I am interested in calculating, or bounding in some way, the following determinant \begin{equation} \det\left[\mathcal{I}-Rxx^t\right] \end{equation} Here, $Rxx^t$ is clearly a singular matrix. Im the case that I am particularily interested in, $R$ is an orthogonal matrix with determinant $+1$. I understand that there are ways of doing this using characteristic polynomials and such, but im not educated in those things. Any help greatly appreciated!

NOTE: I do not know $R$, it can be any special orthogonal matrix. I started my derivation with the `matrix determinant lemma' and am not looking to return to this step, but rather to attack the above determinant


You can use the theorem $\det\left(I+uv^\mathrm{T}\right) = 1+u^\mathrm{T}v$:

\begin{align} \det(I-Rxx^\mathrm{T}) = 1-(Rx)^\mathrm{T}x = 1-x^\mathrm{T}R^\mathrm{T}x \end{align}

Furthermore $-\left\|x\right\|^2_2\leq x^\mathrm{T}R^\mathrm{T}x\leq \left\|x\right\|^2_2$ if $R$ is orthogonal and no other information is known. Thus, we have for the determinant

\begin{align} 1 - \left\|x\right\|^2_2\leq \det(I-Rxx^\mathrm{T}) \leq 1 + \left\|x\right\|^2_2 \end{align}

where both bounds can be reached for some orthogonal $R$.

  • $\begingroup$ ah yes - infact, im trying to avoid using this as this is the starting point in my derivation. I do not know what $R$ is other than it is special orthogonal. I am trying to show, eventually, that i can make the above determinant negative by choosing my $u$ and $v$. $\endgroup$ – jdizzle Jul 19 '14 at 16:33
  • $\begingroup$ I adapted my answer. If you do not have any further information on $R$, the bounds in the answer are least restrictive. I hope I am understanding your question correct. $\endgroup$ – Dimitar Ho Jul 19 '14 at 16:45

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.