• Home
  • Textbooks
  • Linear Algebra in Action
  • Realization theory

Linear Algebra in Action

Harry Dym

Chapter 19

Realization theory - all with Video Answers

Educators


Chapter Questions

Problem 1

Check that the mvf $F(\lambda)$ defined by formula (19.3) does not change if $C$ is replaced by $C S, A$ by $S^{-1} A S$ and $B$ by $S^{-1} B$ for some invertible matrix $S \in \mathbb{C}^{n \times n}$.

Check back soon!

Problem 2

Show that if $F_1(\lambda)=D_1+C_1\left(\lambda I_{n_1}-A_1\right)^{-1} B_1$ is a $p \times q$ mvf and $F_2(\lambda)=D_2+C_2\left(\lambda I_{n_2}-A_2\right)^{-1} B_2$ is a $q \times r$ mvf, then
$$
F_1(\lambda) F_2(\lambda)=D_3+C_3\left(\lambda I_n-A_3\right)^{-1} B_3 \text {, }
$$
where
$$
D_3=D_1 D_2, \quad C_3=\left[\begin{array}{ll}
C_1 & D_1 C_2
\end{array}\right], \quad A_3=\left[\begin{array}{cc}
A_1 & -B_1 C_2 \\
O & A_2
\end{array}\right], \quad B_3=\left[\begin{array}{c}
B_1 D_2 \\
B_2
\end{array}\right]
$$
and $n=n_1+n_2$.

Check back soon!
View

Problem 3

Show that if $D \in \mathbb{C}^{p \times p}$ is invertible, then (19.5)
$$
\left\{D+C\left(\lambda I_n-A\right)^{-1} B\right\}^{-1}=D^{-1}-D^{-1} C\left(\lambda I_n-\left[A-B D^{-1} C\right]\right)^{-1} B D^{-1} \text {. }
$$

Let $C \in \mathbb{C}^{p \times n}, A \in \mathbb{C}^{n \times n}$ and $B \in \mathbb{C}^{n \times q}$. Then the pair $(A, B)$ is said to be controllable if the controllability matrix
$$
\mathfrak{C}=\left[\begin{array}{llll}
B & A B & \cdots & A^{n-1} B
\end{array}\right]
$$
is right invertible, i.e., if
$$
\operatorname{rank} \mathfrak{C}=n
$$
The pair $(C, A)$ is said to be observable if the observability matrix
$$
\mathfrak{O}=\left[\begin{array}{c}
C \\
C A \\
\vdots \\
C A^{n-1}
\end{array}\right]
$$
is left invertible, i.e., if its null space
$$
\mathcal{N}_{\mathfrak{D}}=\{\mathbf{0}\}
$$

Nick Johnson
Nick Johnson
Numerade Educator

Problem 4

Show that $(C, A)$ is an observable pair if and only if the pair $\left(A^H \cdot C^H\right)$ is controllable.

Check back soon!

Problem 5

Show that $(A, B)$ is controllable if and only if condition (4) in Lemma 19.2 is met.

Check back soon!

Problem 6

Show that $(A, B)$ is controllable if and only if condition (5) in Lemma 19.2 is met.

Check back soon!
04:08

Problem 7

Show that $(A, B)$ is controllable if and only if $\mathfrak{C C}^H$ is invertible.

Michael Jacobsen
Michael Jacobsen
Numerade Educator

Problem 8

Verify the equivalence (19.6) and then complete the proof that (3) is equivalent to (1) in Lemma 19.3.

Check back soon!

Problem 9

Show that in Lemma 19.3, (4) is equivalent to (1).

Check back soon!

Problem 10

Show that in Lemma 19.3, (5) is equivalent to (1).

Check back soon!
06:28

Problem 11

Show that the pair $(C, A)$ is observable if and only if the matrix $\mathcal{O}^H \mathcal{O}$ is invertible.

Jimmy Yao
Jimmy Yao
Numerade Educator

Problem 12

Let $F(\lambda)=I_p+C\left(\lambda I_n-A\right)^{-1} B$ and $G(\lambda)=I_p-C_1\left(\lambda I_n-\right.$ $\left.A_1\right)^{-1} B_1$. Show that if $C_1=C$ and $(C, A)$ is an observable pair, then $F(\lambda) G(\lambda)=I_p$ if and only if $B_1=B$ and $A_1=A-B C$.

Check back soon!

Problem 13

Verify the asserted uniqueness of the invertible matrix $Y$ that is constructed in the proof of (2) of Theorem 19.4.

Check back soon!

Problem 14

Verify the assertions in formula (19.8).

Check back soon!

Problem 15

Complete the proof of Lemma 19.6.

Check back soon!

Problem 16

Let $(C, A) \in \mathbb{C}^{p \times n} \times \mathbb{C}^{n \times n}$ be an observable pair and let $\mathbf{u} \in \mathbb{C}^n$. Show that $C\left(\lambda I_n-A\right)^{-1} \mathbf{u}$ has a pole at $\alpha$ if and only if $\left(\lambda I_n-A\right)^{-1} \mathbf{u}$ has a pole at $\alpha$. [HINT: First show that it suffices to focus on the case that $A=C_\alpha^{(n)}$ is a single Jordan cell.]

Check back soon!

Problem 17

Let $(C, A) \in \mathbb{C}^{p \times n} \times \mathbb{C}^{n \times n}$ be an observable pair and let $\mathbf{u}(\lambda)=\mathbf{u}_0+\lambda \mathbf{u}_1+\cdots+\lambda^k \mathbf{u}_k$ be a vector polynomial with coefficients in $\mathbb{C}^n$. Show that $C\left(\lambda I_n-A\right)^{-1} \mathbf{u}(\lambda)$ has a pole at $\alpha$ if and only if $\left(\lambda I_n-A\right)^{-1} \mathbf{u}(\lambda)$ has a pole at $\alpha$. [HINT: Try Exercise 19.16 first to warm up.]

Check back soon!

Problem 18

Show that the coefficients $F_{\ell}^{(j)}$ of $\left(\lambda-\lambda_j\right)^{-\ell}$ in the Laurent expansion of the matrix valued function $F(\lambda)$ considered in Theorem 19.9 are given by the formula
$$
F_{\ell}^{(j)}=C P_j\left(A-\lambda_j I_n\right)^{\ell-1} B \text { for } \ell=1, \cdots, \alpha_j,
$$
where $P_j=U_j V_j$ is the projector defined in the proof of the theorem.

Check back soon!

Problem 19

Show that the projector $P_j$ defined in the proof of Theorem 19.9 is the Riesz projector that is defined by the formula
$$
P_j=\frac{1}{2 \pi i} \int_{\Gamma_j}\left(\lambda I_n-A\right)^{-1} d \lambda,
$$
if $\Gamma_j$ is a small enough circle centered at $\lambda_j$ and directed counterclockwise.

Check back soon!

Problem 20

Show that if $C \in \mathbb{C}^{p \times n}$ and $A \in \mathbb{C}^{n \times n}$, then
$$
\operatorname{rank}\left[\begin{array}{l}
C \\
C A \\
\vdots \\
C A^{\ell}
\end{array}\right]=\operatorname{rank}\left[\begin{array}{l}
C \\
C\left(\alpha I_n+A\right) \\
C\left(\alpha I_n+A\right)^{\ell}
\end{array}\right]
$$
for every positive integer $\ell$ and every point $\alpha \in \mathbb{C}$.

Check back soon!

Problem 21

Show that if $A \in \mathbb{C}^{n \times n}$ and $B \in \mathbb{C}^{n \times q}$, then
$$
\operatorname{rank}\left[B A B \cdots A^{\ell} B\right]=\operatorname{rank}\left[B\left(\alpha I_n+A\right) B \cdots\left(\alpha I_n+A\right)^{\ell} B\right]
$$
for every positive integer $\ell$ and every point $\alpha \in \mathbb{C}$.

Check back soon!

Problem 22

Show that $(A, B)$ is stabilizable if and only if $\left(B^H, A^H\right)$ is detectable.

Check back soon!

Problem 23

Show that if $(C, A)$ is detectable and $(A, B)$ is stabilizable, then there exist matrices $K$ and $L$ such that
$$
\sigma\left(\left[\begin{array}{cc}
A & -B K \\
L C & A-B K-L C
\end{array}\right]\right) \subset \Pi_{-} .
$$

Check back soon!
07:29

Problem 24

Justify assertion (3) of Lemma 19.12.

Ernest Castorena
Ernest Castorena
Numerade Educator
02:50

Problem 25

Verify formulas (19.13) and (19.14) and check that the inner product in the latter is well defined; i.e., if $F X \mathbf{u}_1=F X \mathbf{u}_2$ and $F X \mathbf{v}_1=F X \mathbf{v}_2$, then $\mathbf{v}_1^H X \mathbf{u}_1=\mathbf{v}_2^H X \mathbf{u}_2$.

James Kiss
James Kiss
Numerade Educator
01:13

Problem 26

Show that if $J \in \mathbb{C}^{m \times m}$ is a signature matrix, then either $J= \pm I_m$ or $J=U$ diag $\left\{I_p,-I_q\right\} U^H$, with $U$ unitary, $p \geq 1, q \geq 1$ and $p+q=m$.

Srilakshmi E K
Srilakshmi E K
Numerade Educator
04:59

Problem 27

Complete the proof of Lemma 19.16 by justifying the implications $(4) \Longrightarrow(1) \Longrightarrow(2)$.

Stanley Enemuo
Stanley Enemuo
Numerade Educator
02:10

Problem 28

Verify the statements in (19.32).

Manik Pulyani
Manik Pulyani
Numerade Educator

Problem 29

Verify formula (19.34). [HINT: The trick in this calculation (and others of this kind) is to note that in the product, the two terms
$$
C\left(\lambda I_n-A\right)^{-1} X C^H J+C\left(\lambda I_n-A_1\right)^{-1} X_1 C^H J C\left(\lambda I_n-A\right)^{-1} X C^H J
$$
can be reexpressed as
$$
C\left(\lambda I_n-A_1\right)^{-1}\left\{\lambda I_n-A_1+X_1 C^H J C\right\}\left(\lambda I_n-A\right)^{-1} X C^H J,
$$
which simplifies beautifully.]

Check back soon!

Problem 30

Show that if $f \in \mathcal{M}_2$, then
$$
\left\|\vartheta_1 f\right\|_{\mathcal{M}}=\|f\|_{\mathcal{M}_2} .
$$

Check back soon!

Problem 31

Show that $\operatorname{rank} X_2=\operatorname{rank} X-\operatorname{rank} X_1$.

Check back soon!
01:12

Problem 32

Let $A \in \mathbb{C}^{n \times n}, C \in \mathbb{C}^{m \times n}, J \in \mathbb{C}^{m \times m}$, let $\lambda_1, \ldots, \lambda_k$ denote the distinct eigenvalues of $A$; and let $P$ be a solution of the Stein equation $P-A^H P A=C^H J C$. Show that if $1-\lambda_i \overline{\lambda_j} \neq 0$ for $i, j=1, \ldots, k$, then $\mathcal{N}_{\mathfrak{D}} \subseteq \mathcal{N}_P$. [HINT: $\mathcal{N}_{\mathcal{D}}$ is invariant under A.]

Manik Pulyani
Manik Pulyani
Numerade Educator

Problem 33

Let $A \in \mathbb{C}^{n \times n}, C \in \mathbb{C}^{m \times n}, J \in \mathbb{C}^{m \times m}$; and let $P$ be a solution of the Lyapunov equation $A^H P+P A=C^H J C$. Show that if $\sigma(A) \cap \sigma\left(-A^H\right)=\emptyset$, then $\mathcal{N}_{\mathfrak{D}} \subseteq \mathcal{N}_P$.

Check back soon!