07 - Matrizen

Eine \((m\times n)\)-Matrix ist eine Anordnung von \(m\cdot n\) Zahlen \(a_{ij} \in \mathbb{R},\) \(i=1,\ldots,m,\) \(j=1,\ldots,n\), der Form $$A = (a_{ij})_{i,j} = \left(\begin{array}{cccc}a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \vdots & & & \vdots \\ a_{m1} & a_{m2} & \cdots & a_{mn}\end{array}\right).$$ \((m,n)\) heißt Dimension der Matrix.

Spezielle Matrizen

  1. Die Matrix \(\textbf{0} = \textbf{0}_{m\times n}\), deren Einträge alle Null sind, heißt Nullmatrix.
  2. Eine \((n \times n)\)-Matrix mit \(a_{ij}=0\) für alle \(i\neq j\) heißt Diagonalmatrix.
  3. Die Diagonalmatrix \(\textbf{I}=\textbf{I}_{n\times n} = diag(1,\ldots,1)\), deren Diagonaleinträge alle 1 sind, heißt Einheitsmatrix.

Rechenregeln

Matrixaddition: Die Matrix \(\textbf{C}=\textbf{A}+\textbf{B}\) ergibt sich aus zwei Matrizen \(\textbf{A}\) und \(\textbf{B}\) mit gleicher Dimension durch \(c_{ij}=a_{ij}+b_{ij}\).

Skalare Multiplikation: Ein Skalar \(c \in \mathbb{R}\) wird mit einer Matrix \(\textbf{A}\) elementweise multipliziert: \(c\textbf{A}=(ca_{ij})_{i,j}\).

Für Matrizen \(\textbf{A}, \textbf{B}, \textbf{C}\) gleicher Dimension und Skalare \(c, d \in \mathbb{R}\) gilt:

  1. \((\textbf{A} + \textbf{B}) + \textbf{C} = \textbf{A} + (\textbf{B} + \textbf{C})\).
  2. \(c(\textbf{A} + \textbf{B}) = c\textbf{A} + c\textbf{B}\).
  3. \((c + d)\textbf{A} = c\textbf{A} + d\textbf{A}\).

Matrix-Vektor-/Matrix-Multiplikation

Sei \(\textbf{A}\) eine \((m \times n)\)-Matrix. Dann werden mit \((\textbf{a}_1,\ldots,\textbf{a}_m)\) die Zeilen und mit \((a^{(1)},\ldots,a^{(n)})\) die Spalten von \(\textbf{A}\) bezeichnet.

Matrix-Vektor-Multiplikation

Sei \(\textbf{A}=(a_{ij})_{i,j}\) eine \((m\times n)\)-Matrix und \(\textbf{x}\) ein \(n\)-dimensionaler Vektor. Dann ist die Multiplikation von \(\textbf{A}\) mit \(\textbf{x}\) definiert durch $$\textbf{y} = \textbf{Ax} = \left(\begin{array}{c}\textbf{a}_1'\textbf{x} \\ \vdots \\ \textbf{a}_m'\textbf{x}\end{array}\right),$$ wobei \(\textbf{a}_i'\textbf{x}\) das Skalarprodukt der \(i\)-ten Zeile von \(\textbf{A}\) mit \(\textbf{x}\) ist.

Rechenregeln

Für \((m \times n)\)-Matrizen \(\textbf{A}, \textbf{B}\), Vektoren \(\textbf{x},\textbf{y} \in \mathbb{R}^n\) und \(c \in \mathbb{R}\) gilt:

  1. \((\textbf{A} + \textbf{B})\textbf{x} = \textbf{Ax} + \textbf{Bx}\).
  2. \(\textbf{A}(\textbf{x} + \textbf{y}) = \textbf{Ax} + \textbf{Ay}\).
  3. \(\textbf{A}(c\cdot\textbf{x}) = c\cdot\textbf{Ax}\) (Linearität von \(\textbf{x} \mapsto \textbf{Ax}\)).

Matrizenmultiplikation

Sei \(\textbf{A}\) eine \((m \times n)\)-Matrix und \(\textbf{B}\) eine \((n \times r)\)-Matrix. Dann ist das Produkt von \(\textbf{A}\) und \(\textbf{B}\) definiert durch die \((m \times r)\)-Matrix $$\textbf{C} = \textbf{AB} = (c_{ij})_{i,j} \in \mathbb{R}^{m \times r},$$ wobei $$c_{ij}=\sum_{k=1}^n \, a_{i,k}b_{k,j}$$ das Skalarprodukt der \(i\)-ten Zeile von \(\textbf{A}\) mit der \(j\)-ten Spalte von \(\textbf{B}\) ist.

Zwei Matrizen heißen multiplikations-kompatibel, wenn die Spaltenzahl von \(\textbf{A}\) und die Zeilenzahl von \(\textbf{B}\) übereinstimmen, dann kann das Produkt \(\textbf{AB}\) gebildet werden. Daraus folgt aber nicht, dass auch das Produkt \(\textbf{BA}\) exisiert!

Rechenregeln

Für Matrizen \(\textbf{A}, \textbf{B}, \textbf{C}\), einen Vektor \(\textbf{x} \in \mathbb{R}^n\) und \(c \in \mathbb{R}\), gilt, sofern die Matrixprodukte gebildet werden können:

  1. \((\textbf{A} + \textbf{B})\textbf{C} = \textbf{AC} + \textbf{BC}\).
  2. \(\textbf{A}(\textbf{Bx}) = (\textbf{AB})\textbf{x}\).
  3. \(\textbf{A}(\textbf{BC}) = (\textbf{AB})\textbf{C}\).
  4. In der Regel gilt: \(\textbf{AB} \neq \textbf{BA}\) (selbst wenn beide Produkte existieren).

Rang

Der Zeilenrang einer Matrix ist die maximale Anzahl linear unabhängiger Zeilenvektoren der Matrix, analog dazu ist der Spaltenrang einer Matrix als die maximale Anzahl linear unabhängiger Spaltenvektoren der Matrix definiert. Da Zeilen- und Spaltenrang bei jeder Matrix übereinstimmen, spricht man meistens nur vom Rang \(\text{rg}(\textbf{A})\) der Matrix \(\textbf{A}\).

Determinanten

Die Determinante einer \((2 \times 2)\)-Matrix $$\textbf{A}=\left(\begin{array}{cc}a_{11} & a_{12} \\ a_{21} & a_{22}\end{array}\right)$$ ist gegeben durch $$\text{det}(\textbf{A})=a_{11}a_{22} - a_{12}a_{21}.$$

Die Determinante einer \((3 \times 3)\)-Matrix $$\textbf{A}=\left(\begin{array}{cc}a_{11} & a_{12} & a_{13} \\ a_{21} & a_{22} & a_{23} \\ a_{31} & a_{32} & a_{33}\end{array}\right)$$ ist gegeben durch $$\text{det}(\textbf{A})=a_{11}\text{det}\left(\begin{array}{cc}a_{22} & a_{23} \\ a_{32} & a_{33}\end{array}\right) - a_{12}\text{det}\left(\begin{array}{cc}a_{21} & a_{23} \\ a_{31} & a_{33}\end{array}\right) + a_{13}\text{det}\left(\begin{array}{cc}a_{21} & a_{22} \\ a_{31} & a_{32}\end{array}\right).$$

Allgemein ist die Determinante (entwickelt nach der \(i\)-ten Zeile) einer \(n \times n\)-Matrix \(\textbf{A}\) gegeben durch $$\text{det}(\textbf{A})=\sum_{j=1}^n \,(-1)^{i+j}a_{ij}\text{det}(\textbf{A}_{ij}),$$ wobei sich \(\textbf{A}_{ij}\) durch Streichen der \(i\)-ten Zeile und der \(j\)-ten Spalte ergibt. Insbesondere ist $$\text{det}(\textbf{A})=a_{11}\text{det}(\textbf{A}_{11}) - a_{12}\text{det}(\textbf{A}_{12}) \pm \cdots \pm (-1)^{n+1}\text{det}(\textbf{A}_{1n}).$$

Rechenregeln

Für multiplikations-kompatible Matrizen \(\textbf{A}, \textbf{B}\) und \(c \in \mathbb{R}\) gilt:

  1. Das Vertauschen zweier Zeilen oder Spalten ändert das Vorzeichen der Determinante.
  2. \(\text{det}(\textbf{AB})=\text{det}(\textbf{A})\text{det}(\textbf{B})\).
  3. \(\text{det}(c\textbf{A})=c^n\text{det}(\textbf{A})\).
  4. \(\text{det}(\textbf{A})=\text{det}(\textbf{A}')\).
  5. \(\text{det}(\textbf{A})=0\) genau dann, wenn \(\text{rg}(\textbf{A})<n\).
  6. \(\text{det}(\textbf{A})\neq0\) genau dann, wenn die Zeilen (Spalten) von \(\textbf{A}\) linear unabhängig sind.
  7. Die Determinante ist in jeder Zeile bzw. Spalte linear.
  8. Haben alle Elemente unterhalb der Hauptdiagonalen den Wert 0, dann gilt $$\text{det}(\textbf{A})=a_{11}a_{22}\cdot\ldots\cdot a_{nn}.$$

Wegen Regel 4 kann man auch nach der \(j\)-ten Spalte entwickeln: $$\text{det}(\textbf{A})=\sum_{i=1}^n \,(-1)^{i+j}a_{ij}\text{det}(\textbf{A}_{ij}).$$  Generell enwtickelt man nach derjenigen Zeile oder Spalte, die die meisten Nullen enthält.

Inverse Matrix

Sei \(\textbf{A}\) eine \((n \times n)\)-Matrix. Dann heißt eine Matrix \(\textbf{B}\) mit $$\textbf{AB}=\textbf{BA}=\textbf{I}$$ inverse Matrix von \(\textbf{A}\) und wird mit \(\textbf{A}^{-1}\) bezeichnet. Die Matrix \(\textbf{A}\) heißt dann invertierbar.

Eine Matrix \(\textbf{A}\) ist genau dann invertierbar, wenn \(\text{det}(\textbf{A})\neq0\).

Für eine invertierbare \((n \times n)\)-Matrix \(\textbf{A}\) und \(c \in \mathbb{R}^n\) gilt:

  1. Ist \(\textbf{AB}=\textbf{I}\) oder \(\textbf{BA}=\textbf{I}\), dann folgt \(\textbf{B}=\textbf{A}^{-1}\).
  2. \((\textbf{A}')^{-1}=(\textbf{A}^{-1})'\).
  3. \((c\textbf{A})^{-1}=\frac{1}{c}\textbf{A}^{-1}\).
  4. Ist \(\textbf{A}\) symmetrisch (d.h. \(\textbf{A}=\textbf{A}'\)), dann ist auch \(\textbf{A}^{-1}\) symmetrisch.
  5. Mit \(\textbf{A}\) und \(\textbf{B}\) sind auch \(\textbf{AB}\) und \(\textbf{BA}\) invertierbar, wobei $$(\textbf{AB})^{-1}=\textbf{B}^{-1}\textbf{A}^{-1}, \,\,\, (\textbf{BA})^{-1}=\textbf{A}^{-1}\textbf{B}^{-1}.$$