I have this parametrization of the sphere that seems quite a mess \begin{equation} R_{ij}=\cos r\left(\delta_{ij}-\hat{\theta}_{i}\hat{\theta}_{j}\right)+\hat{\theta}_{i}\hat{\theta}_{j}+\sin r\sum\epsilon_{ijk}\hat{\theta}_{k}, \end{equation}

$i,j=1,2,3$ and $\hat{\theta}_{i}=\theta_{i}/r$ and $r=\sqrt{\left(\theta_{1}\right)^{2}+\left(\theta_{2}\right)^{2}+\left(\theta_{3}\right)^{2}}$ I have to show that $$\sum R_{ik}R_{jk} =\delta_{ij}.$$ Does anybody have seen this parametrization and can give me some references? Thanks.


Explanation: Instead of speaking of sphere parameterization, I would say that it is a variant of the Rodrigues formula for rotations, written with tensor notations.

Let us make the connection. If you apply your formula:

$$\begin{equation} R_{ij}=\cos r\left(\delta_{ij}-\hat{\theta}_{i}\hat{\theta}_{j}\right)+\hat{\theta}_{i}\hat{\theta}_{j}+\sin r\sum\epsilon_{ijk}\hat{\theta}_{k}, \end{equation}$$

to a vector $V_i$, you get:

$$\begin{equation} R_{ij}V_i=\cos r\left(\delta_{ij}-\hat{\theta}_{i}\hat{\theta}_{j}\right)V_i+\hat{\theta}_{i}\hat{\theta}_{j}V_i+\sin r\sum\epsilon_{ijk}\hat{\theta}_{k}V_i, \ \ \ (1) \end{equation}$$

Have now a look at (https://en.wikipedia.org/wiki/Rodrigues'_rotation_formula). Do not stop at its main formula, but proceed a little further in the demontration where you will find:

$$V_{rot}=\cos\theta(V-V_{\parallel})+V_{\parallel}+\sin\theta \ k \times V \ \ \ (2) \ \ \text{(a version of Rodrigues' formula)}$$

(where $k$ is the unit vector giving the axis of rotation, and $V_{\parallel}=(k.V)k=k(k.V)=k(k^TV)$ is the projection of $V$ onto the axis directed by unit vector $k$).

With your notations, where $\hat{\theta}$ is also a unit vector, (2) can be written under the form:

$$V_{rot}=\cos r(IV-(\hat{\theta}\hat{\theta}^TV))+\hat{\theta}\hat{\theta}^TV+\sin r \ \hat{\theta} \times V$$ The identification between (1) and (2) is almost finished. It remains to understand the last term of (1) which is decrypted as a matrix applied to a vector, this matrix being the matrix associated with the cross product with $\hat{\theta}$:

$$\Lambda=\Lambda_{\hat{\theta}}=\begin{bmatrix} 0 & -\hat{\theta}_3&\hat{\theta}_2 \\ \hat{\theta}_3 & 0 & -\hat{\theta}_1 \\ -\hat{\theta}_2 & \hat{\theta}_1 & 0 \end{bmatrix} \ \ \text{which is such that } \ \ \Lambda V = \hat{\theta} \times V$$

One understands now that tensor $\epsilon_{ijk}$ is there for providing its $+1,-1$ or $0$ at the right places in matrix $\Lambda$ (see below the definition of this tensor).

What is asked to you is to prove that a rotation fulfills


which is of course true (but also true as well for symmetries with respect to planes).

Of course, you can prove this formula using tensor notations; in my opinion, vector-matrix notations (i.e., non tensorial) are more easy to work with, at least for this issue.

Explanation about the Levi-Civita tensor: $\epsilon_{ijk}$ may take 3 values:

  • values $+1$ or $-1$ according to the parity of permutation $(i,j,k)$ in the case all $i,j,k$ are different,

  • value $0$ if two or more indices are identical.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.