1 Vector and Tensor Components
Henceforth, the Einstein summation convention will be employed.
1.1 Vector Components
A vector can be written as a linear combination of a basis, that is a list of linearly independent vectors the span the space. We are typically interested in vectors that belong to \(\b{E}^{n}\).
Definition 1 Orthogonal vectors
Two vectors \(\bs{u}, \bs{v} \in \b{E}^n\) are orthogonal if \(\bs{u} \cdot \bs{v} = 0\).
Definition 2 Orthonormal basis
An orthonormal basis is a basis whose elements are mutually orthogonal and have a unit norm.
We typically denote orthonormal basis vectors with a tophat, for example \(\{\hat{\bs{E}}_{i}\}_{i=1,2,3}\). Further, we will typically make use of orthonormal basis vectors unless specified otherwise.
Let \(\{\hat{\bs{E}}_{i}\}_{i=1,2,3}\) form an orthonormal basis for \(\b{E}^{3}\). That is any vector \(\bs{u} \in \b{E}^{3}\) can be written as a linear combination of these basis vectors.
\[ \begin{align} \bs{x}&=x_1 \hat{\bs{E}}_1+ x_2 \hat{\bs{E}}_2+ x_3 \hat{\bs{E}}_3\\ \bs{x}&=x_i \hat{\bs{E}}_i \end{align} \]
\[ \begin{align} [\bs{x}]_{\{\hat{\bs{E}}_{i}\}_{i=1,2,3}}&= \left[ \begin{array}{c} x_1\\ x_2\\ x_3 \end{array} \right] \end{align} \]
Let \(\{\hat{\bs{F}}_{i}\}_{i=1,2,3}\) form another orthonormal basis for \(\b{E}^{3}\). Then,
\[ \begin{align} \bs{x}&=\tilde{x}_1 \hat{\bs{F}}_1+ \tilde{x}_2 \hat{\bs{F}}_2+ \tilde{x}_3 \hat{\bs{F}}_3\\ \bs{x}&=\tilde{x}_i \hat{\bs{F}}_i \end{align} \]
\[ \begin{align} [\bs{x}]_{\{\hat{\bs{F}}_{i}\}_{i=1,2,3}}&= \left[ \begin{array}{c} \tilde{x}_1\\ \tilde{x}_2\\ \tilde{x}_3 \end{array} \right] \end{align} \]
Notice that the components differ when the chosen basis differs. Additionally, for clarity, the matrix representation of the components may include a subscript denoting the basis. The following are equivalent representations of the vector:
\[ \begin{align} \bs{x} = x_{i} \hat{\bs{E}}_{i} = \tilde{x}_{i} \hat{\bs{F}}_{i} \end{align} \]
The question is, are the components \(x_{i}\) and \(\tilde{x}_i\) the same? In general no. However, a relation between the two represenation can be determined using a change of basis, which will be discussed in a later section.
1.2 Tensor Components
Say \(\bs{T} : \b{E}^{n} \ra \b{E}^n\) is a tensor. Then we define its components \(T_{ij}\) w.r.t. to the two bases \(\hat{\bs{E}}_i\) and \(\hat{\bs{F}}_i\) as
\[ \begin{align} T_{ij}=\hat{\bs{F}}_i\cdot(\bs{T} \hat{\bs{E}}_j) \end{align} \]
Typically, we chose both bases to be the same to simplify our study. That is, the components of \(\bs{T}\) w.r.t the basis \(\hat{\bs{E}}_i\) is given by \[ \begin{align} T_{ij}=\hat{\bs{E}}_i\cdot(\bs{T} \hat{\bs{E}}_j) \end{align} \]
Let \(\bs{T} : \b{E}^{3} \ra \b{E}^3\) be a tensor. We define the matrix representation of \(\bs{T}\) as the \(3\times 3\) matrix that we denote as \(\left[\bs{T}\right]\) that satisfies the following equation \[\begin{align} \left[ \bs{T}\right] \left[ \bs{a}\right]&=\left[ \bs{T}\bs{a}\right] \end{align} \]
for all \(\bs{a}\in \mathbb{E}^3\). Let \(\bs{T}\bs{a}=\bs{b}\), where \(\bs b = b_i \hat{\bs{E}}_i\). The matrix representation of \(\bs{b}\) is,
\[ \begin{equation} \left[\bs{b}\right]= \left[ \begin{array}{c} b_1\\ b_2\\ b_3\\ \end{array} \right] \end{equation} \]
Then we have that, \[ \begin{align} \left[ \bs{T}\right] \left[ \bs{a}\right]&=\left[ \bs{b}\right]\\ \end{align} \]
Let \(T_{ij}\) be the \(i-j^{\text{th}}\) component of the matrix \(\left[\bs{T}\right]\), i.e., \(T_{ij}=\left[\bs{T}\right]_{ij}\). More explicitly, \[ \begin{align} \left[\bs{T}\right]&= \left[ \begin{array}{ccc} T_{11} && T_{12} && T_{13} \\ T_{21} && T_{22} && T_{23} \\ T_{31} && T_{32} && T_{33} \\ \end{array} \right] \end{align} \]
\[ \begin{align} \left[ \begin{array}{ccc} T_{11} && T_{12} && T_{13} \\ T_{21} && T_{22} && T_{23} \\ T_{31} && T_{32} && T_{33} \\ \end{array} \right] \left[ \begin{array}{c} a_1\\ a_2\\ a_3\\ \end{array} \right] &= \left[ \begin{array}{c} b_1\\ b_2\\ b_3\\ \end{array} \right] \end{align} \]
or equivalently, \[ \begin{align} T_{11}a_1 + T_{12}a_2 + T_{13}a_3 = b_1\\ T_{21}a_1 + T_{22}a_2 + T_{23}a_3 = b_2\\ T_{31}a_1 + T_{32}a_2 + T_{33}a_3 = b_3\\ \end{align} \]
1.2.1 Computing the matrix representation of a tensor
Motivating example
Let \(\bs{T}\) be a tensor that transforms the specific vectors \(\bs{a}\) and \(\bs{b}\) as follows:
\[ \begin{align} \bs{T}\bs{a} = \bs{a} + 2\bs{b}, \\ \bs{T}\bs{b} = \bs{a} - \bs{b}. \end{align} \]
Given a vector \(\bs{c} = 2\bs{a} + \bs{b}\), find \(\bs{T}\bs{c}\).
Solution
Using the linearity property of tensors, we have \[ \begin{align} \bs{T}\bs{c} = \bs{T}(2\bs{a} + \bs{b})=2\bs{T}\bs{a} + \bs{T}\bs{b} = 2(\bs{a} + 2\bs{b}) + (\bs{a} - \bs{b}) = 3\bs{a} + 3\bs{b}. \end{align} \]
So let us start by defining the Matrix \[ \begin{align} \left[Q\right] &= \left[ \begin{array}{ccc} Q_{11} && Q_{12} && Q_{13} \\ Q_{21} && Q_{22} && Q_{23} \\ Q_{31} && Q_{32} && Q_{33} \end{array} \right] \end{align} \]
\[ \begin{align} \bs{T}\hat{\bs{E}}_1&=Q_{11}\hat{\bs{E}}_1+Q_{21}\hat{\bs{E}}_2+Q_{31}\hat{\bs{E}}_3\\ \bs{T}\hat{\bs{E}}_2&=Q_{12}\hat{\bs{E}}_1+Q_{22}\hat{\bs{E}}_2+Q_{32}\hat{\bs{E}}_3\\ \bs{T}\hat{\bs{E}}_3&=Q_{13}\hat{\bs{E}}_1+Q_{23}\hat{\bs{E}}_2+Q_{33}\hat{\bs{E}}_3\\ \end{align} \]
\[ \begin{align} \bs{T}\bs{a}&=\bs{T}\left(a_1\hat{\bs{E}}_1+a_2\hat{\bs{E}}_2+a_3\hat{\bs{E}}_3\right)\\ &=\bs{T}\left(a_1\hat{\bs{E}}_1\right)+\bs{T}\left(a_2\hat{\bs{E}}_2\right)+\bs{T}\left(a_3\hat{\bs{E}}_3\right)\\ &=a_1\left(\bs{T}\hat{\bs{E}}_1\right)+a_2 \left(\bs{T}\hat{\bs{E}}_2\right)+a_3\left(\bs{T}\hat{\bs{E}}_3\right)\\ \end{align} \]
\[ \begin{align} \bs{T}\bs{a} &=a_1\left(\bs{T}\hat{\bs{E}}_1\right)+a_2 \left(\bs{T}\hat{\bs{E}}_2\right)+a_3\left(\bs{T}\hat{\bs{E}}_3\right)\\ &=a_1\left(Q_{11}\hat{\bs{E}}_1+Q_{21}\hat{\bs{E}}_2+Q_{31}\hat{\bs{E}}_3\right)+a_2 \left(Q_{12}\hat{\bs{E}}_1+Q_{22}\hat{\bs{E}}_2+Q_{32}\hat{\bs{E}}_3\right)+a_3\left(Q_{13}\hat{\bs{E}}_1+Q_{23}\hat{\bs{E}}_2+Q_{33}\hat{\bs{E}}_3\right) \end{align} \]
\[ \begin{align} \bs{T}\bs{a} &=\left(Q_{11}a_1\hat{\bs{E}}_1+Q_{21}a_1\hat{\bs{E}}_2+Q_{31}a_1\hat{\bs{E}}_3\right)+ \left(Q_{12}a_2\hat{\bs{E}}_1+Q_{22}a_2\hat{\bs{E}}_2+Q_{32}a_2\hat{\bs{E}}_3\right)+\left(Q_{13}a_3\hat{\bs{E}}_1+Q_{23}a_3\hat{\bs{E}}_2+Q_{33}a_3\hat{\bs{E}}_3\right) \\ &= \left( Q_{11}a_1+ Q_{12}a_2+ Q_{13}a_3\right)\hat{\bs{E}}_1 + \left( Q_{21}a_1+ Q_{22}a_2+ Q_{23}a_3\right)\hat{\bs{E}}_2 + \left( Q_{31}a_1+ Q_{32}a_2+ Q_{33}a_3 \right)\hat{\bs{E}}_3 \end{align} \]
\[\begin{align} \left[\bs{T}\bs{a}\right] &= \left[ \begin{array}{c} Q_{11}a_1+Q_{12}a_2+Q_{13}a_3\\ Q_{21}a_1+Q_{22}a_2+Q_{23}a_3\\ Q_{31}a_1+Q_{32}a_2+Q_{33}a_3 \end{array} \right] \\ \left[\bs{b}\right]&= \left[ \begin{array}{ccc} Q_{11} && Q_{12} && Q_{13} \\ Q_{21} && Q_{22} && Q_{23} \\ Q_{31} && Q_{32} && Q_{33} \end{array} \right] \left[ \begin{array}{c} a_1\\ a_2\\ a_3 \end{array} \right] \\ \left[ \begin{array}{c} b_1\\ b_2\\ b_3\\ \end{array} \right]&= \left[ \begin{array}{ccc} Q_{11} && Q_{12} && Q_{13} \\ Q_{21} && Q_{22} && Q_{23} \\ Q_{31} && Q_{32} && Q_{33} \end{array} \right] \left[ \begin{array}{c} a_1\\ a_2\\ a_3 \end{array} \right] \end{align} \]
which proves That \[ \begin{equation} \left[ \bs{T} \right]= \left[Q\right] \end{equation} \]
Example
Obtain the matrix for the tensor \(\bs{T}\) that transforms the base vectors as follows
\[ \begin{align} \bs{T}\bs{e}_1 &= 4\bs{e}_1 + \bs{e}_2, \\ \bs{T}\bs{e}_2 &= 2\bs{e}_1 + 3\bs{e}_3, \\ \bs{T}\bs{e}_3 &= -\bs{e}_1 + 3\bs{e}_2 + \bs{e}_3, \end{align} \]
Solution
By the definition above,
\[ \begin{align} \left[\bs{T}\right] &= \left[ \begin{array}{ccc} 4 && 2 && -1 \\ 1 && 0 && 3 \\ 0 && 3 && 1 \end{array} \right] \end{align} \]
Example
Obtain the matrix for the tensor that transforms the base vectors as follows:
\[ \begin{align} \bs{T}\bs{e}_1 &= \bs{e}_1 + 2\bs{e}_2 + 3\bs{e}_3, \\ \bs{T}\bs{e}_2 &= 4\bs{e}_1 + 5\bs{e}_2 + 6\bs{e}_3, \\ \bs{T}\bs{e}_3 &= 7\bs{e}_1 + 8\bs{e}_2 + 9\bs{e}_3. \end{align} \]
Solution
By inspection,
\[ \begin{align} \left[\bs{T}\right] &= \left[ \begin{array}{ccc} 1 && 2 && 3 \\ 4 && 5 && 6 \\ 7 && 8 && 9 \end{array} \right] \end{align} \]
This example emphasizes again the convention we use to write the matrix of a tensor: The component of \(\bs{T}\bs{e}_{1}\) fills the first column, the component of \(\bs{T}\bs{e}_{2}\) fills the second column, and so on. The reason for this choice of convention will become obvious in the next section.
Example: Reflection across a plane
Let \(\bs{T}\) transform every vector into its mirror image with respect to a fixed plane; if \(\bs{e}_1\) is normal to the reflection plane (\(\bs{e}_{2}\) and \(\bs{e}_{3}\) are parallel to this plane), find a matrix of \(\bs{T}\).
Solution
Because the normal to the reflection plane is transformed into its negative, and vectors parallel to the place are not altered, we have
\[ \begin{align} \bs{T}\bs{e}_1 = -\bs{e}_1, \bs{T}\bs{e}_2 = \bs{e}_2, \bs{T}\bs{e}_3 = \bs{e}_3 \end{align} \]
which corresponds to
\[ \begin{align} \left[\bs{T}\right] &= \left[ \begin{array}{ccc} -1 && 0 && 0 \\ 0 && 1 && 0 \\ 0 && 0 && 1 \end{array} \right]_{\bs{e}_i}. \end{align} \]
Example: 2D Rotation
Let \(\bs{R}\) correspond to a right-hand rotation of a rigid body about the \(x_3\)-axis by an angle \(\theta\). Find a matrix of \(\bs{R}\).
Solution
From the figure, it is clear that
\[ \begin{align} \bs{R}\bs{e}_1 &= \cos\theta\bs{e}_1 + \sin\theta\bs{e}_2, \\ \bs{R}\bs{e}_2 &= -\sin\theta\bs{e}_1 + \cos\theta\bs{e}_2, \\ \bs{R}\bs{e}_3 &= \bs{e}_3. \end{align} \]
which corresponds to
\[ \begin{align} \left[\bs{R}\right] &= \left[ \begin{array}{ccc} \cos\theta && \sin\theta && 0 \\ \sin\theta && \cos\theta && 0 \\ 0 && 0 && 1 \end{array} \right]_{\bs{e}_i}. \end{align} \]
Example: Stretching
Write the tensor \(\bs{T}\), defined by the equation \(\bs{T}\bs{a} = \alpha\bs{a}\), where \(\alpha\) is a constant and \(\bs{a}\) is arbitrary, in terms of the identity tensor, and find its components.
Solution
Using the identity tensor, we can write \(\alpha\bs{a}\) as \(\alpha\bs{I}\bs{a}\), so that
\[ \begin{align} \bs{T}\bs{a} = \alpha\bs{a} = \alpha\bs{I}\bs{a}. \end{align} \]
Since \(\bs{a}\) is arbitrary, therefore,
\[ \begin{align} \bs{T} = \alpha\bs{I} \end{align} \]
The components of this tensor are clearly \(T_{ij} = \alpha\delta_{ij}\).
1.2.2 Representation of tensors as a sum of dyadics products
If \(\hat{\bs{E}}_i\) is an orthonormal basis, it can be shown that
\[ \begin{align} \bs{T}&=T_{ij} \, \hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j} \tag{Dyadic basis} \end{align} \]
Let \[ \begin{align} \bs{T}\bs{c}&=\bs{d}\\ (T_{ij} \, \hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j})\bs{c}&=\bs{e} \end{align} \]
We are going to find the components of \(\bs{e}\) and \(\bs{d}\) and show that they are one and the same. Since \(\bs{c}\) is arbitrary this goes to prove that \(\bs{T}\) and \(T_{ij}\hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j}\) are one and the same.
Components of \(\bs{e}\) \[ \begin{align} \bs{e}&=(T_{ij}\hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j})\bs{c}\\ &= (T_{ij}\hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j})(c_k\hat{\bs{E}}_k)\\ &= T_{ij} c_k(\hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j})\hat{\bs{E}}_k\\ &= T_{ij} c_k(\hat{\bs{E}}_{i}( \hat{\bs{E}}_{j}\cdot\hat{\bs{E}}_k))\\ &= T_{ij} c_k(\hat{\bs{E}}_{i}\delta_{jk})\\ &= T_{ij} c_j\hat{\bs{E}}_{i} \end{align} \]
From the last equation it follows that
\[ \begin{align} \bs{e}\cdot\hat{\bs{E}}_j&=T_{ik}c_k\hat{\bs{E}}_i\cdot\hat{\bs{E}}_j\\ e_j&=T_{ik}c_k\delta_{ij}\\ e_j&=T_{jk}c_k\label{eq:eComponents}\\ \end{align} \]
Components of \(\bs{d}\)
From the definition of the matrix definition of \(\bs{T}\), which is \([\bs{T}]\) it follows that
\[ \begin{align} \left[\bs{T}\right]\left[\bs{c}\right]&=\left[\bs{d}\right] \label{eq:MatrixFrom} \end{align} \]
where \[ \begin{align} \left[\bs{c}\right]_{i}&=c_{i}=\bs{c}\cdot \hat{\bs{E}}_i\\ \left[\bs{d}\right]_{j}&=d_{j}=\bs{d}\cdot \hat{\bs{E}}_j \end{align} \]
In Einstein notation \(\eqref{eq:MatrixFrom}\) reads,
\[ \begin{align} d_i&=T_{ij}c_j\label{eq:dComponents} \end{align} \]
If follows from \(\eqref{eq:eComponents}\) and \(\eqref{eq:dComponents}\) that \(\bs{e}\) and \(\bs{d}\) are the same, from which it follows that \(\bs{T}\) and \(T_{ij}\hat{\bs{E}}_{i}\otimes \hat{\bs{E}}_{j}\) are one and the same.