1: \begin{abstract} The plant to be stabilized is a

2: system node $\Sigma$ with generating triple $(A,B,C)$ and transfer

3: function $\GGG$, where $A$ generates a contraction semigroup on the

4: Hilbert space $X$. The control and observation operators $B$ and $C$

5: may be unbounded and they are not assumed to be admissible. The

6: crucial assumption is that there exists a bounded operator $E$ such

7: that, if we replace $\GGG(s)$ by $\GGG(s)+E$, the new system

8: $\Sigma_E$ becomes impedance passive. An easier case is when $\GGG$ is

9: already impedance passive and a special case is when \mm $\Sigma$ has

10: colocated sensors and actuators. Such systems include many wave, beam

11: and heat equations with sensors and actuators on the boundary. It has

12: been shown for many particular cases that the feedback $u=-\ka y+v$,

13: where $u$ is the input of the plant and $\ka>0$, stabilizes $\Sigma$,

14: strongly or even exponentially. Here, $y$ is the output of \m $\Sigma$

15: and $v$ is the new input. Our main result is that if for some $E\in

16: \Lscr(U)$, $\Sigma_E$ is impedance passive, and \m $\Sigma$ is

17: approximately observable or approximately controllable in infinite

18: time, then for sufficiently small $\ka$ the closed-loop system is

19: weakly stable. If, moreover, $\sigma(A)\cap i\rline$ is countable,

20: then the closed-loop semigroup and its dual are both strongly

21: stable.

22: \end{abstract}

23: