Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I am unfamiliar with advanced Matrix theory (nor am I a mathematician), so please bear with me.

Is there anything significant about the following Matrix structure? Are there any special symmetries or conserved quantities that can be extracted from this?

$$ \pmatrix{-u_2 & 0 & -\sqrt{2} u_1 & 0 \\ 0 & u_2 & 0 & -\sqrt{2} u_1 \\ \sqrt{2} u_1 & 0 & 0 & 0 \\ 0 & \sqrt{2} u_1 & 0 & 0} $$

where $u_1,u_2$ are real and have a maximum value of $1$.

Edit:

I was reading something about the correspondence between the group of traceless matrices and the SU(2) group (are they isomorphic?). Am I on the right track? Anything about the blocks that stand out? I was also reading about trace conserving "volume", but I am not sure what that means.

share|improve this question
    
Did you miss a minus sign on $u_2$ in the second row? –  Kartik Audhkhasi Aug 24 '12 at 14:46
    
Jennifer, thank you. Kartik, $u_2$ in the second column is positive.The matrix is correct as posted. –  Antillar Maximus Aug 24 '12 at 14:55
5  
The only properties I notice are that it is traceless, invertible (unless $u_1=0$, of course), and the determinant only depends on $u_1$, not on $u_2$. However, it seems that the product of two such matrices also give a matrix of that form. Oh, and the factor $\sqrt{2}$ can be included in $u_1$ to make the matrix simpler. –  celtschk Aug 24 '12 at 15:03
    
celtschk, thank you. I'd like to add a +1, but I don't have enough rep to do so. :). Can something more be said if it is broken into 2x2 blocks? –  Antillar Maximus Aug 24 '12 at 15:27
2  
@AntillarMaximus, what are you doing with these matrices? –  Will Jagy Aug 31 '12 at 18:19
show 3 more comments

2 Answers

up vote 5 down vote accepted

To complile some of the obvious results (which hopefully don't contain mistakes):

It has determinant $4(u_1)^4$. So, it's going to be nonsingular when $u_1>0$. Even when $u_1=0$, as long as it's nonzero, it's still not nilpotent.

If $\lambda$ is an eigenvalue, then the rest are $\overline{\lambda}$, $-\lambda$ and $-\overline{\lambda}$.

It has trace $0$, whence Lie algebra singles it out as an "infinitesimal volume preserving transformation". (That last point requires a bit of background in Lie algebra to understand.) It can be reexpressed as $AB-BA$ for two other matrices $A$ and $B$.

This matrix is always diagonalizable.

It is close but not quite anti-symmetric if $u_1>0$. It is easy to see the symmetric part is in the upper left $2\times 2$ block, and the antisymmetric part is everything else.

I started on the singular values, but confirmed in WolframAlpha that they are complicated.

WolframAlpha also confirms that the QR, LU and SVD decmopositions are complicated.

share|improve this answer
    
The overline denotes complex conjugation, btw. –  rschwieb Aug 31 '12 at 18:10
1  
The characteristic polynomial is $$ \lambda^4 + (4 u_1^2 - u_2^2) \lambda^2 + 4 u_1^4 $$ so we can solve for $\lambda^2$ with the quadratic formula. –  Will Jagy Aug 31 '12 at 18:38
    
Thanks for the summary! Keep 'em coming. I am particularly interested in the Lie Algebra connection you brought up. At this point, I am on a fishing expedition. :) –  Antillar Maximus Aug 31 '12 at 19:04
    
@rschwieb, one can make a Lie algebra out of these. It's nice, the nonzero entry is a little 2 by 2 determinant. –  Will Jagy Sep 1 '12 at 21:17
    
Marked this as an answer. I also found the Lie Algebra based answer very useful, so the bounty goes there. Thanks to all who responded. –  Antillar Maximus Sep 2 '12 at 15:51
show 1 more comment

You do get a Lie algebra if you ignore the bounds you briefly mentioned. First, define $$ L \; = \; \left( \begin{array}{cccc} 1 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 1 \end{array} \right). $$ and $$ R \; = \; \left( \begin{array}{cccc} 0 & 0 & 0 & 1 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 1 & 0 & 0 & 0 \end{array} \right). $$

Next, define a Lie bracket for two of your matrices $V,W$ as $$ [V,W ] = L (V W - W V) R. $$ This gives a Lie algebra, as the various closure properties hold (addition, multiplication by a scalar) as well as anticommutativity and the Jacobi identity $$ [[U,V],W] + [[V,W],U] + [[W,U],V] = 0. $$

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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