eizeiw2=(eiz+eiz)(eiw+eiw)(eizeiz)(eiweiw)4=coszcoswsinzsinw\cos(z+w) = \frac{e^{i(z+w)} + e^{-i(z+w)}}{2} = \frac{e^{iz}e^{iw} + e^{-iz}e^{-iw}}{2} = \frac{(e^{iz} + e^{-iz})(e^{iw} + e^{-iw}) - (e^{iz} - e^{-iz})(e^{iw} - e^{-iw})}{4} = \cos z \cos w - \sin z \sin w

(4)

sin(z+w)=ei(z+w)ei(z+w)2i=eizeiweizeiw2i=(eiz+eiz)(eiweiw)+(eizeiz)(eiw+eiw)4i=sinzcosw+coszsinw\sin(z+w) = \frac{e^{i(z+w)} - e^{-i(z+w)}}{2i} = \frac{e^{iz}e^{iw} - e^{-iz}e^{-iw}}{2i} = \frac{(e^{iz} + e^{-iz})(e^{iw} - e^{-iw}) + (e^{iz} - e^{-iz})(e^{iw} + e^{-iw})}{4i} = \sin z \cos w + \cos z \sin w

(5)

cos(z+2π)=ei(z+2π)+ei(z+2π)2=eize2πi+eize2πi2=eiz+eiz2=cosz\cos(z + 2\pi) = \frac{e^{i(z + 2\pi)} + e^{-i(z + 2\pi)}}{2} = \frac{e^{iz}e^{2\pi i} + e^{-iz}e^{-2\pi i}}{2} = \frac{e^{iz} + e^{-iz}}{2} = \cos z

sin(z+2π)=ei(z+2π)ei(z+2π)2i=eize2πieize2πi2i=eizeiz2i=sinz\sin(z + 2\pi) = \frac{e^{i(z + 2\pi)} - e^{-i(z + 2\pi)}}{2i} = \frac{e^{iz}e^{2\pi i} - e^{-iz}e^{-2\pi i}}{2i} = \frac{e^{iz} - e^{-iz}}{2i} = \sin z

命题
三角函数全纯且

ddzcosz=sinz,ddzsinz=cosz\frac{d}{dz} \cos z = -\sin z,\quad \frac{d}{dz} \sin z = \cos z

证明

绝对收敛确保逐项可微,所以

ddzcosz=ddzn=0(1)nz2n(2n)!=n=1(1)n2nz2n1(2n)!=n=0(1)nz2n+1(2n+1)!=sinz\frac{d}{dz} \cos z = \frac{d}{dz} \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n}}{(2n)!} = \sum_{n=1}^{\infty} (-1)^n \frac{2n z^{2n-1}}{(2n)!} = -\sum_{n=0}^{\infty} (-1)^n \frac{z^{2n+1}}{(2n+1)!} = -\sin z

ddzsinz=ddzn=0(1)nz2n+1(2n+1)!=n=0(1)n(2n+1)z2n(2n+1)!=n=0(1)nz2n(2n)!=cosz\frac{d}{dz} \sin z = \frac{d}{dz} \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n+1}}{(2n+1)!} = \sum_{n=0}^{\infty} (-1)^n \frac{(2n+1) z^{2n}}{(2n+1)!} = \sum_{n=0}^{\infty} (-1)^n \frac{z^{2n}}{(2n)!} = \cos z

特别提醒:实数三角函数的取值范围只在 [1,1][-1,1] 之间,而复数三角函数的取值范围为整个复平面,并非有界

示例

  • cos(it)=et+et2(t+)\cos(it) = \dfrac{e^{-t} + e^{t}}{2} \to \infty \quad (t \to +\infty)
  • Im(sin(it))=etet2(t+)\mathrm{Im}(\sin(it)) = \dfrac{e^{t} - e^{-t}}{2} \to \infty \quad (t \to +\infty)

# 双曲函数

定义
对于 zCz \in \mathbb C,称

coshz=n=0z2n(2n)!,sinhz=n=0z2n+1(2n+1)!\cosh z = \sum_{n=0}^{\infty} \frac{z^{2n}}{(2n)!},\quad \sinh z = \sum_{n=0}^{\infty} \frac{z^{2n+1}}{(2n+1)!}

复双曲函数

根据 D'Alembert 判别法

z2n+2/(2n+2)!z2n/(2n)!=z2(2n+2)(2n+1)0(n)\frac{|z^{2n+2}/(2n+2)!|}{|z^{2n}/(2n)!|} = \frac{|z|^2}{(2n+2)(2n+1)} \to 0\quad (n \to \infty)

z2n+3/(2n+3)!z2n+1/(2n+1)!=z2(2n+3)(2n+2)0(n)\frac{|z^{2n+3}/(2n+3)!|}{|z^{2n+1}/(2n+1)!|} = \frac{|z|^2}{(2n+3)(2n+2)} \to 0\quad (n \to \infty)

得到两级数全域绝对收敛

同样,双曲函数可以转为指数函数

命题
zCz \in \mathbb C

coshz=ez+ez2,sinhz=ezez2\cosh z = \frac{e^{z} + e^{-z}}{2},\quad \sinh z = \frac{e^{z} - e^{-z}}{2}

证明

ez+ez2=12n=0zn+(z)nn!=n=0z2n(2n)!=coshz\frac{e^{z} + e^{-z}}{2} = \frac{1}{2} \sum_{n=0}^{\infty} \frac{z^n + (-z)^n}{n!} = \sum_{n=0}^{\infty} \frac{z^{2n}}{(2n)!} = \cosh z

ezez2=12n=0zn(z)nn!=n=0z2n+1(2n+1)!=sinhz\frac{e^{z} - e^{-z}}{2} = \frac{1}{2} \sum_{n=0}^{\infty} \frac{z^n - (-z)^n}{n!} = \sum_{n=0}^{\infty} \frac{z^{2n+1}}{(2n+1)!} = \sinh z

双曲函数将继承指数函数的计算性质

命题 双曲函数基本性质
以下 z,wCz,w \in \mathbb C

  1. cosh2zsinh2z=1\cosh^2 z - \sinh^2 z = 1
  2. cosh(iz)=cosz,cos(iz)=icoshz\cosh(iz) = \cos z,\quad \cos(iz) = i \cosh z
  3. sinh(iz)=isinz,sin(iz)=isinhz\sinh(iz) = i \sin z,\quad \sin(iz) = i \sinh z
  4. cosh(x+iy)=coshxcosy+isinhxsiny,x,yR\cosh(x+iy) = \cosh x \cos y + i \sinh x \sin y,\quad x,y \in \mathbb R
  5. sinh(x+iy)=sinhxcosy+icoshxsiny,x,yR\sinh(x+iy) = \sinh x \cos y + i \cosh x \sin y,\quad x,y \in \mathbb R
证明

(1)

cosh2zsinh2z=(ez+ez2)2(ezez2)2=e2z+2+e2z4e2z2+e2z4=1\cosh^2 z - \sinh^2 z = \left( \frac{e^{z} + e^{-z}}{2} \right)^2 - \left( \frac{e^{z} - e^{-z}}{2} \right)^2 = \frac{e^{2z} + 2 + e^{-2z}}{4} - \frac{e^{2z} - 2 + e^{-2z}}{4} = 1

(2)

cosh(iz)=eiz+eiz2=cosz,cos(iz)=eizeiz2i=iezez2=isinhz\cosh(iz) = \frac{e^{iz} + e^{-iz}}{2} = \cos z,\quad \cos(iz) = \frac{e^{iz} - e^{-iz}}{2i} = i \cdot \frac{e^{z} - e^{-z}}{2} = i \sinh z

(3)

sinh(iz)=eizeiz2=ieizeiz2i=isinz,sin(iz)=eizeiz2i=iezez2=isinhz\sinh(iz) = \frac{e^{iz} - e^{-iz}}{2} = i \cdot \frac{e^{iz} - e^{-iz}}{2i} = i \sin z,\quad \sin(iz) = \frac{e^{iz} - e^{-iz}}{2i} = i \cdot \frac{e^{z} - e^{-z}}{2} = i \sinh z

(4)

cosh(x+iy)=ex+iy+e(x+iy)2=exeiy+exeiy2=(ex+ex)(cosy+isiny)+(exex)(cosyisiny)4=coshxcosy+isinhxsiny\cosh(x+iy) = \frac{e^{x+iy} + e^{-(x+iy)}}{2} = \frac{e^x e^{iy} + e^{-x} e^{-iy}}{2} = \frac{(e^x + e^{-x})(\cos y + i \sin y) + (e^x - e^{-x})(\cos y - i \sin y)}{4} = \cosh x \cos y + i \sinh x \sin y

(5)

sinh(x+iy)=ex+iye(x+iy)2=exeiyexeiy2=(exex)(cosy+isiny)+(ex+ex)(cosyisiny)4=sinhxcosy+icoshxsiny\sinh(x+iy) = \frac{e^{x+iy} - e^{-(x+iy)}}{2} = \frac{e^x e^{iy} - e^{-x} e^{-iy}}{2} = \frac{(e^x - e^{-x})(\cos y + i \sin y) + (e^x + e^{-x})(\cos y - i \sin y)}{4} = \sinh x \cos y + i \cosh x \sin y

命题
双曲函数全纯且

ddzsinhz=coshz,ddzcoshz=sinhz\frac{d}{dz} \sinh z = \cosh z,\quad \frac{d}{dz} \cosh z = \sinh z

证明

绝对收敛确保逐项可微,所以

ddzsinhz=ddzn=0z2n+1(2n+1)!=n=0(2n+1)z2n(2n+1)!=n=0z2n(2n)!=coshz\frac{d}{dz} \sinh z = \frac{d}{dz} \sum_{n=0}^{\infty} \frac{z^{2n+1}}{(2n+1)!} = \sum_{n=0}^{\infty} \frac{(2n+1) z^{2n}}{(2n+1)!} = \sum_{n=0}^{\infty} \frac{z^{2n}}{(2n)!} = \cosh z

ddzcoshz=ddzn=0z2n(2n)!=n=12nz2n1(2n)!=n=0z2n+1(2n+1)!=sinhz\frac{d}{dz} \cosh z = \frac{d}{dz} \sum_{n=0}^{\infty} \frac{z^{2n}}{(2n)!} = \sum_{n=1}^{\infty} \frac{2n z^{2n-1}}{(2n)!} = \sum_{n=0}^{\infty} \frac{z^{2n+1}}{(2n+1)!} = \sinh z

# 对数函数

以下为了区分,将实对数函数记为 lnx\ln x,复对数函数记为 logz\log z
作为指数函数的反函数,通过取满足 ew=ze^w = zww 作为 logz\log z,可以自然地定义对数函数
注意 z = |z| e^

定义
对于 z0z \neq 0,称

logz=lnz+iargz\log z = \ln |z| + i \arg z

复对数函数
并且,称

Logz=lnz+iArg\mathrm{Log} z = \ln |z| + i \mathrm{Arg}

为其 主值

注意:对数的运算法则

log(zw)=logz+logw\log (zw) = \log z + \log w

作为集合成立,但是不一定对主值 Log\mathrm{Log} 成立


复对数函数并非如前几个初等函数一样依靠级数定义,所以其连续性和可微性都需要单独讨论

由于继承了偏角的多值性,在每次旋转到 2π2\pi 的位置,一定会发生一个 2π2 \pi 的跳跃。这导致哪怕只考虑主值的 Log\mathrm{Log},在复平面上也并非是连续

不妨如下考虑这一点:取一点 xRCx \in \mathbb R \subset \mathbb C
让偏角主值函数以两个不同的方向趋于 xx
当从上方趋近时,由于主值的范围是 (π,π](-\pi, \pi],所以 Arg(x)=π\mathrm{Arg}(x) = \pi
当从下方趋近时,Arg(x)=π\mathrm{Arg}(x) = -\pi,所以

  • Log(x+iϵ)=lnx+iArg(x+iϵ)lnx+iπ(ϵ0+)Log(x + i\epsilon) = \ln |x| + i Arg(x + i\epsilon) \to \ln |x| + i \pi \quad (\epsilon \to 0^+)
  • Log(xiϵ)=lnx+iArg(xiϵ)lnxiπ(ϵ0+)Log(x - i\epsilon) = \ln |x| + i Arg(x - i\epsilon) \to \ln |x| - i \pi \quad (\epsilon \to 0^+)

显然可以看出二者差了 2πi2\pi i,所以 Log\mathrm{Log}xx 处不连续

但是实际上,通过对定义域进行适当的限制(切割射线),可以使得对数函数连续且可微
令一个去除掉角度为 θ\theta 的射线的区域为

Dθ={reiϕC×r>0,θ<ϕ<θ+2π}D_\theta = \{ re^{i\phi} \in \mathbb C^\times \mid r > 0,\ \theta < \phi < \theta + 2\pi \}

命题
logz\log zDθD_\theta 上连续,可微,且

ddz(logzDθ)=1z\frac{d}{dz} (\log z \bigg |_{D_\theta}) = \frac{1}{z}

证明

(连续性)
z0Dθz_0 \in D_\theta,令 θ0=argz0(θ,θ+2π),r0=z0\theta_0 = \arg z_0 \in (\theta, \theta + 2\pi),\ r_0 = |z_0|
注意事实:实数三角函数的反函数在其定义域内连续
此时

cosθ0=Rez0r0,sinθ0=Imz0r0\cos \theta_0 = \frac{\mathrm{Re} z_0}{r_0},\quad \sin \theta_0 = \frac{\mathrm{Im} z_0}{r_0}

所以

θ0=arccosRez0r0=arcsinImz0r0\theta_0 = \arccos \frac{\mathrm{Re} z_0}{r_0} = \arcsin \frac{\mathrm{Im} z_0}{r_0}

argDθ(z)=arccosRezzcontinuousarccosRez0r0=θ0(zz0)\arg \bigg |_{D_\theta} (z) = \underbrace{\arccos \frac{\mathrm{Re} z}{|z|}}_{\text{continuous}} \to \arccos \frac{\mathrm{Re} z_0}{r_0} = \theta_0 \quad (z \to z_0)

(可微性)
先证明以下引理:对于 w0Cw_0 \in \mathbb C

Ew0(w):={ew0w=w0ewew0ww0ww0E_{w_0}(w) := \begin{cases} e^{w_0} & w = w_0 \\[4pt] \dfrac{e^w - e^{w_0}}{w - w_0} & w \neq w_0 \end{cases}

C\mathbb C 上连续,证明如下

limww0Ew0(w)=limww0ewew0ww0=ddweww=w0=ew0=Ew0(w0)\lim_{w \to w_0} E_{w_0}(w) = \lim_{w \to w_0} \frac{e^w - e^{w_0}}{w - w_0} = \left. \frac{d}{dw} e^w \right|_{w=w_0} = e^{w_0} = E_{w_0}(w_0)

根据此引理
接下来,取 f=logDθf = \log |_{D_\theta},则对于固定的 z0Dθz_0 \in D_\theta

f(z0+h)f(z0)h=f(z0+h)exp(f(z0+h))exp(f(z0))=1Ef(z0)(f(z0+h))\frac{f(z_0 + h) - f(z_0)}{h} = \frac{f(z_0 + h)}{\exp(f(z_0 + h)) - \exp(f(z_0))} = \frac{1}{E_{f(z_0)}(f(z_0 + h))}

由于 f,Ef(z0)f, E_{f(z_0)} 均在各自定义域内连续,所以

Ef(z0)(f(z0+h))Ef(z0)(f(z0))=ef(z0)=z0(h0)E_{f(z_0)}(f(z_0 + h)) \to E_{f(z_0)}(f(z_0)) = e^{f(z_0)} = z_0 \quad (h \to 0)

从而

f(z0)=limh0f(z0+h)f(z0)h=limh01Ef(z0)(f(z0+h))=1z0f'(z_0) = \lim_{h \to 0} \frac{f(z_0 + h) - f(z_0)}{h} = \lim_{h \to 0} \frac{1}{E_{f(z_0)}(f(z_0 + h))} = \frac{1}{z_0}

特别地,对于主值函数,定义域的限制可以放宽为 C(,0]\mathbb C \setminus (-\infty,0],即去除负实轴

命题
Logz\mathrm{Log} zC(,0]\mathbb C \setminus (-\infty,0] 上连续,可微,且

ddz(Logz)=1z\frac{d}{dz} (\mathrm{Log} z) = \frac{1}{z}

证明

直接套用上命题的结论即可,略


虽然对数函数本身不依靠级数定义,但其可微性给出级数展开可能性,规避掉负实轴

命题

Log(1+z)=n=1(1)n1znn,zD(0,1)\mathrm{Log}(1+z) = \sum_{n=1}^{\infty} (-1)^{n-1} \frac{z^n}{n},\quad z \in D(0,1)

证明

f(z)=n=1(1)n1znn,zD(0,1)f(z) = \displaystyle\sum_{n=1}^{\infty} (-1)^{n-1} \dfrac{z^n}{n}, \ z \in D(0,1),系数 an=(1)n1na_n = \dfrac{(-1)^{n-1}}{n},需证 Log(1+z)=f(z)\mathrm{Log}(1+z) = f(z)

由于 ffD(0,1)D(0,1) 上可微,且微分系数由等比级数给出

f(z)=n=1(1)n1zn1=n=1(z)n1=11+zf'(z) = \sum_{n=1}^{\infty} (-1)^{n-1} z^{n-1} = \sum_{n=1}^{\infty} (-z)^{n-1} = \frac{1}{1+z}

另一边,1+z1+zD(0,1)D(0,1) 可微,由链式法则也可以给出

ddz(Log(1+z))=11+z\frac{d}{dz} (\mathrm{Log}(1+z)) = \frac{1}{1+z}

联合得到

ddz(Log(1+z)f(z))=0,zD(0,1)\frac{d}{dz} (\mathrm{Log}(1+z) - f(z)) = 0,\quad z \in D(0,1)

由于 D(0,1)D(0,1) 为领域,且微分系数恒为零,由正则性章节的结论得到

Log(1+z)f(z)=C,zD(0,1)\mathrm{Log}(1+z) - f(z) = C,\quad z \in D(0,1)

z=0z=0,得到 C=0C=0,所以

Log(1+z)=f(z)=n=1(1)n1znn,zD(0,1)\mathrm{Log}(1+z) = f(z) = \sum_{n=1}^{\infty} (-1)^{n-1} \frac{z^n}{n},\quad z \in D(0,1)

# 幂函数

幂函数有两种

  • 底数为变量的幂函数 f(z)=zaf(z) = z^a
  • 指数为变量的幂函数 g(z)=azg(z) = a^z

在实数域上幂函数满足 xa=ealnxx^a = e^{a \ln x},借由已经定义的对数函数,可以推广如下

定义
对于 zC×,aCz \in \mathbb C^\times,\ a \in \mathbb C,称

za=ealogzz^a = e^{a \log z}

复幂函数
并且,称

eaLogze^{a \mathrm{Log} z}

为其主值

指数函数的多值性继承于对数函数
但是如果 aZa \in \mathbb Z,则幂函数变为单纯的多项式,成为单值函数,这很好验证:

enlogz=en(lnz+iargz)=enlnzeinargz=znei(argz+2kπn)=zneiargz=zn,kZe^{n \log z} = e^{n (\ln |z| + i \arg z)} = e^{n \ln |z|} \cdot e^{i n \arg z} = |z|^n \cdot e^{i (\arg z + 2k\pi n)} = |z|^n \cdot e^{i \arg z} = z^n,\quad k \in \mathbb Z

反过来也可以定义另外一个形式的幂函数

az=ezlogaa^z = e^{z \log a}

虽然这个函数也是多值,但仅当 a=ea=e,时,此函数成为单纯的指数函数,规定仅取主值,也就是说

ez=ezLogee^z = e^{z \mathrm{Log} e}


由于对数函数的可微性需要限制幅角的变化量不能超过 2π2\pi,所以幂函数可微性也依赖于相同的限制

命题
对于 aC,zDθa \in \mathbb C, z \in D_\theta

ddz(zaDθ)=aza1\frac{d}{dz} (z^a \bigg |_{D_\theta}) = a z^{a-1}

证明

f=logDθf = \log |_{D_\theta} \quad

ddz(zaDθ)=ddz(eaf(z))=eaf(z)af(z)=zaa1z=aza1\frac{d}{dz} (z^a \bigg |_{D_\theta}) = \frac{d}{dz} (e^{a f(z)}) = e^{a f(z)} \cdot a f'(z) = z^a \cdot a \cdot \frac{1}{z} = a z^{a-1}

命题
对于 aDθa \in D_\theta

ddz(az)=azloga\frac{d}{dz} (a^z) = a^z \log a

证明

=logDθ(a)C\ell = \log |_{D_\theta}(a) \in \mathbb C,则 az=eza^z = e^{\ell z} \quad

ddz(az)=ddz(ez)=ez=azlogaDθ\frac{d}{dz} (a^z) = \frac{d}{dz} (e^{\ell z}) = e^{\ell z} \cdot \ell = a^z \log a \bigg |_{D_\theta}


并且,通过推广组合数,对于 αC\alpha \in \mathbb C

αCn=(αk)=α(α1)(αk+1)k!{}_\alpha C_n = \binom{\alpha}{k} = \frac{\alpha (\alpha - 1) \cdots (\alpha - k + 1)}{k!}

从而可以对 (1+z)α(1+z)^\alpha主值 进行二项式展开

(1+z)α=k=0(αk)zk,zD(0,1)(1+z)^\alpha = \sum_{k=0}^\infty \binom{\alpha}{k} z^k,\quad z \in D(0,1)

# 1 的 n 次根

满足 wn=1w^n = 1 的复数 ww 称为 1 的 n 次根
问题在于解具体是什么

命题
1 的 n 次根共有 n 个,分别为

wk=e2kπi/n,k=0,1,2,,n1w_k = e^{2k\pi i/n},\quad k = 0, 1, 2, \ldots, n-1

证明

由定义,ww 满足

wn=1enlogw=1nlogw=2kπi,kZw^n = 1 \implies e^{n \log w} = 1 \implies n \log w = 2k\pi i,\quad k \in \mathbb Z

所以

logw=2kπinw=e2kπi/n,kZ\log w = \frac{2k\pi i}{n} \implies w = e^{2k\pi i/n},\quad k \in \mathbb Z

由于 e2(k+n)πi/n=e2kπi/ne2πi=e2kπi/ne^{2(k+n)\pi i/n} = e^{2k\pi i/n} \cdot e^{2\pi i} = e^{2k\pi i/n},所以只需取 k=0,1,2,,n1k = 0, 1, 2, \ldots, n-1 即可得到所有不同的解,共计 n 个

事实上,1 的 n 次根是均匀分布在单位圆上的 n 个点,称为 n 次单位根,并且它们构成一个乘法群,记为 μn\mu_n,即

μn={e2kπi/nk=0,1,2,,n1}\mu_n = \{ e^{2k\pi i/n} \mid k = 0, 1, 2, \ldots, n-1 \}

这个群在复平面上对应着一个正 n 边形的顶点集合,且该群是循环群,由 e2πi/ne^{2\pi i/n} 生成

μn=e2πi/n\mu_n = \langle e^{2\pi i/n} \rangle

对于任意复数 α\alpha,方程 wn=αw^n = \alpha 可以化归回 1 的 n 次根的问题
取极坐标表示

α=reiθ,r>0,θR\alpha = r e^{i \theta},\quad r > 0,\ \theta \in \mathbb R

wn=αwn=reiθ(wr1/neiθ/n)n=1w^n = \alpha \implies w^n = r e^{i \theta} \implies \left( \frac{w}{r^{1/n} e^{i \theta/n}} \right)^n = 1

所以 ww 的解为

wk=r1/nei(θ+2kπ)/n,k=0,1,2,,n1w_k = r^{1/n} e^{i (\theta + 2k\pi)/n},\quad k = 0, 1, 2, \ldots, n-1