2
$\begingroup$

Let $C$ be a given (known) matrix and let $\theta$ be a given (known) positive real. We would like to determine the matrices $X$ and $Y$ and diagonal matrix $P$ solving the following inequality

\begin{equation} (XC)^{\text{T}}e{}^{(-YC)^{\text{T}}\theta}Pe{}^{(-YC)\theta}(XC)-P\prec0, \end{equation}

which is equivalent to \begin{equation} P-(XC)^{\text{T}}e{}^{(-YC)^{\text{T}}\theta}Pe{}^{(-YC)\theta}(XC)\succ0. \end{equation} or \begin{equation} P-(XC)^{\text{T}}e{}^{(-YC)^{\text{T}}\theta}PP^{-1}Pe{}^{(-YC)\theta}(XC)\succ0. \end{equation} Using Schur complement, it is equivalent to \begin{equation} \begin{bmatrix}P & (XC)^{\text{T}}e{}^{(-YC)^{\text{T}}\theta}\\ e^{(-YC)\theta}(XC) & P^{-1} \end{bmatrix}\succ0, \end{equation} or \begin{equation} \begin{bmatrix}P & (XC)^{\text{T}}e{}^{(-YC)^{\text{T}}\theta}P\\ Pe^{(-YC)\theta}(XC) & P \end{bmatrix}\succ0. \end{equation}

Does anyone know how to transform these inequalities into linear matrix inequalities (LMIs) please? Or does anyone know how to solve these inequalities in order to find the matrices $X$ and $Y$ please? Thanks.

$\endgroup$
  • $\begingroup$ Please provide background and motivation. Does this come from the stability of discrete-time linear systems? $\endgroup$ – Rodrigo de Azevedo May 14 '17 at 11:56
  • $\begingroup$ Thanks Rodrigo. Yes, it comes from the stability of discrete-time linear systems. I am actually dealing with linear impulsive systems. $\endgroup$ – G. Trav May 15 '17 at 19:35
2
$\begingroup$

Choosing $\mathrm X = \mathrm I$ and $\mathrm Y = \mathrm O$, we obtain the following Lyapunov linear matrix inequality (LMI) in $\mathrm P$

$$\mathrm C^{\top} \mathrm P \, \mathrm C - \mathrm P \prec \mathrm O$$

We can choose a positive definite matrix $\mathrm Q$ and solve the following Lyapunov equation in $\mathrm P$

$$\mathrm C^{\top} \mathrm P \, \mathrm C - \mathrm P + \mathrm Q = \mathrm O$$

Vectorizing, we obtain a system of linear equations

$$\left( \mathrm I - (\mathrm C \otimes \mathrm C)^{\top} \right) \mbox{vec} (\mathrm P) = \mbox{vec} (\mathrm Q)$$

$\endgroup$
  • $\begingroup$ Thank you very much for your answer Rodrigo. The case $X = I$ and $Y = 0$ is a trivial case. However, we are more interested in how to solve this inequality in general cases. $\endgroup$ – G. Trav May 15 '17 at 13:10

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.