Covariant basis

Description

I consider

\[\pder{\vec{E}_i}{X^j}.\]

By using the relation

\[\fromctog,\]

this is equal to

\[\pder{}{X^j} \left( \sum_l \pder{x^l}{X^i} \vec{e}_l \right) = \sum_l \pder{}{X^j} \left( \pder{x^l}{X^i} \right) \vec{e}_l + \sum_l \pder{x^l}{X^i} \pder{\vec{e}_l}{X^j}.\]

Since the reference Cartesian basis vectors are constant, the second term disappears and I am left with

\[\begin{split}\sum_l \pder{}{X^j} \left( \pder{x^l}{X^i} \right) \vec{e}_l & = \sum_l \pder{}{X^j} \left( \pder{x^l}{X^i} \right) \left( \sum_k \pder{X^k}{x^l} \vec{E}_k \right) \\ & = \sum_{kl} \pder{}{X^j} \left( \pder{x^l}{X^i} \right) \pder{X^k}{x^l} \vec{E}_k \\ & = \sum_k \left\{ \sum_l \pder{}{X^j} \left( \pder{x^l}{X^i} \right) \pder{X^k}{x^l} \right\} \vec{E}_k \\ & = \sum_k \Gamma_{ij}^k \vec{E}_k,\end{split}\]

where \(\Gamma_{ij}^k\) is the Christoffel symbol of the second kind:

\[\christoffel,\]

and by using the relation between the symbol and the scale factors:

\[\fromchristoffeltoscalefactor,\]

the change of the basis vector is given by

\[\begin{split}\newcommand{\tmp}[3]{ \pder{}{X^{#3}} \left( \delta_{#1#2} H_{#1} H_{#2} \right) } & \sum_k \frac{1}{2 H_k H_k} \left\{ - \tmp{i}{j}{k} + \tmp{k}{i}{j} + \tmp{j}{k}{i} \right\} \vec{E}_k \\ = & - \sum_k \delta_{ij} \frac{H_j}{2 H_k H_k} \pder{H_i}{X^k} \vec{E}_k - \sum_k \delta_{ij} \frac{H_i}{2 H_k H_k} \pder{H_j}{X^k} \vec{E}_k + \frac{1}{H_i} \pder{H_i}{X^j} \vec{E}_i + \frac{1}{H_j} \pder{H_j}{X^i} \vec{E}_j,\end{split}\]

and in summary:

\[\dedx{i}{j}{k}.\]

This relation is written down explicitly below for an intuitive understanding.

\[\begin{split}\newcommand{\diag}[1]{ \sum_k \frac{H_#1}{H_k H_k} \pder{H_#1}{X^k} \vec{E}_k } \newcommand{\nodiag}[2]{ \frac{1}{H_{#1}} \pder{H_{#1}}{X^{#2}} \vec{E}_{#1} + \frac{1}{H_{#2}} \pder{H_{#2}}{X^{#1}} \vec{E}_{#2} } \begin{pmatrix} \pder{\vec{E}_1}{X^1} & \pder{\vec{E}_2}{X^1} & \pder{\vec{E}_3}{X^1} \\ \pder{\vec{E}_1}{X^2} & \pder{\vec{E}_2}{X^2} & \pder{\vec{E}_3}{X^2} \\ \pder{\vec{E}_1}{X^3} & \pder{\vec{E}_2}{X^3} & \pder{\vec{E}_3}{X^3} \\ \end{pmatrix} = & - \begin{pmatrix} \diag{1} & 0 & 0 \\ 0 & \diag{2} & 0 \\ 0 & 0 & \diag{3} \\ \end{pmatrix} \\ & + \begin{pmatrix} 0 & \nodiag{1}{2} & \nodiag{1}{3} \\ sym. & 0 & \nodiag{2}{3} \\ sym. & sym. & 0 \\ \end{pmatrix}.\end{split}\]

Example

The changes of the basis vectors

\[\begin{split}\begin{pmatrix} \pder{\vec{E}_1}{X^1} & \pder{\vec{E}_2}{X^1} & \pder{\vec{E}_3}{X^1} \\ \pder{\vec{E}_1}{X^2} & \pder{\vec{E}_2}{X^2} & \pder{\vec{E}_3}{X^2} \\ \pder{\vec{E}_1}{X^3} & \pder{\vec{E}_2}{X^3} & \pder{\vec{E}_3}{X^3} \\ \end{pmatrix}\end{split}\]

read as follows.

Cylindrical coordinates

\[\begin{split}\begin{pmatrix} 0 & \frac{1}{X^1} \vec{E}_2 & 0 \\ \frac{1}{X^1} \vec{E}_2 & - X^1 \vec{E}_1 & 0 \\ 0 & 0 & 0 \\ \end{pmatrix}.\end{split}\]

Rectilinear coordinates

\[\begin{split}\begin{pmatrix} - \frac{1}{H_1} \pder{H_1}{X^1} \vec{E}_1 & 0 & 0 \\ 0 & - \frac{1}{H_2} \pder{H_2}{X^2} \vec{E}_2 & 0 \\ 0 & 0 & - \frac{1}{H_3} \pder{H_3}{X^3} \vec{E}_3 \\ \end{pmatrix}.\end{split}\]

Application

\[\begin{split}\begin{pmatrix} - \frac{1}{H_1} \pder{H_1}{X^1} \vec{E}_1 & \frac{1}{H_2} \pder{H_2}{X^1} \vec{E}_2 & 0 \\ \frac{1}{H_2} \pder{H_2}{X^1} \vec{E}_2 & - \frac{H_2}{H_1 H_1} \pder{H_2}{X^1} \vec{E}_1 & 0 \\ 0 & 0 & 0 \\ \end{pmatrix}.\end{split}\]