Planeamento

Aulas Teóricas

Números Complexos

 

Recorde-se que os números reais são números racionais (números com uma expansão decimal finita ou periódica) juntamete com os números irracionais (números que tem uma expansão decimal infinita e não periódica).  Designa-se por \(\mathbb{R}\) o conjunto dos números reais.

Os números complexos é o conjunto \(\mathbb{C} = \mathbb{R}^2\). Define-se

  • adição por \( (x,y) + (u,v) = (x+u,y+v) \) 
  • multiplicação por \( (x,y)\cdot(u,v) = (xu-yv,xv+yu) \)

Esta operações são comutativas e associativas.

Observe-se que  \begin{align} (1,0)\cdot(x,y) &= (x,y)\\ (0,1)\cdot(0,1) &= (-1,0) \end{align} 

 

Designa-se \(1=(1,0)\), \(i=(0,1)\) e \((x,y) = x + i y\). Segue-se que a polinómio \(t^2+1\) tem uma raiz em \(\mathbb{C}\); nomeadamente \(t=i\). Mais tarde vamos ver que qualquer polinómio, com coeficientes complexas, tem uma raiz em \(\mathbb{C}\).


Parte real, parte imaginária e conjugação

Seja \(z=x+iy\) um número complexo.

  • A parte real de \(z\) é \( \def\re{\operatorname{Re}} \def\im{\operatorname{Im}}\re z = x\)
  • A parte imaginária de \(z\) é \(\im z = y\).
  • A conjugada de \(z\) é \(\overline{z} = x - i y\). 

Propriedades:

  • \(\re (z+w) = \re z + \re w\)
  • \(\im (z+w) = \im z + \im w\)
  • \(\overline{z+w} = \overline{z}+\overline{w}\)
  • \(\overline{z\cdot w} = \overline{z}\cdot\overline{w}\)


Módulo


Define-se o módulo de \(z=x+iy\) por \[ |z| = \sqrt{x^2 + y^2}.\]

Observe-se que \(|z|^2 = z\cdot\overline{z}\).

Propriedades:

  • \(|z\cdot w| = |z|\cdot|w|\)
  • \(|z+w| \leq |z| + |w|\) Desigualdade triangular
  • \(| \, |z| - |w|\,|\ \leq  |z-w|\)
  • \(|\overline{z}| = |z|\)

Topologia do plano complexo


Disco aberto

Para um ponto \(p\in\mathbb{C}\) e um número \(r>0\), designa-se por \[D(p,r) =\{z\in\mathbb{C} \colon |z-p|< r\}\] e chama-se disco aberto com centro \(p\) e raio \(r\).


Disco fechado

O disco fechado com centro \(p\) e raio \(r\) é o conjunto \[\overline{D}(p,r) =\{z\in\mathbb{C}\colon |z-p| \leq r\}.\]

Conjuntos abertos

Diz-se que um subconjunto \(U \subseteq \mathbb{C}\) é aberto quando para cada ponto \(p\in U\) existe um número \(r>0\) tal que \[D(p,r) \subseteq U.\]

Conjuntos fechados

Diz-se que um subconjunto \(A\subseteq \mathbb{C}\) é fechado quando \(\mathbb{C}\setminus A\) é aberto. 

Note-se que existem subconjuntos de \(\mathbb{C}\) nem abertos nem fechados.

Conjuntos limitados

Diz-se que um subconjunto \(U\subseteq \mathbb{C}\) é limitado quando existe \(M>0\) tal que \(U\subseteq D(0,M)\).  Isto é para qualquer \(z\in U\) tem-se \(|z| <M\). 

Um subconjunto chama-se compacto quando é fechado e limitado. 

Curvas

Para números \(a<b\), considere o intervalo fechado \([a,b]\).  Um função \(\gamma\colon [a,b]\rightarrow \mathbb{C}\) pode ser vista como uma função \( (\gamma_1,\gamma_2)\colon[a,b]\rightarrow \mathbb{R}^2\) 

Diz-se que \[\gamma(t) = \gamma_1(t) + i \gamma_2(t) \quad \text{ou}\quad \bigl(\gamma_1(t),\gamma_2(t)\bigr) \]

é contínua, diferenciável, ou regular por partes quando \(\bigl(\gamma_1(t),\gamma_2(t)\bigr)\) é contínua, diferenciável, ou regular por partes respectivamente.  Diz-se que \(\gamma(t)\) é fechado quando $\(\gamma(a)=\gamma(b)\), e diz-se que \(\gamma(t)\) é uma  curva de Jordan quando é contínua, fechado e a restrição de \(\gamma\) a \(]a,b[\) é injectiva.

A linha quebrada de vértices \(z_0,z_1,z_2,\ldots,z_n\) é a curva \[\gamma(t)= \bigl(1-(t-k)\bigr)z_k + (t-k)z_{k+1}\quad \text{para \(t\in[k,k+1]\) e \(k=0,1,\ldots,n-1\)}\]  

Subconjuntos conexos

Diz-se que um subconjunto aberto \(U\) é conexo quando para cada par de pontos \(p_1,p_2\in U\) existe uma linha quebrada \(\gamma(t) \subset U\) que liga os ponto \(p_1\) e \(p_2\). Chama-se domínio  a todo o cinjunto aberto e conexo.

Seja \(\gamma(t)\) uma curva de Jordan. Um teorema de Jordan diz que \(\mathbb{C}\setminus \text{imagem de \(\gamma(t)\)}\) é a união disjunta de dois domínios um de quais é limitado.  Chama-se interior de \(\gamma\) ao domínio limitado de \(\mathbb{C}\setminus  \text{imagem de \(\gamma(t)\)}\) e escreva \(\operatorname{int}\gamma\).

Funções Holomorfas

\( \def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\R{\mathbb{R}}\)

 

O conjunto de funções holomorfas é uma extensão do conjunto dos polinómios na variável \(z\). Por exemple considere o polinómio \(f(z) =z\). A inversa \(\frac{1}{z}\), definida no aberto \(\{z \colon |z|>0\}\), não é um polinómio, mas é uma função holomorfa.

Funções holomorfas são definidas usando a dervidada complexa \[\lim_{h\rightarrow 0}\frac{f(z+h)-f(z)}{h}\] ou um sistema de derivadas parciais, que se chama as equações de Cauchy-Riemann em que \[u(x,y) = \re f(x +iy) \quad \text{e} \quad v(x,y) = \im f(x+iy).\]

Seja \(U\subseteq \R^2 =\C\) um domínio. Diz-se que uma função \(u \colon U \rightarrow \R\) é de classe \(\mathcal{C}^k\) quando \[ \frac{\partial^{j} u}{\partial x^{a} \partial y^b}\] existe e é contínua para \(0\leq a,b,j\leq k\) e \(a+b=j\). Para uma função complexa \(f\colon U\rightarrow \C\) podemos escrever \[f(x+iy) = u(x,y) + iv(x,y)\] com \(u = \re f\) e \(v=\im f\). Diz-se que \(f\) é de classe \(\mathcal{C}^k\) quando \(u\) e \(v\) são ambos de classe \( \mathcal{C}^k\).


Definição: Seja \(f(x+iy=u(x,y)+iv(x,y)\) uma função complexa de classe \(\mathcal{C}^1\) definida num domínio \(U\).  Diz-se que \(f\) é holomorfa se \begin{equation}\label{eq:CR}\frac{\partial u}{\partial x} = \frac{\partial v}{\partial y} \qquad \frac{\partial u}{\partial y}= -\frac{\partial v}{\partial y}  \end{equation} em cada ponto de \(U\). 

Exemplo: A função \(f(z) = z\) é holomorfa.  De facto \( f(x+iy) = x + iy\). Logo \(u(x,y)=x\) , \( v(x,y)=y\) e \begin{align} \frac{\partial u}{\partial x}&= 1 & \frac{\partial v}{\partial x}&=0\\ \frac{\partial u}{\partial y}&= 0 & \frac{\partial v}{\partial y}&=1.\end{align} Segue-se que \(f(z)\) é holomorpha.

Tem-se

  • Se \(f\) e \(g\) são holomorfas, então \(f+g\) e \(f\cdot g\) são holomorfas.
  • Se \(f\) e \(g\) são holomorfas e \(g\) é não nula, então \(\frac{f}{g}\) é holomorfa. 

Segue-se que cada polinómio \(p(z) = a_0 + a_1 z + a_2 z^2 + \cdots a_n z^n\) é holomorfo, e cada função racional \[\frac{p(z)}{q(z)} = \frac{a_0 + a_1 z + a_2 z^2 + \cdots a_n z^n}{b_0 + b_1 z + b_2 z^2 + \cdots b_m z^m}\] é uma função holomorf em \(\C\setminus \{z : q(z)=0\}\).

Equações de Cauchy-Riemann

Considere uma função complexa \(f(x,y)\) em variáveis reais. Temos \[x=\frac{z+\overline{z}}{2} \qquad y = \frac{z-\overline{z}}{2}\] Usando as regras de cáculo temos \begin{align*} \frac{\partial}{\partial z}&= \frac{\partial x}{\partial z}\frac{\partial}{\partial x} + \frac{\partial y}{\partial z}\frac {\partial }{\partial y} & \frac{\partial}{\partial \overline{z}}&= \frac{\partial x}{\partial \overline{z}}\frac{\partial}{\partial x} + \frac{\partial y}{\partial \overline{z}}\frac {\partial }{\partial y} \\ &=\frac{1}{2}\frac{\partial}{\partial x} + \frac{1}{2i}\frac{\partial}{\partial y} & &= \frac{1}{2}\frac{\partial}{\partial x} - \frac{1}{2i}\frac{\partial}{\partial y} \\ &= \frac{1}{2}\left(\frac{\partial}{\partial x} - i \frac{\partial}{\partial y}\right) & &= \frac{1}{2}\left(\frac{\partial}{\partial x} + i \frac{\partial}{\partial y}\right) \end{align*} Para \(f(x,y)= u(x,y) + iv(x,y) \) com \(u (x,y) = \re f(x,y)\) e \(v(x,y) = \im f(x,y)\), obtemos 

\begin{align} \frac{\partial f}{\partial \overline{z}} &=  \frac{1}{2}\left(\frac{\partial}{\partial x} + i \frac{\partial}{\partial y}\right)(u+iv) \\ &=\frac{1}{2}\left(\frac{\partial u}{\partial x} - \frac{\partial v}{\partial y} \right) +  \frac{i}{2}\left(\frac{\partial u}{\partial y}+\frac{\partial v}{\partial x}\right) \\ \end{align} Recorde-se que as equações de Cauchy-Riemann são  \[ \frac{\partial u}{\partial x} = \frac{\partial v}{\partial y} \qquad  \frac{\partial u}{\partial y}= - \frac{\partial v}{\partial x} \] Portanto uma função \(f\) de classe \(\mathcal{C}^1\) num domínio \(U \subseteq \C\) é holomorfa se e só se\[\frac{\partial f}{\partial\overline{z}} = 0\] Quando \(f(x,y) = u(x,y) + iv(x,y)\) é uma função holomorfa obtemos \begin{align}\frac{\partial f}{\partial z} &= \frac{1}{2}\left(\frac{\partial u}{\partial x} + i \frac{\partial v}{\partial x} - i\frac{\partial u}{\partial y} + \frac{\partial v}{\partial y}\right) \\&=\frac{\partial u}{\partial x}+ i\frac{\partial v}{\partial x} \\&=\frac{\partial f}{\partial x}\end{align}

Recorde-se que se \(u(x,y)\) e \(v(x,y)\) são de classe \(\mathcal{C}^1\) então \begin{align} u(x+p,y+q) - u(x,y) &= \frac{\partial u}{\partial x}p + \frac{\partial u}{\partial y}q + \epsilon_1 \\ v(x+p,y+q)-v(x,y) &= \frac{\partial v}{\partial x}p + \frac{\partial v}{\partial y}q + \epsilon_2\end{align} com \begin{align}\lim_{p+iq\rightarrow 0}\frac{\epsilon_1}{p+iq} &=0 \\ \lim_{p+iq\rightarrow 0}\frac{\epsilon_2}{p+iq} &=0.\end{align}

Obtemos \[f(z+p+iq) - f(z) = \left(\frac{\partial u}{\partial x} + i \frac{\partial v}{\partial x}\right){(p+iq)} + \epsilon_1 + \epsilon_2,\] e logo \[\frac{df}{dz} =\lim_{p+iq \rightarrow 0} \frac{f(z+p+iq)-f(z)}{p+iq} = \frac{\partial u}{\partial x} + i\frac{\partial v}{\partial x}\]

Funções Harmónicas

\( \def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\grad{\operatorname{grad}} \)

Considere uma função real \(u(x,y)\) de classe \(\mathcal{C}^2\) em variáveis reais \(x,y\). Diz-se que \(u(x,y)\) é uma função harmónica quando satifaz a equação de Laplace \[ \frac{\partial^2 u}{\partial x^2} +\frac{\partial^2 u}{\partial y^2} = \Delta u = 0. \]

Recorde-se que temos operadores \begin{align*} \frac{\partial}{\partial z} &= \frac{1}{2}\left(\frac{\partial}{\partial x} - i \frac{\partial}{\partial y}\right) & \frac{\partial}{\partial \overline{z}} &= \frac{1}{2}\left(\frac{\partial}{\partial x} + i \frac{\partial}{\partial y}\right) \end{align*} Logo \[ \frac{\partial^2}{\partial z \partial \overline{z}} = \frac{1}{4}\left(\frac{\partial^2}{\partial x^2} + \frac{\partial^2}{\partial y^2} \right) \]

Vamos ver mais tarde o seguinte.

Para uma função holomorfa \(f(z)\) a sua derivada \(f'(z)\) é homolomorfa.

Portanto a parte real de \(f\) bem como a parte imaginária de \(f\) são de classe \(\mathcal{C}^\infty\).

Ora, para \(f(x+iy)= u(x,y) + iv(x,y) \) com \(u (x,y) = \re f(x+iy)\) e \(v(x,y) = \im f(x+iy)\), uma função holomorfa obtemos \begin{align} \left(\frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2}\right) + i \left(\frac{\partial^2 v}{\partial x^2} + \frac{\partial^2 v}{\partial y^2} \right) &= \left(\frac{\partial^2}{\partial x^2} + \frac{\partial^2}{\partial y^2} \right)(u+iv) \\ &= 4 \frac{\partial^2}{\partial z \partial \overline{z}}(u+iv) \\ &= 4\frac{\partial^2 f}{\partial z \partial \overline{z}} \\ &= 0 \end{align} Portanto \( u(x,y) \) e \(v(x,y)\) são funções harmónicas.

Agora suponha-se que \(u(x,y)\) é uma função harmónica no plano real \(\mathbb{R}^2\). Vamos ver que existe uma função harmónica \[ v(x,y) \], que se chama uma conjugada harmónica de \(u\),  tal que \[ f(x+iy) = u(x,y) + i v(x,y)\] é uma função holomorfa.

Se uma função \(v\) existe, então \[ \frac{\partial u}{\partial x} = \frac{\partial v}{\partial y} \qquad \frac{\partial u}{\partial y}=-\frac{\partial v}{\partial y} \] e \[ \grad v = \left(\frac{\partial v}{\partial x},\frac{\partial v}{\partial y}\right) = \left(-\frac{\partial u}{\partial y},\frac{\partial u}{\partial x}\right). \] Dado um caminho regular por partes \(\sigma\colon[a,b]\rightarrow \mathbb{C}\) temos \[ v(\sigma(b)) - v(\sigma(a)) = \int_a^b \grad u \cdot \dot{\sigma}\, dt \] Dado um ponto \((x,y)\in\mathbb{R}^2\) considere o caminho \(\sigma(t) = t\cdot(x,y)\). O vetor tangente é \(\dot{\sigma}=(x,y)\). Define-se \begin{align} v(x,y) &= \int_0^1 \left(-\frac{\partial u}{\partial y},\frac{\partial u}{\partial x}\right)\cdot(x,y)\,dt \\ &= \int_0^1 -x\frac{\partial u}{\partial y}+y\frac{\partial u}{\partial x}\,dt \\ &= \int_{\sigma}- \frac{\partial u}{\partial y} dx + \frac{\partial u}{\partial x}dy. \end{align}

Para ver que \(v\) é bem definida observamos se \(\gamma(t)\) é qualquer caminho regular por partes com ponto inicial \((0,0)\), ponto final \((x,y)\) e \(\sigma - \gamma\) uma curva de Jordan, temos \begin{align} \int_{\sigma}-\frac{\partial u}{\partial y} dx + \frac{\partial u}{\partial x}dy - \int_{\gamma} -\frac{\partial u}{\partial y} dx + \frac{\partial u}{\partial x}dy &= \int_{\sigma -\gamma}- \frac{\partial u}{\partial y} dx + \frac{\partial u}{\partial x}dy \\ &=\iint_{\operatorname{int}(\sigma-\gamma)} \frac{\partial^2 u}{\partial x^2}+\frac{\partial^2 u}{\partial y^2}\, dxdy \\ &=0 \end{align} Segue-se que \(f(x+iy) = u(x,y) + iv(x,y)\)  é uma função holomorfa.

Funções Harmónicas II

Exemplo da função harmónica \[ u(x,y) = \log\sqrt{x^2+y^2}.\]

Séries de Potências

\( \def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\grad{\operatorname{grad}} \def\sen{\operatorname{sen}} \def\sh{\operatorname{sh}} \def\ch{\operatorname{ch}} \)

No cáculo estude-se séries de pontências. Por exemplo tem-se \begin{align*} e^x &= 1 + x + \frac{x^2}{2!} + \cdots + \frac{x^n}{n!} + \cdots, \\ \cos x &= 1-\frac{x^2}{2!} + \frac{x^4}{4!} + \cdots +(-1)^n\frac{x^{2n}}{(2n)!}+ \cdots \\ \ch x &= 1+\frac{x^2}{2!} + \frac{x^4}{4!} + \cdots +\frac{x^{2n}}{(2n)!}+ \cdots \\ \sen x &= x-\frac{x^3}{3!} + \frac{x^5}{5!} + \cdots +(-1)^n\frac{x^{2n+1}}{(2n+1)!}+ \cdots \\ \sen x &= x+\frac{x^3}{3!} + \frac{x^5}{5!} + \cdots +\frac{x^{2n+1}}{(2n+1)!}+ \cdots \\ \end{align*}

Seja \(P\in\C\) um ponto. Uma série de potências \begin{equation}\label{eq:1} \sum_{n=0}^\infty a_n(z-P)^n \end{equation} é definida pelo limite das sumas parciais \[ S_N(z) = \sum_{n=0}^n a_n (z-P)^n. \] Diz-se que a série em \(\sum_{n=0}^\infty a_n(z-P)^n \) converge quando o limite \(\lim_{N\rightarrow \infty}S_N(z)\) existe. Note-se que \[ \lim_{N\rightarrow \infty}S_N(z) \quad \text{existe}\qquad \text{se e só se } \qquad \lim_{N\rightarrow \infty}\re S_N(z) \text{ e } \lim_{N\rightarrow \infty}\im S_N(z) \text{ existem.} \]

Propriedades Gerais de Séries:

  • Se a série \(\sum a_n\) converge, então \(\lim_{n\rightarrow\infty} a_n =0\).
  • Se \(\sum_{n=0}^\infty a_n \) e \(\sum_{n=0}^\infty b_n\) convergem, então para qualquer \(k\in\C\) a série \(\sum_{n=0}^\infty a_n+kb_n\) converge.
  • Se a série real \(\sum_{n=0}^\infty |a_n|\) converge, então a série complexa \(\sum_{n=0}^\infty a_n \) converge.
  • Se \(b_n>0\) e \(\sum_{n=0}^\infty b_n\) converge e existe um constante \(k>0\) tal que \(|a_n|\leq k\cdot b_n\), então \(\sum a_n\) converge.
  • Se \[ L = \lim_{n\rightarrow\infty} \left|\frac{a_{n+1}}{a_n}\right|, \] então \(\sum |a_n|\) e \(\sum a_n\) convergem quando \(L<1\) e \(\sum |a_n|\) diverge se \(L>1\).
  • Se \[ L = \lim_{n\rightarrow\infty} \sqrt[n]{|a_{n}|}, \] então \(\sum |a_n|\) e \(\sum a_n\) convergem quando \(L<1\) e \(\sum |a_n|\) diverge quando \(L>1\).

Série Geométrica

Recorde-se \[ (1-z)(1+z+z^2 + \cdots + z^n)=1-z^{n+1}. \] Portanto \begin{equation}\label{eq:2} 1+z+z^2 + \cdots +z^n = \frac{1-z^{n+1}}{1-z} \quad\text{para \(z\neq 0\)}. \end{equation} Quando \(|z|1\) o limite \(\lim_{n\to\infty}|z|^n\) não existe. Portanto \[ \sum z^n \begin{cases} \text{converge e } \sum z^n=\frac{1}{1-z} & \text{se \(|z|<1\)} \\ \text{não converge } & \text{se \(|z|>1\)} \end{cases} \]

Exemplos:

\[1-z+z^2 - z^3 + \cdots + (-1)^nz^n + \cdots = \frac{1}{1+z}\quad  \text{ se  } |z|<1.\]

\[1+z^2 + z^4 + \cdots z^{2n} +\cdots = \frac{1}{1-z^2}\quad\text{ se } |z|< 1.\]

\[\frac{1}{z-4} = -\frac{1}{4}\frac{1}{1-z/4} = -\sum_{n=0}^\infty\frac{z^n}{4^{n+1}}\quad\text{ se } |z|<4.\]

Para \( a\neq b\) temos \[\frac{1}{(z-a)(z-b)} = \frac{1}{a-b}\left( \frac{1}{z-a} - \frac{1}{z-b} \right) = \frac{1}{a-b}\left( \frac{1}{b^{n+1}} - \frac{1}{a^{n+1}} \right) z^n\quad\text{ se } |z|<\min\{|a|,|b|\}.\]

\[\frac{1}{1+z+z^2} = \frac{1-z}{1-z^3} = \sum_{n=0}^\infty (z^{3n}-z^{3n+1})\quad\text{ se } |z|<1.\]

Séries de Potências II

\(\def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\R{\mathbb{R}} \def\grad{\operatorname{grad}} \def\sen{\operatorname{sen}} \def\sh{\operatorname{sh}} \def\ch{\operatorname{ch}} \)

Raio de Convergência.

Seja \(P\in\C\) um ponto, e considere-se uma série de potências \begin{equation*} f(z)=\sum_{n=0}^\infty a_n(z-P)^n. \end{equation*} Define-se raio de convergência de \(f\) por \[ R = \sup \left\{ |z-P| : \text{\(\sum_{n=0}^\infty |a_n(z-P)^n|\) converge }\right\} \] Escreve-se \(R=\infty\) quando a série \(\sum_{n=0}^\infty |a_n(z-P)^n|\) converge para \(|z-P|\) arbitrária grande.

Exemplos:

  1. Considere a série \(\sum_{n=0}^\infty nz^n\). Aplicando a teste de razão, obtemos para \(z\neq 0\)  \[ \lim_{n\to \infty}\left|\frac{(n+1)z^{n+1}}{nz^n}\right| = \lim_{n\to \infty}\left|\frac{(n+1)}{n}\right|\cdot|z| = |z|. \] Portanto \(\sum_{n=0}^\infty nz^n\) converge para \(|z|1\). Logo o raio da convergência da série é 1.
  2. Considere a série \(\sum_{n=0}^\infty \frac{z^n}{n!}\). Aplicando a teste de razão, obtemos para \(z\neq 0\) \[ \lim_{n\to \infty}\left|\frac{z^{n+1}}{(n+1)!}\cdot\frac{n!}{z^{n}}\right| = \lim_{n\to \infty}\left|\frac{1}{n}\right|\cdot|z| = 0. \] Portanto o raio da convergência da série \(\sum_{n=0}^\infty \frac{z^n}{n!}\) é \(\infty\).
  3. Considere a série \(\sum_{n=0}^\infty n^nz^n\). Aplicando a teste de raizes, obtemos \[ \lim_{n\to \infty}\sqrt[n]{|n^nz^n|} = \lim_{n\to \infty}n|z| = \begin{cases} 0 & \text{se \(z=0\)} \\ \infty & \text{se \(z\neq 0\).} \end{cases} \] Portanto o raio da é 0.

Para a série \(f(z)=\sum_{n=0}^\infty a_n(z-P)^n\), o teorem de Cauchy-Hadamard diz- que o raio de convergência é \[ R = \frac{1}{\limsup_{n\to\infty}\sqrt[n]{|a_n|} }. \]

Recorde-se que para uma sucessão de números reais \(\{b_n\}_{n=0}^\infty\), \[ \limsup_{n\to\infty} b_n = \lim_{n\to\infty} \sup\{ b_k : k\geq n\}. \] Para \(c_n = \sup\{b_k : k\geq n \}\in \{-\infty,+\infty\}\cup \R \) tem-se \[ c_0 \geq c_1 \geq c_2 \geq \cdots c_n \geq \cdots . \] Portanto \(\limsup_{n\to\infty}b_n\) existe em \(\{-\infty,+\infty\}\cup \R\). Além disso, \(\limsup_{n\to\infty} b_n = L\) se e só se

(i) existe uma subsucessão \(\{b_{n_j}\}\) de \(\{b_n\}\) que converge a \(L\) e

(ii) se \(L'>L\), então existe um inteiro \(K\) tal que \(b_n < L' \) para cada \(n\geq K\).

Quando \(\lim_{n\to\infty}b_n=b\) com \(-\infty\leq b\leq +\infty\) tem-se \(b=\limsup_{n\to\infty}b_n\).

Considere uma série de potências \[ f(z) = \sum_{n=0}^\infty a_n (z-P)^n \] e as suas somas parciais \[ S_N(z) = \sum_{n=0} a_n(z-P)^n \] Como, para cada \(N\), \(S_N(z)\) é um polinómio, a função \(S_N(z)\) é holomorfa.

Se a série \(f\) tem raio de convergência \(R>0\), então \[ S_N(z) \to f(z) \quad \text{uniformemente no disco aberto \( D(P,R) = \{z : |z-P| < R \}\)}. \] Isto é a convergência é independente de \(z\in D(P,R)\).

Temos \[ \frac{\partial S_N}{\partial z}(z) = S_N'(z) = a_1 + 2a_2(z-P) +\cdots +na_n(z-P)^{n-1}+\cdots \]

Lema.

A série \(\sum_{n=1}^\infty na_n(z-P)^{n-1}\) tem o mesmo raio de convergencia de \(\sum_{n=0}^\infty_z(z-P)^n\).

Demostração.

Seja \( 0 < \rho < R \). Para \(z\neq P\) temos \[ \left|na_n(z-P)^{n-1}\right| = \frac{1}{|z-P|} n\left(\frac{|z-P|}{\rho}\right)^n|a_n\rho|^n \] Aplicando o teste de razão, obtemos que \(\sum n\left(\frac{|z-P|}{\rho}\right)^n\) converge para \(|z-P|<\rho\). Portanto
\(\lim_{n\to\infty}n\left(\frac{|z-P|}{\rho}\right)^n =0\).  Logo existe \(M>0\)  tal que \[ n\left(\frac{|z-P|}{\rho}\right)^n < M \quad \text{para cada \(n\)} \]

Assim, \[ \left|na_n(z-P)^{n-1}\right| \leq \frac{M}{|z-P|}\cdot|a_n\rho|^n. \] Portanto \(\sum n a_n(z-P)^{n-1}\) converge e tem raio de convergência \(\geq R \).

Agora, se \(\sum n a_n(z-P)^{n-1}\) converge, temos \[ |a_n(z-P)^n| \leq n |a_n(z-P)^n| = |z-P|\cdot | na_n(z-P)^{n-1}| \quad \text{ para \(n\geq 1\)}. \] Pelo critério de comparação segue-se que \(\sum a_n(z-P)^n\) converge. Logo as séries têm o mesmo raio de convergência. \(\square\)

 

Agora vamos ver que \[f'(z) = \sum_{n=1}^\infty n a_n(z-P)^{n-1}.\]

Sejam \[ g(z) = \sum_{n=1}^\infty na_n(z-P)^{n-1}\qquad \text{ e } \qquad R_N(z) = \sum_{n=N+1}^\infty a_n(z-P)^n \]

Temos \(f(z) = \sum a_n(z-p)^n = S_N(z) + R_N(z)\)  e

\begin{align} \frac{f(z) - f(z_0)}{z-z_0} - g(z_0) &= \frac{S_N(z)-S_N(z_0)}{z-z_0} - S_N'(z_0) + \left(S_N'(z_0) - g(z_0)\right) + \frac{R_N(z)-R_N(z_0)}{z-z_0} \\ \frac{R_N(z)-R_N(z_0)}{z-z_0} &= \sum_{n=N+1}^\infty a_n \frac{(z-P)^n-(z_0-P)^n}{(z-P) - (z_0-P)} \\ & = \sum_{n=N+1}^\infty a_n\sum_{k=1}^n (z-P)^{n-k}(z_0-P)^k \end{align}

Portanto, para \( z \neq z_0\) e \(|z-P|,|z_0-P| < \rho < R \)

\begin{align} \left| \frac{R_N(z)-R_N(z_0)}{z-z_0}\right| &\leq \sum_{n=N+1}^\infty |a_n|n\rho^{n} \to 0 \quad \text{como \(N\to\infty\)} \end{align}

Seja \(\epsilon >0\). Existe \(n_0\) tal que

\[ \left| \frac{R_N(z)-R_N(z_0)}{z-z_0}\right| \leq \frac{\epsilon}{3} \]

quando \(n\geq n_0\).

Como \(S_N'(z_0) \to g(z_0)\) existe \(n_1\) tal que

\[\left| S_N'(z_0) - g(z_0) \right| < \frac{\epsilon}{3} \]

quando \(0<|z-z_0|<\delta\).

Finalmente, pela definição de derivada existe \(\delta> 0 \) tal que

\[\left|\frac{S_N(z)-S_N(z_0)}{z-z_0} - S_N'(z)\right| < \frac{\epsilon}{3} \]
quando \(0<|z-z_0|<\delta\).

Obtemos \[\left|\frac{f(z) - f(z_0)}{z-z_0} - g(z_0)\right| < \epsilon\] e \(f'(z) = \sum_{n=1}^\infty na_n(z-P)^{n-1}.\)z

Integrais complexos

\( \def\class{\mathcal{C}} \def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\R{\mathbb{R}} \def\grad{\operatorname{grad}} \def\sen{\operatorname{sen}} \def\sh{\operatorname{sh}} \def\ch{\operatorname{ch}} \def\bz{\overline{z}} \)

Curvas regulares

Um função \(\varphi \colon [a,b]\to \R\) é regular quando

  1. \(\varphi\) é contínua em \([a,b]\);
  2. \(\varphi'\) existe e é contínua em \(]a,b[\);
  3. \(\varphi'\) tem uma extensão contínua a \([a,b]\), i.e., os limites \[ \lim_{t\to a^+}\varphi'(t) \qquad \lim_{t\to b⁻}\varphi'(t) \] existem.

Tem-se \[\varphi(b) - \varphi(a) = \int_a^b \varphi'(t) \, dt.\] Uma curva \(\gamma\colon [a,b]\to \C\) é regular em \([a,b]\) quando \(\re \gamma(t)\) e \(\im \gamma(t)\) são regulares em \([a,b]\). Quando \(\gamma(t) = \gamma_1(t) + i \gamma_2(t) \) é regular escreve-se \[ \gamma'(t) = \gamma_1'(t) + i \gamma_2(t). \]

Integrais

Para uma função contínua \(\psi\colon[a,b]\to \C\) com \(\re \psi(t) = \psi_1(t)\) e \(\im\psi(t) = \psi_2(t)\), define-se \[\int_a^b \psi(t) \, dt = \int_a^b \psi_1(t) \,dt + i \int_a^b \psi_2(t)\, dt.\] Quando \(\psi (t)\) é regular, temos \[ \psi(b) - \psi(a)= \int_a^b \psi'(t)\, dt.\] Para uma função \(f(z)\) de class \(\class^1\) num aberto \(U\) e uma curva regular \(\gamma\colon[a,b]\to U\) tem-se \[ f(\gamma(b)) - f(\gamma(a)) = \int_a^b \left(\frac{\partial f}{\partial x}(\gamma(t))\cdot\frac{d\gamma_1}{dt} + \frac{\partial f}{\partial y}(\gamma(t))\cdot\frac{d\gamma_2}{dt}\right) \, dt.\] Quando \(f\) é homolorfa, tem-se

  • \(0=\dfrac{\partial f}{\partial \bz}= \dfrac{1}{2}\left(\dfrac{\partial f}{\partial x} + i\dfrac{\partial f}{\partial y}\right) \);
  • \( \dfrac{\partial f}{\partial x} =-i \dfrac{\partial f}{\partial y} \);
  • \(\dfrac{\partial f}{\partial x}(\gamma(t))\cdot\dfrac{d\gamma_1}{dt} + \dfrac{\partial f}{\partial y}(\gamma(t))\cdot\dfrac{d\gamma_2}{dt} = \dfrac{\partial f}{\partial z}(\gamma(t))\cdot \gamma'(t) \).

Portanto \[ f(\gamma(b)) - f(\gamma(a)) = \int_a^b \frac{\partial f}{\partial z}(\gamma(t))\cdot \frac{d\gamma}{dt} \, dt. \] Para \(g(z)\) uma função contínua em \(u\) e \(\gamma\colon[a,b]\to U \) uma curva regular, define-se o integral de \(g\) ao longo \(\gamma\) por \[ \oint_\gamma g(z)\, dz = \int_a^b g(\gamma(t)) \cdot \gamma'(t) \, dt.\]

Teorema de Cauchy

Teorema de Cauchy em um triângulo

Seja \(f\) uma função holomorfa em uma região \(\Omega\) que um triângulo \(\gamma\) e seu interior.  Tem-se \[\oint_\gamma f(z)\, dz = 0.\]

Para pontos distintos \(a,b\in\mathbb {C}\) designa-se por \( [a,b]\) o segmento \(\{ a\cdot (1-t) + b\cdot t : 0\leq t \leq 1\}\).  Diz-se que uma região \(\Omega\subseteq \mathbb{C}\) é conexa quando cada par do pontos \(a,b\in\Omega\) o segmento \([a,b]\subset\Omega\). 

Agora, seja \(f(z)\) uma função contínua em uma região conexa \(\Omega\) tal que para cada triângulo \(\gamma \subset \Omega\) tem-se \[\oint_\gamma f(z)\, dz = 0.\] Seja \(a\in\Omega\). A função \[F(z) = \int_{[a,z]}f(w)\,dw\] é holomorfa em \(\Omega\).

Obtemos que cada função holomorfa em uma região conexa tem uma primitiva.  De facto, dada uma função holomorfa \(f\) em um região conexa \(\Omega\) temos  \[\oint_\gamma f(z)\,dz=0\] para cada triângulo \(\gamma\subset\Omega\), e portanto, a função \[F(z) = \int_{[a,z]} f(w)\, dw\] é holomorfa e \(F'(z) = f(z)\).

Teorema de Cauchy em uma região conexa

Se \(\Omega\) é uma região conexa e \(f\) é holomorfa em \(\Omega\), então para cada caminho \(\gamma \subset \Omega\) regular por partes e fechado, tem-se \[\oint_\gamma f(z)\,dz = 0.\]

Variação das constantes

\( \def\class{\mathcal{C}} \def\re{\operatorname{Re}} \def\im{\operatorname{Im}} \def\C{\mathbb{C}} \def\Z{\mathbb{Z}} \def\R{\mathbb{R}} \def\grad{\operatorname{grad}} \def\sen{\operatorname{sen}} \def\sh{\operatorname{sh}} \def\ch{\operatorname{ch}} \def\bz{\overline{z}} \def\res{\operatorname{Res}} \def\tg{\operatorname{tg}} \def\res{\operatorname{Res}} \def\ind{\operatorname{Ind}} \def\Exp{\operatorname{Exp}} \)

Variação das constantes

Considere a equação diferencial \[ y'' + a_1 y' + a_2 y = b(t) \] com \(b(t)\) uma função contínua em um intervalo aberto \(I\subseteq \R\).

Recorde-se que para uma equação linear do 1ª ordem \[y' + a(t)y = b(t)\] temos um factor integrante \[\mu(t) = \Exp\left[\int^t a(s)\, ds\right]\] e uma solução particular \[y_p(t) = \mu(t)^{-1} \int^t \mu(s)\cdot b(s)\, ds.\] Além disso, a solução geral da equação homogénea \(y' + a(t) y = 0\) é \[y_h(t) K, \quad \text{onde } y_h(t) = \mu(t)^{-1}.\] Logo a solução geral da equção não-homogénea \(y' + a(t) y = b(t)\) é \[ y_h(t)\cdot K + y_h(t)\cdot \int^t y_h(s)^{-1}\cdot b(s)\, ds.\]

Usando vectores e matrizes podemos escrever a equação \(y'' + a_1y' + a_2 y = b(t)\) na forma de um sistema de equações diferenciais de 1ª ordem. Sejam \[ \vec{y} = \begin{bmatrix}y \\ y' \end{bmatrix}, \quad A = \begin{bmatrix} 0 & -1 \\ a_2 & \phantom{-}a_1 \end{bmatrix}\quad \text{e}\quad \vec{b}(t) = \begin{bmatrix}0 \\ b(t) \end{bmatrix}.\] Temos \[\vec{y}' + A\cdot\vec{y} = \vec{b}(t).\]

Como o caso de equações de 1ª ordem, temos um factor integrante \begin{align*}\Exp\left[At\right] =& I + At + \frac{1}{2}A^2 + \cdots \frac{1}{n!}A^n t^n + \cdots \\ =&\sum_{n=0}^\infty \frac{1}{n!} A^n t^n\end{align*} com \[\frac{d}{dt}\Exp[At] = A\Exp[At] = \Exp[At]A \quad\text{e}\quad \Exp[At]^{-1} = \Exp[-At].\] Assim, obtemos uma solução particular \[\vec{y}_p (t) = \Exp[-At]\cdot \int^t \Exp[As]\cdot\vec{b}(s)\, ds\] e a solução geral da equção vectorial \[\vec{y}(t) = \Exp[-At]\cdot \vec{y}_0 + \Exp[-At]\cdot \int^t \Exp[As]\cdot\vec{b}(s)\, ds.\]

Cálcule de \(\Exp[At]\)

Sejam \(w_1(t),w_2(t)\) soluções linearmente independentes da equação homogénea \(y'' + a_1y + a_2y= 0\), e considere a matriz \[W(t) = \begin{bmatrix} w_1 & w_2 \\ w_1' & w_2'\end{bmatrix}. \] Temos \[W' + A\cdot W = 0,\] e portanto \begin{align*} \frac{d}{dt}\biggl[\Exp[At]\cdot W(t) \biggr] &= \Exp[At]W'(t) + \Exp[At]\cdot A W(t) \\ &= \Exp[At]\biggl[W'(t) + A\cdot W(t)\biggr] \\ &= 0.\end{align*} Logo \(\Exp[At]\cdot W(t)\) é contstante. Como \(\Exp[A\cdot 0] = I\) obtemos \[ \Exp[At] \cdot W(t) = W(0), \quad \Exp[At] = W(0)\cdot W(t)^{-1}\quad \text{e} \quad \Exp[-At] = W(t) \cdot W(0)^{-1}.\] Assim, \begin{align*} \vec{y}_p(t) &= \Exp[-At]\cdot \int^t \Exp[As]\cdot\vec{b}(s)\, ds \\ &= W(t)\cdot W(0)^{-1} \int^tW(0)\cdot W(s)^{-1}\vec{b}(s)\,ds \\ &= W(t)\cdot \int^t W(s)^{-1}\cdot \vec{b}(s)\, ds \\ &= \begin{bmatrix} w_1 & w_2 \\ w_1' & w_2'\end{bmatrix} \cdot \int^t \frac{1}{w_1w_2' - w_2w_1'} \cdot \begin{bmatrix} \phantom{-}w_2' & -w_2 \\ -w_1' & \phantom{-}w_1 \end{bmatrix} \begin{bmatrix}0 \\ b(s)\end{bmatrix}\, ds \\ &= \begin{bmatrix} w_1 & w_2 \\ w_1' & w_2'\end{bmatrix} \cdot \int^t \frac{1}{w_1w_2' - w_2w_1'} \cdot \begin{bmatrix}-w_2(s) b(s) \\ \phantom{-}w_1(s) b(s) \end{bmatrix} \, ds \end{align*}

Exemplo

\[y'' + 3y' + 2y = e^{-t} \] O polinómio característico é \(\lambda^2 + 3\lambda + 2 = (\lambda+1)(\lambda+2)\). Obtemos soluções linearmente independentes da equção homogénea \(y'' + 3y' +2y=0\) \[w_1(t) = e^{-t}, \quad w_2(t) = e^{-2t}.\] \begin{align*} W(t) &= \begin{bmatrix} e^{-t} & e^{-2t} \\ - e^{-t} & -2e^{-2t} \end{bmatrix} \\ \det W(t) &= w_1w_2'- w_2w_1' = -e^{-3t} \\ \int^t \frac{1}{w_1w_2' - w_2w_1'} \cdot \begin{bmatrix}-w_2(s) b(s) \\ \phantom{-}w_1(s) b(s) \end{bmatrix}\, ds &= \int^t \begin{bmatrix} \phantom{-}1 \\ -e^s \end{bmatrix}\, ds \\ &= \begin{bmatrix} \phantom{-}t \\ -e^t \end{bmatrix} \\ \vec{y}_p(t) &= W(t) \cdot \begin{bmatrix} \phantom{-}t \\ -e^t \end{bmatrix} \\ &= \begin{bmatrix} (t-1)e^{-t} \\ (2-t)e^{-t} \end{bmatrix} \end{align*} A solução geral da equação é \[y(t) = c_1 e^{-t} + c_2e^{-2t} + (t-1)e^{-t}.\]