Hi am studying orthogonal matrices and I am having difficulties in finding a proof and to show the following :

An n x n matrix is orthogonal if $A^T A = I $, Show that such matrices preserve volumes.

I found that it is related with the determinant. It says that the determinant of an orthogonal matrix is $\pm$1 and orthogonal transformations and isometries preserve volumes. However I do not know how to show it.

Also I would like to show that Orthogonal matrices preserve dot product and I found that:

$A\vec{x}$ $.$ $A\vec{y}$= $\vec{x}$.$\vec{y}$ then,

$A\vec{x}$ $.$ $A\vec{y}$= $A^T $$A\vec{x}$.$\vec{y}$ and because of orthogonality property, $A^T A = I $

this is $\vec{x}$.$\vec{y}$

In this case, I am not sure if this is correct or complete.

Can anyone help me on this?


  • $\begingroup$ You can just take "preserves volume" to literally mean $\det(A) = \pm 1$ which is easily provable from the definition ($A^TA=I$). $\endgroup$ – user137731 Feb 26 '17 at 2:51
  • 4
    $\begingroup$ As for $A$ preserving the dot product: $$Ax \cdot Ay := (Ax)^T(Ay) = (x^TA^T)Ay = x^T(A^TA)y = x^TIy = x^Ty =: x\cdot y$$ $\endgroup$ – user137731 Feb 26 '17 at 2:54
  • $\begingroup$ In the last part do you mean: $\vec{x}$.$\vec{y}$? I do not understand the literally mean of the determinant. I found something that says: that is related to the scalar triple product (a x b) . c- volume of a parallelepiped spanned by the vectors a,b and c. and by using the alternative notation for the derminant the volume of the parallelepiped spanned by a,b,c is |(axb).c | but I do not understand how it shows that preserves volume $\endgroup$ – user290335 Feb 26 '17 at 3:04
  • $\begingroup$ Not "literally mean of the determinant" as if "literally mean" is an operation. I'm just using English vernacular, but I'll rephrase: you should consider "preserving volumes" as the same thing as "$\det(A) = \pm 1$". $\endgroup$ – user137731 Feb 26 '17 at 3:07
  • $\begingroup$ Ah, so showing that preserve volume is the same as proving that determinant is 1 or -1? $\endgroup$ – user290335 Feb 26 '17 at 3:10

Like said in the comments, for orthogonal matrices $A$ have $|\det(A)|=1$ (as a consequence of the multiplicativity of $\det$). The volume preserving property is a special case of a substitution theorem for integrals with multiple variables from analysis (https://en.wikipedia.org/wiki/Integration_by_substitution#Substitution_for_multiple_variables): When you transform an object $U$ with a linear map $\phi:\mathbb{R}^n\rightarrow\mathbb{R}^n,x\mapsto Ax$ you have $\phi'(x)=A$ and we can compute the volume of the transformed object $\phi(U)$ as follows: $$\operatorname{Vol}(\phi(U)) = \int_{\phi(U)}dx = \int_U \underbrace{|\det \phi'(x)|}_1 dx = \int_U dx = \operatorname{Vol}(U).$$ Thats why orthogonal matrices are volume preserving. (Btw orthogonal matrices only allow rotations (case $\det A=1$) or reflections (case $\det A=-1$), which makes the volume preserving property intuitively plausible)

  • $\begingroup$ Thank you very much for your help $\endgroup$ – user290335 Mar 1 '17 at 14:57

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.