0
$\begingroup$

Let $e=(a,b,c)$ be a unit vector in $\mathbb{R}^3$ and let $T$ be the linear transformation on $\mathbb{R}^3$ of rotation by $180^\circ$ about $e$. Find the matrix for $T$ with respect to the standard basis $e_1=(1,0,0)$, $e_2=(0,1,0)$ and $e_3=(0,0,1)$.

The rotation matrix in $\mathbb{R}^3$ by $180^\circ$ is : \begin{bmatrix} -1 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & 1 \end{bmatrix} So rotating $e$ by $180^\circ$ gives : \begin{bmatrix} -a \\ -b \\ c \end{bmatrix} After that how to get the transformation matrix w.r.t the standard basis? P.S.- The answer given involves terms consisting of $a,b,c$.

| cite | improve this question | |
$\endgroup$
  • $\begingroup$ Why are you rotating $e$? That’s the axis of the rotation, so it would remain fixed. $\endgroup$ – amd May 4 '17 at 19:42
1
$\begingroup$

The matrix that you gave is for a 180° rotation about $e_3$. A general way to proceed from there would be via a change-of-basis operation, which involves finding an appropriate basis and performing a couple of matrix multiplications and perhaps an inversion, to boot.

Observe, however, that a 180° rotation about a vector $e$ is equivalent to a reflection in the span of $e$, so you can save yourself quite a bit of work by using the formula for reflection of a vector $v$ in a subspace $W$. This is $Rv=2\mathbf\pi_Wv-v$, where $\mathbf\pi_Wv$ is the projection of $v$ onto $W$. In this problem $W$ is the span of $e$, so for this transformation we have $$Tv=2{ee^T\over e^Te}v-v$$ and we know that $\|e\|=\sqrt{e^Te}=1$, so the matrix that you’re looking for is $$2ee^T-I_3.$$ You can easily verify for yourself that $Te=e$ and that if $v$ is orthogonal to $e$, then $Tv=-v$, which is exactly what we want for this rotation.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Ok. I found that the transformation you mentioned about that reflection $Tv=2\frac{ee^T}{e^Te}v-v$, it's a negative of the Householder transformation. But why that sign is taken negative here? $\endgroup$ – am_11235... May 5 '17 at 5:31
  • $\begingroup$ @am_11235... You ought to be able to work that out for yourself. Reflection of a vector in a space reverses the orthogonal rejection of the vector from that space, so the reflection of $v$ in $W$ is $\mathbf\pi_Wv-(v-\mathbf\pi_Wv)$. $\endgroup$ – amd May 5 '17 at 6:00
0
$\begingroup$

HINT: The matrix you give, let's call it $M$, is rotation by $180^{\circ}$ about the vector $e_3$. If you can find an orthogonal matrix $T$ that maps $e$ to $e_3$, then $T^{-1}MT$ is the rotation by $180^{\circ}$ about the vector $e$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ $T$ need not be orthogonal: it’s enough for the images of $e_1$ and $e_2$ to span $e^\perp$. On the other hand, if $T$ is orthogonal, then $T^{-1}=T^T$ and the inversion can be avoided. $\endgroup$ – amd May 4 '17 at 20:06
  • $\begingroup$ @amd You're absolutely right. I included orthogonality because it's always possible to take $T$ orthogonal, and it simplifies the computations significantly. $\endgroup$ – Servaes May 6 '17 at 10:01

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.