3
$\begingroup$

Is the determinant of any submatrix of an ORTHOGONAL matrix extracted from the intersection of $k$ row and $k$ columns equal to that of the $(n-k)(n-k)$ submatrix remaining after deletion of these rows and columns (up to a sign)?! this is something I arrived at in trials to show that the Hodge Star operator’s definition does not depend on the choice of an orthonormal basis of $T*_xM$, $M$ a Reimannian manifold.

$\endgroup$
1
  • $\begingroup$ Yes, and you arrived at this exactly the way I did. I mentioned it to other differential geometry people, it was unfamiliar to all. But I also showed it to a matrix expert in the late 1980's, he showed me a one-line proof, typeset in my answer below. Note that column permutations and row permutations allow you to take a standard block pattern without loss of generality. $\endgroup$
    – Will Jagy
    Jun 23, 2013 at 17:32

2 Answers 2

8
$\begingroup$

Given a square matrix $M \in SO_n$ decomposed as illustrated with square blocks $A,D$ and rectangular blocks $B,C,$

$$M = \left( \begin{array}{cc} A & B \\\ C & D \end{array} \right) ,$$

then $\det A = \det D.$

What this says is that, in Riemannian geometry with an orientable manifold, the Hodge star operator is an isometry, a fact that has relevance for Poincare duality.

http://en.wikipedia.org/wiki/Hodge_duality

http://en.wikipedia.org/wiki/Poincar%C3%A9_duality

But the proof is a single line:

$$ \left( \begin{array}{cc} A & B \\\ 0 & I \end{array} \right) \left( \begin{array}{cc} A^t & C^t \\\ B^t & D^t \end{array} \right) = \left( \begin{array}{cc} I & 0 \\\ B^t & D^t \end{array} \right). $$

$\endgroup$
2
  • $\begingroup$ I first put this at mathoverflow.net/questions/3134/… in March 2010. So sue me. $\endgroup$
    – Will Jagy
    Jun 23, 2013 at 17:26
  • $\begingroup$ Can anything be said about the invertibility of the block C? $\endgroup$
    – gen
    Oct 16, 2021 at 1:38
2
$\begingroup$

Let $M\in{\bf GL}_n(k)$ be written blockwise $$M=\begin{pmatrix} A & B \\\\ C & D \end{pmatrix},$$ with $A$ a square, invertible, submatrix. The Sherman-Morrison formula says that $$\det M=\det A\cdot\det(D-CA^{-1}B).$$ On the other hand, it can be shown that $$M^{-1}=\begin{pmatrix} \cdot & \cdot \\\\ \cdot & (D-CA^{-1}B)^{-1} \end{pmatrix}.$$ We therefore have $$\det M\cdot\det(D-CA^{-1}B)^{-1}=\det A.$$ When $M$ is orthogonal, one has $\det M=\pm1$, and the previous formula is exactly what you guessed.

$\endgroup$
2
  • $\begingroup$ Recent work (2012) of Richard Brent and Judy-ann Osborn on determinants of binary matrices uses a similar (if not identical) observation by F. Szollozi to get better bounds on such determinants. Gerhard "Ask Me About Binary Matrices" Paseman, 2013.06.23 $\endgroup$ Jun 23, 2013 at 16:56
  • $\begingroup$ Denis, in this special case there is a one line proof. I cannot recall the name of the guy who found it for me, he was a matrix (and other) expert at U.C. San Diego in the late 1980's. $\endgroup$
    – Will Jagy
    Jun 23, 2013 at 17:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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