0
$\begingroup$

Suppose there are 2 orthogonal vectors $v_1$ and $v_2$ so $v_1^Tv_2=0$. Suppose we take linear transformations of the two vectors. Let $A$ be the transformation matrix. Let $u_1=Av_1; u_2=Av_2;$ Are $u_1$ are $u_2$ orthogonal? i.e. is $u_1^Tu_2=0$? How to prove it. I tried the following way, but I'm stuck.

$u_1^Tu_2=0$
$(Av_1)^T(Av_2)=0$
$v_1^TA^TAv_2=0$

I'm stuck here, unable to proceed further.

If the above equation is not true for any random vectors $v_1$ and $v_2$, is it true for any specific vectors? What if $v_1$ and $v_2$ are in $Row$ $space$ $of$ $A$?

$\endgroup$
1
$\begingroup$

It is not true in general, but it will be true if $A^TA=\lambda\operatorname{Id}$, for some scalar $\lambda$. In particular, it is true if $A$ is an orthogonal matrix ($\iff A^TA=\operatorname{Id}$).

$\endgroup$
  • $\begingroup$ What if $v_1$ and $v_2$ are in row space of A? Is it true then? $\endgroup$ – Nagabhushan S N May 19 '18 at 8:17
  • $\begingroup$ @NagabhushanSN No, not even then. $\endgroup$ – José Carlos Santos May 19 '18 at 8:23
  • $\begingroup$ thank you. I was watching Video Lecture by Gilbert Strang on Singular Value Decomposition. In that he takes two orthonormal vectors $v_1$, $v_2$ in row-space of A. Let $\sigma_1$*$u_1$=A*$v_1$ and $\sigma_2$*$u_2$=A*$v_2$. He takes $u_1$ is orthogonal to $u_2$. How? $\endgroup$ – Nagabhushan S N May 19 '18 at 8:26
  • $\begingroup$ @NagabhushanSN Post this as another question, with a link to the video. $\endgroup$ – José Carlos Santos May 19 '18 at 8:29
  • $\begingroup$ here it is math.stackexchange.com/q/2777511/516816 $\endgroup$ – Nagabhushan S N May 19 '18 at 8:36
1
$\begingroup$

Not it is not true in general, just consider $e_1=(1,0)$, $e_2=(0,1)$ and

$$A=\begin{bmatrix}1&0\\1&1\end{bmatrix}$$

It is true when $A^TA=k\,I$, indeed

$$\langle Ae_i,Ae_j \rangle=e_i^TA^TAe_j=ke_i^Te_j=\langle e_i,e_j \rangle=0$$

$\endgroup$

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.