Formulate a large LMI

47 Views Asked by At

I have read that this inequality

$$ P - \sum_{j=0}^{m} \left( A^{(j)} + B^{(j)} K \right)^{T} P \left(A^{(j)} + B^{(j)} K\right) \succ 0 $$

can be reformulated into a large linear matrix inequality (LMI)

$$\left[\begin{array}{ccccc}S & \left(A^{(0)} S+B^{(0)} Y\right)^{T} & \left(A^{(1)} S+B^{(1)} Y\right)^{T} & \cdots & \left(A^{(m)} S+B^{(m)} Y\right)^{T} \\ \star & S & 0 & \cdots & 0 \\ \star & \star & S & \cdots & 0 \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ \star & \star & \star & \cdots & S\end{array}\right] \succ 0$$

by defining $S=P^{-1}$, and $Y=K P^{-1}$.

I know how to reformulate the inequality into a single LMI by using Schur complements, if the inequality does not take the sum of these terms. However, I do not know by which lemma this large LMI can be formulated.

Any hint would be appreciated. Many thanks in advance.

1

There are 1 best solutions below

0
On BEST ANSWER

I'm gonna answer my question myself, thanks for Panos' hint.

According to the Schur complement, $$ X \succ 0 $$ iff $$ C \succ 0, A-B C^{-1} B^{\top}$$ where $$ X=\left[\begin{array}{cc} A & B \\ B^{\top} & C \end{array}\right]. $$

Let $A = S$, $B^{T} = \left[ \left(A^{(0)} S+B^{(0)} Y\right)^{T} \quad\left(A^{(1)} S+B^{(1)} Y\right)^{T} \quad \cdots \quad\left(A^{(m)} S+B^{(m)} Y\right)^{T} \right]$ and $C=\left[ \begin{array}{cccc} S & 0 & \cdots & 0 \\ \star & S & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ \star & \star & \cdots & S \end{array} \right]$.

Then, $A - B^{T}C^{-1}B = S - \Sigma_i (A^{(i)}S + B^{(i)}Y)^{T}S^{-1}(A^{(i)}S + B^{(i)}Y)$. After inserting $S=P^{-1}, \text { and } Y=K P^{-1}$ into the equation above, we prove the first inequality is equivalent to the large LMI.