Suppose I have the system:

$\dot{x} = Ax+Bu$


and the following Hamiltonian matrix:

$H=\begin{pmatrix} A & \frac{1}{2}B^TB\\ -CC^T&-A \end{pmatrix}$

I want to find the value of $\gamma$ which is the bound of the $H_{\infty}$ norm, so it is the value such that $\left |T(j\omega) \right |_{\infty }<\gamma$.

I know that for the bounded real lemma, if the eigenvalues of $A$ have negative real part, and $I\gamma^2-DD^T>0$, then the Hamiltonian have no eigenvalues on the imaginary axis. I also know that if the eigenvalues of $A$ have negative real part, then $\left |T(j\omega) \right |_{\infty }<\gamma$ holds.

But my question is : how do I find the value of $\gamma$?

I have been told the result is $\gamma=0.5$ but I really can't get to this result. I have tried using the Shur's complement to see if this matrix is negative definite (so before doing that i switched sign to the Hamiltonian). In this way i thiught that if the A matrix is negative definite, it has all eigenvalues with negative real part, so the resulting value of $\gamma$ form the computation would have been the searched value. But I don't find the desired result. Maybe I am missing a point and doing something wrong, or maybe I am complitely on the wrong path.

Can somebody please help me? Thank's in advance.

  • $\begingroup$ You may have better luck if you migrate this question to the computational science group scicomp.stackexchange.com or the maths stack. Yours is essentially a linear algebra problem and there may be experts in those groups. Also, you may want to define the operator $T$. $\endgroup$ Commented Sep 25, 2019 at 20:32

1 Answer 1


I believe you should take a look of the real bounded lemma. Because the hamiltonian matrix has another shape. I quote the book from Zhou - Essentials of Robust Control p.238. $$ H=\begin{bmatrix} A+BR^{-1}D^{\top}C & BR^{-1}B^{\top} \\ -C^{\top}(I+DR^{-1}D^{\top})C & -(A+BR^{-1}D^{\top}C)^{\top} \end{bmatrix} \ $$ where $R=\gamma^2I-D^{\top}D$.

I would compare both hamiltionian matrixes and try to find $\gamma$.


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.