____ _ _ _ _
| _ \ ___ | |_ (_) _ __ ___ __| | (_) __ _
| |_) | / _ \ | __| | | | '_ \ / _ \ / _| | | | / _ |
| _ < | __/ | |_ | | | |_) | | __/ | (_| | | | | (_| |
|_| \_\ \___| \__| |_| | .__/ \___| \__,_| |_| \__,_|
|_|
- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b- `b
ยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏยฏ
Nabla-Operator
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
top
Der Nabla-Operator ist ein Symbol, das in der Vektor- und Tensoranalysis benutzt wird, um kontextabhรคngig einen der drei Differentialoperatoren Gradient, Divergenz oder Rotation zu notieren. Das Formelzeichen des Operators ist das Nabla-Symbol โ โ {\displaystyle \nabla } (auch โ โ โ โ {\displaystyle {\vec {\nabla }}} oder โ โ _ _ {\displaystyle {\underline {\nabla }}} , um die formale รhnlichkeit zu รผblichen vektoriellen Grรถรen zu betonen).
Der Name โNablaโ leitet sich ab von einem harfenยญรคhnlichen phรถnizischencite-ref-1[1] Saiteninstrument, das in etwa die Form dieses Zeichens hatte. Die Schreibweise wurde von William Rowan Hamilton (1805โ1865) eingefรผhrt und vom Mathematiker Peter Guthrie Tait (1831โ1901) weiterentwickelt.cite-ref-2[2] Im Englischen wird der Operator als โdelโ bezeichnet.cite-ref-3[3]
Contents
โข Definition
โข Laplace-Operator
โข Rechenregeln
โข Siehe auch
โข Weblinks
โข Literatur
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Definition
Formal ist der Nabla-Operator ein Vektor, dessen Komponenten die partiellen Ableitungsoperatoren โ โ โ โ x i {\displaystyle \textstyle {\frac {\partial }{\partial x_{i}}}} sind:
โ โ โ โ = ( โ โ โ โ x 1 , โฆ โฆ , โ โ โ โ x n ) {\displaystyle {\vec {\nabla }}=\left({\frac {\partial }{\partial x_{1}}},\ldots ,{\frac {\partial }{\partial x_{n}}}\right)}
Er kann dabei sowohl als Spaltenvektor (zum Beispiel grad) als auch als Zeilenvektor (zum Beispiel div) auftreten.cite-ref-4[4] Im dreidimensionalen kartesischen Koordinatensystem schreibt man auch:
โ โ โ โ = ( โ โ โ โ x , โ โ โ โ y , โ โ โ โ z ) = e โ โ x โ โ โ โ x + e โ โ y โ โ โ โ y + e โ โ z โ โ โ โ z {\displaystyle {\vec {\nabla }}=\left({\frac {\partial }{\partial x}},{\frac {\partial }{\partial y}},{\frac {\partial }{\partial z}}\right)={\vec {e}}_{x}{\frac {\partial }{\partial x}}+{\vec {e}}_{y}{\frac {\partial }{\partial y}}+{\vec {e}}_{z}{\frac {\partial }{\partial z}}}
Dabei sind e โ โ x {\displaystyle {\vec {e}}_{x}} , e โ โ y {\displaystyle {\vec {e}}_{y}} und e โ โ z {\displaystyle {\vec {e}}_{z}} die Einheitsvektoren des Koordinatensystems. In allgemein krummlinigen Koordinaten ฮ ฮ i {\displaystyle \Theta _{i}} sind die Einheitsvektoren durch die kontravarianten Basisvektoren zu ersetzen:
โ โ โ โ = โ โ i = 1 n g โ โ i โ โ โ โ ฮ ฮ i mit g โ โ i := grad โก โก ฮ ฮ i . {\displaystyle {\vec {\nabla }}=\sum _{i=1}^{n}{\vec {g}}^{i}{\frac {\partial }{\partial \Theta _{i}}}\quad {\text{mit}}\quad {\vec {g}}^{i}:=\operatorname {grad} \Theta _{i}\,.}
Darin ist grad {\displaystyle \operatorname {grad} } der Gradientenoperator. Bei der Anwendung dieses Nabla-Operators auf ein Vektorfeld ist zu beachten, dass die Basisvektoren in krummlinigen Koordinatensystemen im Allgemeinen von den Koordinaten ฮ ฮ i {\displaystyle \Theta _{i}} abhรคngen und ebenfalls zu differenzieren sind.
Gerechnet wird mit dem Nabla-Operator wie mit einem Vektor, wobei das โProduktโ von beispielsweise โ โ โ โ x i {\displaystyle \textstyle {\frac {\partial }{\partial x_{i}}}} mit einer rechts davon stehenden Funktion f {\displaystyle f} als partielle Ableitung โ โ f โ โ x i {\displaystyle \textstyle {\frac {\partial f}{\partial x_{i}}}} interpretiert wird.
Darstellung anderer Differentialoperatoren
Im n-dimensionalen Raum
Sei D โ โ R n {\displaystyle D\subset \mathbb {R} ^{n}} eine offene Teilmenge, f : : D โ โ R {\displaystyle f\colon D\to \mathbb {R} } eine differenzierbare Funktion und V โ โ = ( V 1 , โฆ โฆ , V n ) โค โค : : D โ โ R n {\displaystyle {\vec {V}}=(V_{1},\dots ,V_{n})^{\top }\colon D\to \mathbb {R} ^{n}} ein differenzierbares Vektorfeld. Das hochgestellte โฌ bezeichnet die Transposition.
Das (formale) Produkt von โ โ โ โ {\displaystyle {\vec {\nabla }}} mit der Funktion f {\displaystyle f} ergibt deren Gradienten:
โ โ โ โ f = grad โก โก f = ( โ โ f โ โ x 1 , โฆ โฆ , โ โ f โ โ x n ) โค โค . {\displaystyle {\vec {\nabla }}f=\operatorname {grad} f=\left({\frac {\partial f}{\partial x_{1}}},\ldots ,{\frac {\partial f}{\partial x_{n}}}\right)^{\top }\,.}
Das transponierte (formale) dyadische Produkt โ โ โ {\displaystyle \otimes } โ von โ โ โ โ {\displaystyle {\vec {\nabla }}} mit dem Vektorfeld V โ โ {\displaystyle {\vec {V}}} ergibt dessen Gradienten oder Jacobi-Matrix:
( โ โ โ โ โ โ V โ โ ) โค โค = grad โก โก V โ โ = J V โ โ = ( โ โ V 1 โ โ x 1 โฆ โฆ โ โ V 1 โ โ x n โฎ โฎ โฑ โฑ โฎ โฎ โ โ V n โ โ x 1 โฆ โฆ โ โ V n โ โ x n ) . {\displaystyle ({\vec {\nabla }}\otimes {\vec {V}})^{\top }=\operatorname {grad} {\vec {V}}=J_{\vec {V}}={\begin{pmatrix}{\frac {\partial V_{1}}{\partial x_{1}}}&\ldots &{\frac {\partial V_{1}}{\partial x_{n}}}\\\vdots &\ddots &\vdots \\{\frac {\partial V_{n}}{\partial x_{1}}}&\ldots &{\frac {\partial V_{n}}{\partial x_{n}}}\end{pmatrix}}\,.}
Das (formale) Skalarprodukt mit dem Vektorfeld V โ โ {\displaystyle {\vec {V}}} ergibt dessen Divergenz:
โ โ โ โ โ
โ
V โ โ = div โก โก V โ โ = โ โ i = 1 n โ โ V i โ โ x i . {\displaystyle {\vec {\nabla }}\cdot {\vec {V}}=\operatorname {div} {\vec {V}}=\sum _{i=1}^{n}{\frac {\partial V_{i}}{\partial x_{i}}}\,.}
Sie ist die Spur des Gradienten.
Laplace-Operator
Das (formale) Skalarprodukt โ โ โ โ 2 {\displaystyle {\vec {\nabla }}^{2}} von โ โ โ โ {\displaystyle {\vec {\nabla }}} mit sich selbst ergibt den Laplace-Operator ฮ ฮ {\displaystyle \Delta } , denn es gilt
โ โ 2 = โ โ โ โ โ
โ
โ โ โ โ = โ โ i = 1 n โ โ 2 โ โ x i 2 = ฮ ฮ . {\displaystyle \nabla ^{2}={\vec {\nabla }}\cdot {\vec {\nabla }}=\sum _{i=1}^{n}{\frac {\partial ^{2}}{\partial x_{i}^{2}}}=\Delta \,.}
Manche Autoren verwenden das Symbol โ โ 2 {\displaystyle \nabla ^{2}} oder โ โ โ โ โ โ โ โ {\displaystyle {\vec {\nabla }}{\vec {\nabla }}} auch fรผr die Hesse-Matrix, deshalb sollte immer der Kontext angegeben werden.
Richtungsableitung
Bei einem gegebenen Vektor H โ โ {\displaystyle {\vec {H}}} kann mit dem Operator
D H โ โ := H โ โ โ
โ
โ โ โ โ = โ โ i = 1 n H i โ โ โ โ x i {\displaystyle \operatorname {D} _{\vec {H}}:={\vec {H}}\cdot {\vec {\nabla }}=\sum _{i=1}^{n}H_{i}{\frac {\partial }{\partial x_{i}}}}
die Richtungsableitung von differenzierbaren Funktionen f {\displaystyle f} in Richtung des Vektors H โ โ {\displaystyle {\vec {H}}} berechnet werden:
D H โ โ โก โก f = ( H โ โ โ
โ
โ โ โ โ ) f = H โ โ โ
โ
โ โ โ โ f = H โ โ โ
โ
grad โก โก ( f ) = grad โก โก ( f ) โ
โ
H โ โ {\displaystyle \operatorname {D} _{\vec {H}}f=({\vec {H}}\cdot {\vec {\nabla }})f={\vec {H}}\cdot {\vec {\nabla }}f={\vec {H}}\cdot \operatorname {grad} (f)=\operatorname {grad} (f)\cdot {\vec {H}}}
siehe den Zusammenhang zwischen Gradient und Richtungsableitung. Ist die Funktion ein Vektorfeld V โ โ {\displaystyle {\vec {V}}} , dann berechnet sich das Produkt aus der Jacobi-Matrix des Feldes und dem Vektor:
D H โ โ โก โก V โ โ = ( H โ โ โ
โ
โ โ โ โ ) V โ โ โ โ = H โ โ โ
โ
( โ โ โ โ โ โ V โ โ ) = ( โ โ โ โ โ โ V โ โ ) โค โค โ
โ
H โ โ = g r a d ( V โ โ ) โ โ โ
โ
H โ โ = ( H 1 โ โ โ โ x 1 V 1 + โฆ โฆ + H n โ โ โ โ x n V 1 โฎ โฎ โฑ โฑ โฎ โฎ H 1 โ โ โ โ x 1 V n + โฆ โฆ + H n โ โ โ โ x n V n ) = ( โ โ V 1 โ โ x 1 โฆ โฆ โ โ V 1 โ โ x n โฎ โฎ โฑ โฑ โฎ โฎ โ โ V n โ โ x 1 โฆ โฆ โ โ V n โ โ x n ) ( H 1 โฎ โฎ H n ) {\displaystyle {\begin{aligned}\operatorname {D} _{\vec {H}}{\vec {V}}&=\underbrace {({\vec {H}}\cdot {\vec {\nabla }}){\vec {V}}} ={\vec {H}}\cdot ({\vec {\nabla }}\otimes {\vec {V}})=({\vec {\nabla }}\otimes {\vec {V}})^{\top }\cdot {\vec {H}}\!\!\!\!\!\!\!\!\!\!&=&\underbrace {\mathrm {grad} ({\vec {V}})} \cdot {\vec {H}}\\&={\begin{pmatrix}H_{1}{\frac {\partial }{\partial x_{1}}}V_{1}+&\ldots &+H_{n}{\frac {\partial }{\partial x_{n}}}V_{1}\\\vdots &\ddots &\vdots \\H_{1}{\frac {\partial }{\partial x_{1}}}V_{n}+&\ldots &+H_{n}{\frac {\partial }{\partial x_{n}}}V_{n}\end{pmatrix}}\!\!\!\!\!\!\!\!\!\!&=&{\begin{pmatrix}{\frac {\partial V_{1}}{\partial x_{1}}}&\ldots &{\frac {\partial V_{1}}{\partial x_{n}}}\\\vdots &\ddots &\vdots \\{\frac {\partial V_{n}}{\partial x_{1}}}&\ldots &{\frac {\partial V_{n}}{\partial x_{n}}}\end{pmatrix}}{\begin{pmatrix}H_{1}\\\vdots \\H_{n}\end{pmatrix}}\end{aligned}}}
siehe Vektorgradient und die Anwendung in der Kontinuumsmechanik unten.
Im dreidimensionalen Raum
Sei D โ โ R 3 {\displaystyle D\subset \mathbb {R} ^{3}} nun eine offene Teilmenge, f : : D โ โ R {\displaystyle f\colon D\to \mathbb {R} } eine differenzierbare Funktion und V โ โ = ( V x , V y , V z ) โค โค : : D โ โ R 3 {\displaystyle {\vec {V}}=(V_{x},V_{y},V_{z})^{\top }\colon D\to \mathbb {R} ^{3}} ein differenzierbares Vektorfeld. Die Indizes โฆx,y,z bezeichnen hier die Vektorkomponenten und keine Ableitungen. Im dreidimensionalen Raum R 3 {\displaystyle \mathbb {R} ^{3}} mit den kartesischen Koordinaten x {\displaystyle x} , y {\displaystyle y} , z {\displaystyle z} stellen sich die obigen Formeln wie folgt dar:
Der Nabla-Operator angewandt auf das Skalarfeld f {\displaystyle f} ergibt den Gradienten des Skalarfeldes
grad โก โก f = โ โ โ โ f = ( โ โ f โ โ x , โ โ f โ โ y , โ โ f โ โ z ) โค โค = โ โ f โ โ x e โ โ x + โ โ f โ โ y e โ โ y + โ โ f โ โ z e โ โ z . {\displaystyle \operatorname {grad} f={\vec {\nabla }}f=\left({\frac {\partial f}{\partial x}},{\frac {\partial f}{\partial y}},{\frac {\partial f}{\partial z}}\right)^{\top }={\frac {\partial f}{\partial x}}{\vec {e}}_{x}+{\frac {\partial f}{\partial y}}{\vec {e}}_{y}+{\frac {\partial f}{\partial z}}{\vec {e}}_{z}\,.}
Das Ergebnis ist ein Vektorfeld. Hierbei sind e โ โ x , e โ โ y , e โ โ z {\displaystyle {\vec {e}}_{x},\,{\vec {e}}_{y},\,{\vec {e}}_{z}} die Einheitsvektoren des R 3 {\displaystyle \mathbb {R} ^{3}} .
Der Nabla-Operator angewandt auf das Vektorfeld V โ โ {\displaystyle {\vec {V}}} ergibt die Divergenz des Vektorfeldes als formales Skalarprodukt mit dem Vektorfeld zu
div โก โก V โ โ = โ โ โ โ โ
โ
V โ โ = โ โ V x โ โ x + โ โ V y โ โ y + โ โ V z โ โ z , {\displaystyle \operatorname {div} {\vec {V}}={\vec {\nabla }}\cdot {\vec {V}}={\frac {\partial V_{x}}{\partial x}}+{\frac {\partial V_{y}}{\partial y}}+{\frac {\partial V_{z}}{\partial z}},}
also ein Skalarfeld.
Eine Besonderheit des dreidimensionalen Raums ist die Rotation eines Vektorfelds. Sie ergibt sich durch (rechtsseitige) Verknรผpfung รผber das formale Kreuzprodukt als
rot โก โก V โ โ = โ โ โ โ ร ร V โ โ = ( โ โ V z โ โ y โ โ โ โ V y โ โ z โ โ V x โ โ z โ โ โ โ V z โ โ x โ โ V y โ โ x โ โ โ โ V x โ โ y ) {\displaystyle \operatorname {rot} {\vec {V}}={\vec {\nabla }}\times {\vec {V}}={\begin{pmatrix}{\frac {\partial V_{z}}{\partial y}}-{\frac {\partial V_{y}}{\partial z}}\\{\frac {\partial V_{x}}{\partial z}}-{\frac {\partial V_{z}}{\partial x}}\\{\frac {\partial V_{y}}{\partial x}}-{\frac {\partial V_{x}}{\partial y}}\end{pmatrix}}}
also wieder ein Vektorfeld.
Zylinderkoordinaten ( ฯ ฯ , ฯ ฯ , z ) {\displaystyle (\rho ,\varphi ,z)} und Kugelkoordinaten ( r , ฮธ ฮธ , ฯ ฯ ) {\displaystyle (r,\theta ,\varphi )} sind Beispiele fรผr krummlinige Koordinaten. Die Formeln fรผr den Gradienten in Zylinder- und Kugelkoordinaten ergeben sich aus den Nabla-Operatoren
in Zylinderkoordinaten: โ โ โ โ = e โ โ ฯ ฯ โ โ โ โ ฯ ฯ + 1 ฯ ฯ e โ โ ฯ ฯ โ โ โ โ ฯ ฯ + e โ โ z โ โ โ โ z bzw. Kugelkoordinaten: โ โ โ โ = e โ โ r โ โ โ โ r + 1 r e โ โ ฮธ ฮธ โ โ โ โ ฮธ ฮธ + 1 r sin โก โก ฮธ ฮธ e โ โ ฯ ฯ โ โ โ โ ฯ ฯ . {\displaystyle {\begin{aligned}{\text{in Zylinderkoordinaten:}}\quad &{\vec {\nabla }}={\vec {e}}_{\rho }{\frac {\partial }{\partial \rho }}+{\frac {1}{\rho }}{\vec {e}}_{\varphi }{\frac {\partial }{\partial \varphi }}+{\vec {e}}_{z}{\frac {\partial }{\partial z}}\\{\text{bzw. Kugelkoordinaten:}}\quad &{\vec {\nabla }}={\vec {e}}_{r}{\frac {\partial }{\partial r}}+{\frac {1}{r}}{\vec {e}}_{\theta }{\frac {\partial }{\partial \theta }}+{\frac {1}{r\sin \theta }}{\vec {e}}_{\varphi }{\frac {\partial }{\partial \varphi }}\,.\end{aligned}}}
Bei der Anwendung auf ein Vektorfeld ist wie oben erwรคhnt zu beachten, dass die Basisvektoren in krummlinigen Koordinatensystemen im Allgemeinen wie auch hier von den Koordinaten abhรคngen und ebenfalls zu differenzieren sind. Beispielsweise ergibt sich fรผr die Divergenz eines Vektorfeldes in Zylinderkoordinaten, wo die Basisvektoren e โ โ ฯ ฯ {\displaystyle {\vec {e}}_{\rho }} und e โ โ ฯ ฯ {\displaystyle {\vec {e}}_{\varphi }} vom Winkel ฯ abhรคngen und โ โ โ โ ฯ ฯ e โ โ ฯ ฯ = e โ โ ฯ ฯ , โ โ โ โ ฯ ฯ e โ โ ฯ ฯ = โ โ e โ โ ฯ ฯ {\displaystyle {\tfrac {\partial }{\partial \varphi }}{\vec {e}}_{\rho }={\vec {e}}_{\varphi },\,{\tfrac {\partial }{\partial \varphi }}{\vec {e}}_{\varphi }=-{\vec {e}}_{\rho }} gilt:
div โก โก V โ โ = โ โ โ โ โ
โ
V โ โ = ( e โ โ ฯ ฯ โ โ โ โ ฯ ฯ + 1 ฯ ฯ e โ โ ฯ ฯ โ โ โ โ ฯ ฯ + e โ โ z โ โ โ โ z ) โ
โ
( V ฯ ฯ e โ โ ฯ ฯ + V ฯ ฯ e โ โ ฯ ฯ + V z e โ โ z ) = โ โ โ โ ฯ ฯ V ฯ ฯ + 1 ฯ ฯ e โ โ ฯ ฯ โ
โ
โ โ โ โ ฯ ฯ ( V ฯ ฯ e โ โ ฯ ฯ + V ฯ ฯ e โ โ ฯ ฯ + V z e โ โ z ) + โ โ V z โ โ z = โ โ โ โ ฯ ฯ V ฯ ฯ + 1 ฯ ฯ e โ โ ฯ ฯ โ
โ
( โ โ V ฯ ฯ โ โ ฯ ฯ e โ โ ฯ ฯ + V ฯ ฯ e โ โ ฯ ฯ + โ โ V ฯ ฯ โ โ ฯ ฯ e โ โ ฯ ฯ โ โ V ฯ ฯ e โ โ ฯ ฯ + โ โ V z โ โ ฯ ฯ e โ โ z ) + โ โ V z โ โ z = โ โ V ฯ ฯ โ โ ฯ ฯ + 1 ฯ ฯ V ฯ ฯ + 1 ฯ ฯ โ โ V ฯ ฯ โ โ ฯ ฯ + โ โ V z โ โ z . {\displaystyle {\begin{aligned}\operatorname {div} {\vec {V}}=&{\vec {\nabla }}\cdot {\vec {V}}=\left({\vec {e}}_{\rho }{\frac {\partial }{\partial \rho }}+{\frac {1}{\rho }}{\vec {e}}_{\varphi }{\frac {\partial }{\partial \varphi }}+{\vec {e}}_{z}{\frac {\partial }{\partial z}}\right)\cdot (V_{\rho }{\vec {e}}_{\rho }+V_{\varphi }{\vec {e}}_{\varphi }+V_{z}{\vec {e}}_{z})\\=&{\frac {\partial }{\partial \rho }}V_{\rho }+{\frac {1}{\rho }}{\vec {e}}_{\varphi }\cdot {\frac {\partial }{\partial \varphi }}(V_{\rho }{\vec {e}}_{\rho }+V_{\varphi }{\vec {e}}_{\varphi }+V_{z}{\vec {e}}_{z})+{\frac {\partial V_{z}}{\partial z}}\\=&{\frac {\partial }{\partial \rho }}V_{\rho }+{\frac {1}{\rho }}{\vec {e}}_{\varphi }\cdot \left({\frac {\partial V_{\rho }}{\partial \varphi }}{\vec {e}}_{\rho }+V_{\rho }{\vec {e}}_{\varphi }+{\frac {\partial V_{\varphi }}{\partial \varphi }}{\vec {e}}_{\varphi }-V_{\varphi }{\vec {e}}_{\rho }+{\frac {\partial V_{z}}{\partial \varphi }}{\vec {e}}_{z}\right)+{\frac {\partial V_{z}}{\partial z}}\\=&{\frac {\partial V_{\rho }}{\partial \rho }}+{\frac {1}{\rho }}V_{\rho }+{\frac {1}{\rho }}{\frac {\partial V_{\varphi }}{\partial \varphi }}+{\frac {\partial V_{z}}{\partial z}}\,.\end{aligned}}}
Notation mit Subskript
Wirkt der Nablaoperator nur auf bestimmte Komponenten einer Funktion mit einem mehrdimensionalen Argument, so wird dies durch ein Subskript angedeutet. Fรผr eine Funktion f ( r โ โ , t ) {\displaystyle f({\vec {r}},t)} mit r โ โ = ( x 1 , x 2 , โฆ โฆ , x n ) {\displaystyle {\vec {r}}=(x_{1},x_{2},\dotsc ,x_{n})} beispielsweise ist
โ โ โ โ r โ โ f = ( โ โ f โ โ x 1 , โ โ f โ โ x 2 , โฆ โฆ , โ โ f โ โ x n ) โค โค {\displaystyle {\vec {\nabla }}_{\vec {r}}f=\left({\frac {\partial f}{\partial x_{1}}},{\frac {\partial f}{\partial x_{2}}},\dots ,{\frac {\partial f}{\partial x_{n}}}\right)^{\top }}
im Gegensatz zu
โ โ โ โ f = ( โ โ f โ โ x 1 , โ โ f โ โ x 2 , โฆ โฆ , โ โ f โ โ x n , โ โ f โ โ t ) โค โค . {\displaystyle {\vec {\nabla }}f=\left({\frac {\partial f}{\partial x_{1}}},{\frac {\partial f}{\partial x_{2}}},\dots ,{\frac {\partial f}{\partial x_{n}}},{\frac {\partial {f}}{\partial t}}\right)^{\top }\,.}
Diese Bezeichnung ist รผblich, wenn mit dem Nabla-Symbol das einfache Differential (d. h. die einzeilige Jacobi-Matrix) bzw. ein Teil davon bezeichnet wird.
Gelegentlich tritt alternativ fรผr die Schreibweise mit dem Nabla-Symbol โ โ โ โ r โ โ {\displaystyle {\vec {\nabla }}_{\vec {r}}} die Schreibweise โ โ โ โ r โ โ {\displaystyle {\tfrac {\partial }{\partial {\vec {r}}}}} auf.cite-ref-5[5]
Darstellung als Quaternion
โ
Hauptartikel
:
Quaternion
Sir William Rowan Hamiltoncite-ref-beutelspacher-la-7-30-6-0[6] definierte den Nabla-Operator als reine Quaternion
โ โ := i โ โ โ โ x + j โ โ โ โ y + k โ โ โ โ z {\displaystyle \nabla :=\mathrm {i} \,{\frac {\partial }{\partial x}}+\mathrm {j} \,{\frac {\partial }{\partial y}}+\mathrm {k} \,{\frac {\partial }{\partial z}}}
mit den komplex-imaginรคren Einheiten i {\displaystyle \mathrm {i} } , j {\displaystyle \mathrm {j} } und k {\displaystyle \mathrm {k} } , die durch die Hamilton-Regeln i 2 = j 2 = k 2 = i j k = โ โ 1 {\displaystyle \mathrm {i^{2}=j^{2}=k^{2}=i\,j\,k=-1} } nicht kommutativ verknรผpft sind. Beispielsweise gilt j k = โ โ k j = i {\displaystyle \mathrm {j\,k=-k\,j=i} } .
Anwendung auf eine reellwertige Funktion f {\displaystyle f} (formale Multiplikation) liefert die quaternionische Entsprechung fรผr deren Gradient und Laplace-Ableitung:
โ โ f = i โ โ f โ โ x + j โ โ f โ โ y + k โ โ f โ โ z = grad โก โก f โ โ โ โ f = โ โ โ
โ
โ โ f = โ โ โ โ 2 f โ โ x 2 โ โ โ โ 2 f โ โ y 2 โ โ โ โ 2 f โ โ z 2 = โ โ ฮ ฮ f {\displaystyle {\begin{aligned}\nabla f=&\mathrm {i} \,{\frac {\partial f}{\partial x}}+\mathrm {j} \,{\frac {\partial f}{\partial y}}+\mathrm {k} \,{\frac {\partial f}{\partial z}}=\operatorname {grad} f\\\nabla \nabla f=&\nabla \cdot \nabla f=-{\frac {\partial ^{2}f}{\partial x^{2}}}-{\frac {\partial ^{2}f}{\partial y^{2}}}-{\frac {\partial ^{2}f}{\partial z^{2}}}=-\Delta f\end{aligned}}}
Anwendung auf eine reine Quaternion q = i u + j v + k w {\displaystyle q=\mathrm {i} \,u+\mathrm {j} \,v+\mathrm {k} \,w} (formale Multiplikation) liefert:
โ โ q = โ โ โ โ u โ โ x โ โ โ โ v โ โ y โ โ โ โ w โ โ z + i ( โ โ w โ โ y โ โ โ โ v โ โ z ) + j ( โ โ u โ โ z โ โ โ โ w โ โ x ) + k ( โ โ v โ โ x โ โ โ โ u โ โ y ) = โ โ โ โ โ
โ
q + โ โ ร ร q = โ โ div โก โก q + rot โก โก q {\displaystyle {\begin{aligned}\nabla q=&-{\frac {\partial u}{\partial x}}-{\frac {\partial v}{\partial y}}-{\frac {\partial w}{\partial z}}+\mathrm {i} \,\left({\frac {\partial w}{\partial y}}-{\frac {\partial v}{\partial z}}\right)+\mathrm {j} \,\left({\frac {\partial u}{\partial z}}-{\frac {\partial w}{\partial x}}\right)+\mathrm {k} \,\left({\frac {\partial v}{\partial x}}-{\frac {\partial u}{\partial y}}\right)\\=&-\nabla \cdot q+\nabla \times q=-\operatorname {div} q+\operatorname {rot} q\end{aligned}}}
Die hier benutzten Definitionen des Skalarprodukts und Kreuzprodukts von Quaternionen sind im Hauptartikel nachzuschlagen.
Rechenregeln
Rechenregeln fรผr den Nabla-Operator lassen sich formal aus den Rechenregeln fรผr Skalar- und Kreuzprodukt zusammen mit den Ableitungsregeln herleiten. Dabei muss man die Produktregel anwenden, wenn der Nabla-Operator links von einem Produkt steht.
Sind ฯ ฯ {\displaystyle \psi } und ฯ ฯ {\displaystyle \varphi } differenzierbare Skalarfelder (Funktionen) und A โ โ {\displaystyle {\vec {A}}} sowie B โ โ {\displaystyle {\vec {B}}} differenzierbare Vektorfelder, so gilt:
โ โ โ โ ฯ ฯ ( ฯ ฯ ) = d ฯ ฯ d ฯ ฯ โ โ โ โ ฯ ฯ {\displaystyle {\vec {\nabla }}\varphi (\psi )={\frac {\mathrm {d} \varphi }{\mathrm {d} \psi }}{\vec {\nabla }}\psi } (Kettenregel fรผr Gradient)
โ โ โ โ ( ฯ ฯ ฯ ฯ ) = ฯ ฯ โ โ โ โ ฯ ฯ + ฯ ฯ โ โ โ โ ฯ ฯ {\displaystyle {\vec {\nabla }}(\psi \varphi )=\psi {\vec {\nabla }}\varphi +\varphi {\vec {\nabla }}\psi } (Produktregel fรผr Gradient)
โ โ โ โ ( A โ โ โ
โ
B โ โ ) = ( A โ โ โ
โ
โ โ โ โ ) B โ โ + ( B โ โ โ
โ
โ โ โ โ ) A โ โ + A โ โ ร ร ( โ โ โ โ ร ร B โ โ ) + B โ โ ร ร ( โ โ โ โ ร ร A โ โ ) {\displaystyle {\vec {\nabla }}({\vec {A}}\cdot {\vec {B}})=({\vec {A}}\cdot {\vec {\nabla }}){\vec {B}}+({\vec {B}}\cdot {\vec {\nabla }}){\vec {A}}+{\vec {A}}\times ({\vec {\nabla }}\times {\vec {B}})+{\vec {B}}\times ({\vec {\nabla }}\times {\vec {A}})}
โ โ โ โ โ
โ
( ฯ ฯ A โ โ ) = ฯ ฯ โ โ โ โ โ
โ
A โ โ + A โ โ โ
โ
โ โ โ โ ฯ ฯ {\displaystyle {\vec {\nabla }}\cdot (\varphi {\vec {A}})=\varphi {\vec {\nabla }}\cdot {\vec {A}}+{\vec {A}}\cdot {\vec {\nabla }}\varphi }
โ โ โ โ โ
โ
( A โ โ ร ร B โ โ ) = B โ โ โ
โ
( โ โ โ โ ร ร A โ โ ) โ โ A โ โ โ
โ
( โ โ โ โ ร ร B โ โ ) {\displaystyle {\vec {\nabla }}\cdot ({\vec {A}}\times {\vec {B}})={\vec {B}}\cdot ({\vec {\nabla }}\times {\vec {A}})-{\vec {A}}\cdot ({\vec {\nabla }}\times {\vec {B}})}
โ โ โ โ โ
โ
( โ โ โ โ ฯ ฯ ) = d i v ( g r a d โก โก ฯ ฯ ) = ฮ ฮ ฯ ฯ {\displaystyle {\vec {\nabla }}\cdot ({\vec {\nabla }}\varphi )=\operatorname {div(grad} \varphi )=\Delta \varphi } (siehe auch Laplace-Operator)
โ โ โ โ โ
โ
( โ โ โ โ ร ร A โ โ ) = d i v ( r o t โก โก A โ โ ) = 0 {\displaystyle {\vec {\nabla }}\cdot ({\vec {\nabla }}\times {\vec {A}})=\operatorname {div(rot} {\vec {A}})=0}
โ โ โ โ ร ร ( โ โ โ โ ฯ ฯ ) = r o t ( g r a d โก โก ฯ ฯ ) = 0 {\displaystyle {\vec {\nabla }}\times ({\vec {\nabla }}\varphi )=\operatorname {rot(grad} \varphi )=0}
โ โ โ โ ร ร ฯ ฯ A โ โ = ฯ ฯ โ โ โ โ ร ร A โ โ โ โ A โ โ ร ร โ โ โ โ ฯ ฯ {\displaystyle {\vec {\nabla }}\times \varphi {\vec {A}}=\varphi {\vec {\nabla }}\times {\vec {A}}-{\vec {A}}\times {\vec {\nabla }}\varphi }
โ โ โ โ ร ร ( A โ โ ร ร B โ โ ) = ( B โ โ โ
โ
โ โ โ โ ) A โ โ โ โ B โ โ ( โ โ โ โ โ
โ
A โ โ ) + A โ โ ( โ โ โ โ โ
โ
B โ โ ) โ โ ( A โ โ โ
โ
โ โ โ โ ) B โ โ {\displaystyle {\vec {\nabla }}\times ({\vec {A}}\times {\vec {B}})=({\vec {B}}\cdot {\vec {\nabla }}){\vec {A}}-{\vec {B}}({\vec {\nabla }}\cdot {\vec {A}})+{\vec {A}}({\vec {\nabla }}\cdot {\vec {B}})-({\vec {A}}\cdot {\vec {\nabla }}){\vec {B}}}
โ โ โ โ ร ร ( โ โ โ โ ร ร A โ โ ) = r o t ( r o t โก โก A โ โ ) = g r a d ( d i v โก โก A โ โ ) โ โ ฮ ฮ A โ โ {\displaystyle {\vec {\nabla }}\times ({\vec {\nabla }}\times {\vec {A}})=\operatorname {rot(rot} {\vec {A}})=\operatorname {grad(div} {\vec {A}})-\Delta {\vec {A}}} (siehe auch vektorieller Laplace-Operator)
Anwendung in der Kontinuumsmechanik
In der Kontinuumsmechanik wird der Nabla-Operator dazu verwendet, zusรคtzlich zu den oben genannten Operatoren den Gradienten eines Vektorfeldes und die Divergenz sowie Rotation eines Tensorfeldes zu definieren. Hier kann der Nabla-Operator gelegentlich auch nach links wirken.cite-ref-7[7]
Die Darstellung erfolgt wegen der Wichtigkeit der Rotation fรผr die Kontinuumsmechanik in drei Dimensionen. Sei also D โ โ R 3 {\displaystyle D\subset \mathbb {R} ^{3}} eine offene Teilmenge, V โ โ = ( V x , V y , V z ) โค โค : : D โ โ R 3 {\displaystyle {\vec {V}}=(V_{x},V_{y},V_{z})^{\top }\colon D\to \mathbb {R} ^{3}} ein differenzierbares Vektorfeld mit Komponenten Vx,y,z, die wie รผblich nach dem Schema xโ1, yโ2 und zโ3 durchnummeriert werden, und T : : D โ โ R 3 ร ร 3 {\displaystyle \mathbf {T} \colon D\to \mathbb {R} ^{3\times 3}} ein differenzierbares Tensorfeld zweiter Stufe mit Komponenten T i j , i , j = 1 , 2 , 3 {\displaystyle T_{ij}\,,\;i,j=1,2,3} bezรผglich eines kartesischen Koordinatensystems.
Das transponierte dyadische Produkt des Nabla-Operators mit einem Vektorfeld V โ โ {\displaystyle {\vec {V}}} ergibt โ wie oben dargelegt โ den Gradienten eines Vektorfeldes
( โ โ โ โ โ โ V โ โ ) โค โค = grad โก โก ( V โ โ ) := โ โ j = 1 3 โ โ V โ โ โ โ x j โ โ e โ โ j = โ โ i = 1 3 e โ โ i โ โ grad โก โก ( V i ) = โ โ i , j = 1 3 โ โ V i โ โ x j e โ โ i โ โ e โ โ j {\displaystyle ({\vec {\nabla }}\otimes {\vec {V}})^{\top }=\operatorname {grad} ({\vec {V}}):=\sum _{j=1}^{3}{\frac {\partial {\vec {V}}}{\partial x_{j}}}\otimes {\vec {e}}_{j}=\sum _{i=1}^{3}{\vec {e}}_{i}\otimes \operatorname {grad} (V_{i})=\sum _{i,j=1}^{3}{\frac {\partial V_{i}}{\partial x_{j}}}{\vec {e}}_{i}\otimes {\vec {e}}_{j}}
also ein Tensorfeld zweiter Stufe. Der so definierte Gradient stimmt mit der Frรฉchet-Ableitung รผberein:
grad โก โก ( V โ โ ) โ
โ
h โ โ = d d s V โ โ ( x โ โ + s h โ โ ) | s = 0 = lim s โ โ 0 V โ โ ( x โ โ + s h โ โ ) โ โ V โ โ ( x โ โ ) s fรผr alle x โ โ , h โ โ โ โ D {\displaystyle \operatorname {grad} ({\vec {V}})\cdot {\vec {h}}=\left.{\frac {\mathrm {d} }{\mathrm {d} s}}{\vec {V}}({\vec {x}}+s{\vec {h}})\right|_{s=0}=\lim _{s\rightarrow 0}{\frac {{\vec {V}}({\vec {x}}+s{\vec {h}})-{\vec {V}}({\vec {x}})}{s}}\quad {\text{fรผr alle}}\;{\vec {x}},{\vec {h}}\in D}
und nรคhert das Vektorfeld in der Nรคhe eines Punktes x โ โ {\displaystyle {\vec {x}}} linear an:
V โ โ ( y โ โ ) โ โ V โ โ ( x โ โ ) = g r a d ( V โ โ ) โ
โ
( y โ โ โ โ x โ โ ) + O ( | y โ โ โ โ x โ โ | ) {\displaystyle {\vec {V}}({\vec {y}})-{\vec {V}}({\vec {x}})=\mathrm {grad} ({\vec {V}})\cdot ({\vec {y}}-{\vec {x}})+{\mathcal {O}}(|{\vec {y}}-{\vec {x}}|)} wenn y โ โ โ โ x โ โ {\displaystyle {\vec {y}}\to {\vec {x}}}
Das Landau-Symbol O ( x ) {\displaystyle {\mathcal {O}}(x)} stellt eine Grรถรe dar, die langsamer wรคchst als ihr Argument x {\displaystyle x} .
Das linksseitige Skalarprodukt des Nabla-Operators mit einem transponierten Tensorfeld zweiter Stufe ergibt formal die Divergenz des Tensorfeldes:cite-ref-hbphys-8-0[8]
โ โ โ โ โ
โ
( T โค โค ) = div โก โก ( T ) = โ โ k = 1 3 โ โ T โ โ x k โ
โ
e โ โ k = โ โ i , j 3 โ โ T i j โ โ x j e โ โ i {\displaystyle {\vec {\nabla }}\cdot (\mathbf {T} ^{\top })=\operatorname {div} (\mathbf {T} )=\sum _{k=1}^{3}{\frac {\partial \mathbf {T} }{\partial x_{k}}}\cdot {\vec {e}}_{k}=\sum _{i,j}^{3}{\frac {\partial T_{ij}}{\partial x_{j}}}{\vec {e}}_{i}}
also ein Vektorfeld. Sie entspricht der Definition
d i v ( T ) โ
โ
c โ โ = d i v ( T โค โค โ
โ
c โ โ ) fรผr alle c โ โ โ โ V {\displaystyle \mathrm {div} (\mathbf {T} )\cdot {\vec {c}}=\mathrm {div} \left(\mathbf {T} ^{\top }\cdot {\vec {c}}\right)\quad {\text{fรผr alle}}\;{\vec {c}}\in \mathbb {V} } .
Es wird auch die nicht-transponierte Version benutzt, โ โ โ โ โ
โ
T {\displaystyle {\vec {\nabla }}\cdot \mathbf {T} } , die bei symmetrischen Tensoren zum selben Ergebnis fรผhrt.
Das Kreuzprodukt des Nabla-Operators mit einem transponierten Tensor zweiter Stufe liefert dessen Rotation:cite-ref-hbphys-8-1[8]
โ โ โ โ ร ร ( T โค โค ) = rot โก โก ( T ) = โ โ i , j , l = 1 3 e โ โ i ร ร โ โ T l j โ โ x i ( e โ โ j โ โ e โ โ l ) = โ โ i , j , k , l = 1 3 ฯต ฯต i j k โ โ T l j โ โ x i ( e โ โ k โ โ e โ โ l ) {\displaystyle {\vec {\nabla }}\times (\mathbf {T} ^{\top })=\operatorname {rot} (\mathbf {T} )=\sum _{i,j,l=1}^{3}{\vec {e}}_{i}\times {\frac {\partial T_{lj}}{\partial x_{i}}}({\vec {e}}_{j}\otimes {\vec {e}}_{l})=\sum _{i,j,k,l=1}^{3}\epsilon _{ijk}{\frac {\partial T_{lj}}{\partial x_{i}}}({\vec {e}}_{k}\otimes {\vec {e}}_{l})}
also ein Tensorfeld zweiter Stufe. Darin ist ฯตijk = (รชi ร รชj) ยท รชk das Permutationssymbol. Obige Form der Rotation entspricht der Definition
r o t ( T ) โ
โ
c โ โ = r o t ( T โค โค โ
โ
c โ โ ) fรผr alle c โ โ โ โ V {\displaystyle \mathrm {rot} (\mathbf {T} )\cdot {\vec {c}}=\mathrm {rot} \left(\mathbf {T} ^{\top }\cdot {\vec {c}}\right)\quad {\text{fรผr alle}}\;{\vec {c}}\in \mathbb {V} }
Es wird auch die Form ohne Transposition benutzt, โ โ โ โ ร ร T {\displaystyle {\vec {\nabla }}\times \mathbf {T} } , die bei symmetrischen Tensoren zum selben Ergebnis fรผhrt.
Siehe auch
Weblinks
Wikibooks: Formelsammlung Physik: Nabla-Operator
โ Lern- und Lehrmaterialien
Einzelnachweise und Fuรnoten
cite-note-11. โ K. E. Georges: Ausfรผhrliches lateinisch-deutsches Handwรถrterbuch. Hrsg.: Karl-Maria Guth. 1. Auflage. Band 4 (MโQ). Hofenberg, Berlin 2014, ISBN 978-3-8430-4923-8 (Vollstรคndige Neuausgabe der 8. Auflage von 1913).
cite-note-44. โ Zeilen- und Spaltenvektoren werden in der Differentialgeometrie und im mathematischen Formalismus der Relativitรคtstheorie auch als kovariant beziehungsweise kontravariant bezeichnet. Der Ableitungsoperator nach den kovarianten Koordinaten bildet dabei einen kontravarianten Vektor und umgekehrt.
cite-note-55. โ Jรผrgen Schnakenberg: Elektrodynamik. John Wiley & Sons, 2003, ISBN 3-527-40369-8, S. 31 ff., (eingeschrรคnkte Vorschau in der Google-Buchsuche).
cite-note-beutelspacher-la-7-30-66. โ H.-D. Ebbinghaus, H. Hermes, F. Hirzebruch, M. Koecher, K. Mainzer, A. Prestel, R. Remmert: Zahlen. Band 1 Grundwissen und Mathematik. Springer-Verlag, Berlin u. a. 1983, ISBN 978-3-540-12666-9, doi:10.1007/978-3-642-96783-2 (eingeschrรคnkte Vorschau in der Google-Buchsuche).
cite-note-77. โ P. Haupt: Continuum Mechanics and Theory of Materials. 2. Auflage. Springer, 2002, ISBN 978-3-540-43111-4 (englisch).
cite-note-hbphys-88. โ C. Truesdell: Festkรถrpermechanik II. In: S. Flรผgge (Hrsg.): Handbuch der Physik. Band VIa/2. Springer, 1972, ISBN 3-540-05535-5.
Literatur
โข Bronstein, Semendjajew, Musiol, Mรผhlig: Taschenbuch der Mathematik. 5. Auflage. Harri Deutsch, 2001, ISBN 3-8171-2005-2 (Enthรคlt alle hier genannten Eigenschaften, jedoch ohne Beweis.).
โข Jรคnich: Vektoranalysis. Springer, 1992, ISBN 3-540-55530-7 (Enthรคlt nur die grundlegende Definition.).
โข Groรmann: Mathematischer Einfรผhrungskurs fรผr die Physik. Teubner, Stuttgart 1991 (siehe insbesondere Abschnitt 3.6).
โข H. Altenbach: Kontinuumsmechanik. Springer, 2012, ISBN 978-3-642-24118-5 (siehe Abschnitt 2.3 Tensoranalysis).