I am haunted by a problem of angles of rotation. Here's my nightmare. At the origin of an inertial frame $R_0(XYZ)$, there is a ball. Another reference frame $R(xyz)$ is fixed to the center of this ball. At the beginning, the origin of $R_0$ coincides the origin of $R$, and $x\parallel X$, $y\parallel Y$, $z\parallel Z$ and in the same orientation. Then the ball begins to roll. Firstly, it rotates by an angle of $\psi$ around the axe $Z$ in $R_0$; secondly, it rotates by an angle of $\theta$ around the 'new' axe $x$ in $R$; finally, it rotates by an angle of $\phi$ around the 'final' axe $y$ in $R$. My question is, after these three rotations, could we express the angles of rotation ($\theta_X$,$\theta_Y$,$\theta_Z$)of the ball with respect to the 3 axes $X$,$Y$ and $Z$ of inertial frame $R_0$? I tried different ways, but it seems to be too complexe to express them without using integration. Is there any existing mathematical formulation to deal this problem? Thank you for taking a look!



You can solve this problem representing the rotations by means of quaternions. (see here and, more briefly: Representing rotations using quaternions ).

Let $\mathbf{i},\mathbf{j},\mathbf{k}$ the three versors of the axis $X,Y,Z$ in $R_0$. The first rotation of the $x,y,z$ axis around $Z$ of angle $\psi$, is given, in exponential notation, by: $$ \mathbf{x_1}=e^{\mathbf{k}\psi/2}\mathbf{i}e^{\mathbf{k}\psi/2}=\mathbf{i}\cos \psi+\mathbf{j} \sin \psi $$ $$ \mathbf{y_1}=e^{\mathbf{k}\psi/2}\mathbf{j}e^{\mathbf{k}\psi/2}=-\mathbf{i}\sin \psi+\mathbf{j} \cos \psi $$ $$ \mathbf{z_1}=\mathbf{z}=\mathbf{k} $$ where $\mathbf{x_1},\mathbf{y_1},\mathbf{z_1}$ are the rotated versors of the moving frame $R$(note that this is a simple rotation in the $XY$ plane and the quaternion representation seems a unnecessary complication).

Now the second rotation is about the versor $\mathbf{x_1}$ of an angle $\theta$ so , with obvious notation, the rotated versors are given by:

$$ \mathbf{x_2}=\mathbf{x_1} $$ $$ \mathbf{y_2}=e^{\mathbf{x_1}\theta/2}\mathbf{y_1}e^{-\mathbf{x_1}\theta/2} $$ $$ \mathbf{z_2}=e^{\mathbf{x_1}\theta/2}\mathbf{k}e^{-\mathbf{x_1}\theta/2} $$

Here the calculus becomes more complex and the use of quaternions is well justified. And, in the same way we can represent the other rotation around $\mathbf{y_2}$:

$$ \mathbf{x_3}=e^{\mathbf{y_2}\phi/2}\mathbf{x_2}e^{-\mathbf{y_2}\phi/2} $$ $$ \mathbf{y_3}=\mathbf{y_2} $$ $$ \mathbf{z_3}=e^{\mathbf{y_2}\phi/2}\mathbf{z_2}e^{-\mathbf{y_2}\phi/2} $$

From these you can easily found the angles between the rotated versors $ \mathbf{x_3},\mathbf{y_3},\mathbf{z_3}$, and the axis of the fixed reference system using the scalar product.

Clearly the explicit calculation of the components of $\mathbf{x_3},\mathbf{y_3},\mathbf{z_3}$ requires a bit of algebraic work. And, if you prefer, you can find the same result also using the matrix representation of the rotations, obviously in the correct order.

  • $\begingroup$ Thank you very much! It helps a lot. $\endgroup$ – Zihan Shen Feb 19 '16 at 7:56
  • $\begingroup$ You are welcome! I've added a reference also to the matrix representation :) $\endgroup$ – Emilio Novati Feb 19 '16 at 8:26
  • $\begingroup$ Thank you very much! Could I ask another question? $\endgroup$ – Zihan Shen Feb 19 '16 at 10:15
  • $\begingroup$ If it is a clarification about this question you can add a comment. If it is a different question it is better to ask another question ant link the two questions if these are related. $\endgroup$ – Emilio Novati Feb 19 '16 at 11:27
  • $\begingroup$ Thank you for your advice. I am rethinking about it and going to ask another question. ;) $\endgroup$ – Zihan Shen Feb 19 '16 at 11:40

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.