The general theory of dynamical systems is the paradigm for modeling and studying phenomena that undergo spatial and temporal evolution. The application of dynamical systems has nowadays spread to a wide spectrum of disciplines including physics, chemistry, biochemistry, biology, economy and even sociology. In particular, theory of $C^{*}$  -dynamical systems concerns the theory of derivations in $C^{*}$  -algebras and is motivated by questions in quantum physics and statistical mechanics.

In this paper We investigate the generalized derivations This notion was first appeared in the context of operator algebras [9 ]  $$  . Later, it was introduced in the framework of pure algebra [5 ]  $$  . Recently the stability of these maps are investigated by the second author [11 ]  $$  .

We shall show that every generalized derivation $\delta$  on a simple Hilbert $C^{*}$  -module $M$  either is closable or has a dense range in $M$  .

We also describe dynamical systems on a full Hilbert $C^{*}$  -module $M$  over a $C^{*}$  -algebra $A$  as a one-parameter group of unitaries on $M$  and prove that if $\alpha$  is such a dynamical system, then we can associate a $C^{*}$  -dynamical system ${\alpha }^{{}^{\prime }}$  on $A$  such that if $\delta$  and $d$  are the infinitesimal generators of $\alpha$  and $d$  respectively, then $\delta$  is a $d$  -generalized derivation.

The reader is referred to [8 ]  $$  and [15 ]  $$  for more details on Hilbert $C^{*}$  -modules and to [17 ]  $$  formore information on $C^{*}$  -dynamical systems.

2 Preliminaries.

Throughout this section $M$  and $N$  are assumed to be Hilbert modules over $C^{*}$  -algebras $A$  and $B$  respectively, and $\phi :A\to B$  is a morphism of $C^{*}$  -algebras.

A map $\Phi :M\to N$  is said to be a $\phi$  -morphism of Hilbert $C^{*}$  -modules if $$<\Phi \left(x\right),\Phi \left(y\right)>=\phi (<x,y>)(x,y\in M)$$  Applying the polarization we can immediately conclude that $\Phi$  is a $\phi$  -morphism if and only if $<\Phi \left(x\right),\Phi \left(x\right)>=\phi (<x,x>)$  for all $x\in M$  . Each $\phi$  -morphism is necessarily a linear operator since for any $x,y,z\in M$  and $\alpha \in \mathbf{C}$  we have $$<\Phi (\alpha x+y),\Phi \left(z\right)>=\alpha \phi (<x,z>)+\phi (<y,z>)=\alpha <\Phi \left(x\right),\Phi \left(z\right)>+<\Phi \left(y\right),\Phi \left(z\right)>$$  and hence $<\Phi (\alpha x+y)-\alpha \Phi \left(x\right)-\Phi \left(y\right),\Phi \left(z\right)>=0$  for all $z\in M$  . If we replace $z$  with $x,y$  and $\alpha x+y$  then we can infer that $$<\Phi (\alpha x+y)-\alpha \Phi \left(x\right)-\Phi \left(y\right),\Phi (\alpha x+y)-\alpha \Phi \left(x\right)-\Phi \left(y\right)>=0$$  Thus $\Phi (\alpha x+y)-\alpha \Phi \left(x\right)-\Phi \left(y\right)=0$  and therefore $\Phi (\alpha x+y)=\alpha \Phi \left(x\right)+\Phi \left(y\right)$  . Simillary every $\phi$  -morphism is necessarily a module map in the sense that $\Phi \left(ax\right)=\phi \left(a\right)\Phi \left(x\right)$  for all $x\in M$  and for all $a\in A$  .

Further, let $\psi :B\to C$  be a morphism of $C^{*}$  -algebras and let $Q$  be a Hilbert $C^{*}$  -module over $C$  . If $\Phi :M\to N$  is a $\phi$  -morphism and $\Psi :N\to Q$  is a $\psi$  -morphism, then obviously $\Psi \Phi :M\to Q$  is a $\psi \phi$  -morphism of Hilbert $C^{*}$  -modules. By following [2 ]  $$  we call a map $\Phi :M\to N$  a unitary operator if there exists an injective morphism of $C^{*}$  -algebras $\phi :A\to B$  such that $\Phi$  is a surjection $\phi$  -morphism.

Remark 2.1.

(i) If $\Phi :M\to N$  be a $\phi$  -morphism of Hilbert $C^{*}$  -modules and $\phi$  is injective then $\Phi$  is an isometry; cf. [2 ]  $$  . Thus each unitary operator of Hilbert $C^{*}$  -modules is necessarily an isometry.

(ii) It is known [2 ]  $$  that if $\Phi :M\to N$  be a $\phi$  -morphism of Hilbert $C^{*}$  -modules then $Im\Phi$  is a closed subspace of N and a Hilbert $C^{*}$  -module over the $C^{*}$  -algebra $Im\phi \subset B$  such that $<Im\Phi ,Im\Phi >=\phi (<M,M>)$  . Moreover if $\Phi$  is surjective, and if $N$  is a full $B$  -module, then $\phi$  is also a surjection. Thus if $\Phi$  is a unitary then it is surjective and hence $<N,N>=\phi (<M,M>)\simeq <M,M>$  . If in addition $N$  is a full Hilbert $B$  -module, then $\phi$  is also surjective and so is an isomorphism of $C^{*}$  -algebras.

Example 2.2. Let $H$  be a Hilbert space. Then $H$  can be regarded as a $K\left(H\right)$  -module by $T.x=T\left(x\right)(T\in K(H),x\in H)$  . If we define a $K\left(H\right)$  -inner product on $H$  by $<x,y>=x\otimes yx,y\in H$  , where $x\otimes y\left(z\right)=(z,y)xz\in H$  and $(.,.)$  denotes the complex inner product on $H$  , then $H$  can be regarded as a Hilbert $C^{*}$  -module over $K\left(H\right)$  . In this case, $U:H\to H$  is a unitary as an operator on Hilbert space $H$  if and only if $U$  is a unitary operator as above definition on Hilbert $K\left(H\right)$  -module $H$  . This is a consequence of the facts that $<U\left(x\right),U\left(y\right)>=U\left(x\right)\otimes U\left(y\right)=U(x\otimes y)U^{*}$  , and that $AdU:K\left(H\right)\to K\left(H\right)$  defined by $AdU\left(V\right)=UV{U}^{*},V\in K\left(H\right)$  is a $*$  -isomorphism and each $*$  -isomorphism on $K\left(H\right)$  is of this form; cf. [12 ]  $$  .

We denote by $U\left(M\right)$  the group of all unitary operators of $M$  onto $M$  . If $M$  is full and $\alpha :M\to M$  is a unitary operator then by Remark 2.1.ii there is a $*$  -isomorphism ${\alpha }^{{}^{\prime }}:A\to A$  such that $\alpha$  is a ${\alpha }^{{}^{\prime }}$  -morphism.

We end this section with the following useful lemma which can be found in [1 ]  $$  and [10 ]  $$  .

Lemma 2.3. Let $M$  be a full Hilbert module over the $C^{*}$  -algebra $A$  and let $a\in A$  . Then $a=0$  if and only if $ax=0$  for all $x\in M$  .

3 Generalized Derivation

This section is devoted to study of generalized derivations. Our aim is to show that every generalized derivation $\delta$  on a simple Hilbert $C^{*}$  -module either is closable or has a dense range.

Definition 3.1. Let $M$  be full Hilbert $A$  -module. A linear map $\delta :D\left(\delta \right)\subseteq M\to M$  , where $D\left(\delta \right)$  is a dense subspace of $M$  , is called a generalized derivation if there exists a mapping $d:D\left(d\right)\to A$  , where $D\left(d\right)$  is a dense subalgebra of $A$  such that $D\left(\delta \right)$  is an algebraic left $D\left(d\right)$  -module, and $\delta \left(ax\right)=a\delta \left(x\right)+d\left(a\right)x$  for all $x\in D\left(\delta \right)$  and all $a\in D\left(d\right)$  .

In this case $d$  must be a derivation since for any $a,b\in D\left(d\right)$  and $x\in D\left(\delta \right)$  we have $\delta \left(abx\right)=ab\delta \left(x\right)+d\left(ab\right)x$  .

On the other hand $\delta \left(abx\right)=\delta \left(a\right(bx\left)\right)=a\delta \left(bx\right)+d\left(a\right)bx=ab\delta \left(x\right)+ad\left(b\right)x+d\left(a\right)bx$  .

So $\left(d\right(ab)-(ad\left(b\right)+d\left(a\right)b)x=0$  for all $x\in D\left(\delta \right)$  . $D\left(\delta \right)$  is dense in $M$  thus by lemma 2.3 we obtain $d\left(ab\right)=ad\left(b\right)+d\left(a\right)b$  .

Similarly we can show $d$  is linear so $d:D\left(d\right)\subseteq A\to A$  is a derivation. Sometimes we call $\delta$  a $d$  -derivation.

Denote by $GDer\left(M\right)$  the set of all generalized derivations on $M$  then it is easy to see that $GDer\left(M\right)$  is a linear space. In fact if ${\delta }_1,{\delta }_2\in GDer\left(M\right)$  , ${\delta }_1$  is a $d_1$  -derivation and ${\delta }_2$  is a $d_2$  -derivation, then $\alpha {\delta }_1+\beta {\delta }_2$  is a $\alpha d_1+\beta d_2$  -derivation and so $\alpha {\delta }_1+\beta {\delta }_2\in GDer\left(M\right)$  .

Also the Lie product $[{\delta }_1,{\delta }_2]$  is $[d_1,d_2]$  -derivation , so $[{\delta }_1,{\delta }_2]\in GDer\left(M\right).$  Now we show a similar result as in [13 ]  $$  for generalized derivations:

Theorem 3.2. Let $M$  be a simple full Hilbert $C^{*}$  -module in the sense that it has no trivial left $A$  -module and let $\delta :D\left(\delta \right)\subseteq M\to M$  be a $d$  -derivation. Then either $\delta$  is closable or the range of $\delta$  is dense in $M$  .

Proof. Let $\sigma \left(\delta \right)$  be the separating space of $\delta$  that is $$\sigma \left(\delta \right)=\{x\in M;thereisasequence(x_n)inD(\delta )with{x}_n\to 0,\delta (x_n)\to x\}$$  It is obvious that $\sigma \left(\delta \right)$  is closed subspace of $M$  . We show that $\sigma \left(\delta \right)$  is a left submodule of $M$  . Let $a\in A,x\in \sigma \left(\delta \right)$  thus there exists a sequence $\left(x_n\right)\subseteq D\left(\delta \right)$  such that $x_n\to 0$  and $\delta \left(x_n\right)\to x$  so we have $a{x}_n\to 0$  and $\delta \left(a{x}_n\right)=a\delta \left(x_n\right)+d\left(a\right)x_n\to ax$  So $ax\in \sigma \left(\delta \right)$  . By the hypothesis $\sigma \left(\delta \right)=\left\{0\right\}$  or $\sigma \left(\delta \right)=M$  . Therefore $\delta$  is closable or range $\delta$  is dense in $M.\square$ 

4 Dynamical Systems On Full Hilbert $C^{*}$  -Modules

We start this section with a basic definition. From now on let us $M$  denote a full Hilbert module over the $C^{*}$  -algebra $A$  .

Definition 4.1. Let $M$  be a full Hilbert $A$  -module. A map $\alpha$  from the real line $\mathbf{R}$  to $U\left(M\right)$  which maps $t$  to ${\alpha }_t$  is said to be a one-parameter group of unitaries if (i) ${\alpha }_0=I$  (ii) ${\alpha }_{t+s}={\alpha }_t{\alpha }_s(t,s\in \mathbf{R})$  Further, $\alpha$  is said to be strongly continuous one-parameter group of unitaries if, in addition, for every $x\in M,{\alpha }_t\left(x\right)\to x$  when t tends to zero. In this case we call $\alpha$  a dynamical system on $M$  .

We can define the infinitesimal generator of a dynamical system as follows:

Definition 4.2. Let $\alpha :\mathbf{R}\to U\left(H\right)$  be a dynamical system on $M$  , we define the infinitesimal generator $\delta$  of $\alpha$  as a mapping $\delta :D\left(\delta \right)\subseteq M\to M$  where $D\left(\delta \right)=\{x\in M;{lim}_{t\to 0}\frac{{\alpha }_t\left(x\right)-x}{t}exists\}$  and $$\delta \left(x\right)={lim}_{t\to 0}\frac{{\alpha }_t\left(x\right)-x}{t},x\in D\left(\delta \right)$$  Now we are ready to prove the main theorem of this paper:

Theorem 4.3. Let $M$  be a full Hilbert $A$  -module and $\alpha$  be a dynamical system on $M$  and $\delta$  be the infinitesimal generator of $\alpha$  then $D\left(\delta \right)$  is a dense subspace of $M$  and there exists a derivation $d:D\left(d\right)\subseteq A\to A$  such that $D\left(\delta \right)$  is a left $D\left(d\right)$  -module and $\delta \left(ax\right)=a\delta \left(x\right)+d\left(a\right)x,a\in D\left(d\right),x\in D\left(\delta \right)$  Proof. By Hille-Yosida theorem [4 ]  $$  , $D\left(\delta \right)$  is a dense subspace of $M$  . Since $\alpha$  is a dynamical system on $M$  then for each $t\in \mathbf{R}$  , the mapping ${\alpha }_t:M\to M$  is a unitary. So there exists $*$  -isomorphism ${\alpha }_t^{{}^{\prime }}:A\to A$  such that $<{\alpha }_t\left(x\right),{\alpha }_t\left(y\right)>={\alpha }_t^{{}^{\prime }}(<x,y>)$  and hence ${\alpha }_t\left(ax\right)={\alpha }_t^{{}^{\prime }}\left(a\right){\alpha }_t\left(x\right)(a\in A,x\in M).$  Now we show that ${\alpha }^{\prime }:\mathbf{R}\to Aut\left(A\right)$  is a $C^{*}$  -dynamical system. For each $a\in A,x\in M$  we have $ax={\alpha }_0\left(ax\right)={\alpha }_0^{{}^{\prime }}\left(a\right){\alpha }_0\left(x\right)={\alpha }_0^{{}^{\prime }}\left(a\right)x$  thus by lemma 2.3 ${\alpha }_0^{{}^{\prime }}\left(a\right)=a$  for all $a\in A$  .

Therefore ${\alpha }_0^{{}^{\prime }}=I$  .

Also for all $t,s\in \mathbf{R}$  we have ${\alpha }_{t+s}^{{}^{\prime }}\left(a\right){\alpha }_{t+s}\left(x\right)={\alpha }_{t+s}\left(ax\right)={\alpha }_t\left({\alpha }_s\right(ax\left)\right)={\alpha }_t\left({\alpha }_s^{{}^{\prime }}\right(a\left){\alpha }_s\right(x\left)\right)={\alpha }_t^{{}^{\prime }}\left({\alpha }_s^{{}^{\prime }}\right(a\left)\right){\alpha }_t\left({\alpha }_s\right(x\left)\right)={\alpha }_t^{{}^{\prime }}\left({\alpha }_s^{{}^{\prime }}\right(a\left)\right){\alpha }_{t+s}\left(x\right)$  and so ${\alpha }_{t+s}^{{}^{\prime }}\left(a\right)={\alpha }_t^{{}^{\prime }}\left({\alpha }_s^{{}^{\prime }}\right(a\left)\right)$  , thus ${\alpha }_{t+s}^{{}^{\prime }}={\alpha }_t^{{}^{\prime }}{\alpha }_s^{{}^{\prime }}.$  Since for each $x\in M$  , ${\alpha }_t\left(x\right)\to x$  as $t\to 0$  we have $$\parallel {\alpha }_t^{{}^{\prime }}\left(a\right)x-ax\parallel \le \parallel {\alpha }_t^{{}^{\prime }}\left(a\right)x-{\alpha }_t^{{}^{\prime }}\left(a\right){\alpha }_t\left(x\right)\parallel +\parallel {\alpha }_t^{{}^{\prime }}\left(a\right){\alpha }_t\left(x\right)-ax\parallel$$  thus ${lim}_{t\to 0}{\alpha }_t^{{}^{\prime }}\left(a\right)x=ax$  for all $x\in M$  . So that ${lim}_{t\to 0}{\alpha }_t^{{}^{\prime }}\left(a\right)=a$  .

Therefore ${\alpha }^{\prime }:\mathbf{R}\to Aut\left(A\right)$  is a $C^{*}$  -dynamical system on $A$  .

If $d$  is the infinitesimal generator of ${\alpha }^{\prime }$  then for each $a\in D\left(d\right),x\in D\left(\delta \right)$  we have

$$\begin{array}{ccc}{lim}_{t\to 0}\frac{{\alpha }_t\left(ax\right)-ax}{t}& =& {lim}_{t\to 0}\frac{a{\alpha }_t\left(x\right)-ax}{t}+{lim}_{t\to 0}\frac{{\alpha }_t^{\prime }\left(a\right){\alpha }_t\left(x\right)-a{\alpha }_t\left(x\right)}{t}\end{array}$$

$$\begin{array}{ccc}& =& a{lim}_{t\to 0}\frac{{\alpha }_t\left(x\right)-x}{t}+{lim}_{t\to 0}\frac{{\alpha }_t^{\prime }\left(a\right)-a}{t}{\alpha }_t\left(x\right)\end{array}$$

$$\begin{array}{ccc}& =& a\delta \left(x\right)+d\left(a\right)x\end{array}$$

So $ax\in D\left(\delta \right)$  , $\delta \left(ax\right)=a\delta \left(x\right)+d\left(a\right)x$  . Furthermore $D\left(\delta \right)$  is a left $D\left(d\right)$  -module. $\square$  Example 4.4. Let $H$  be a Hilbert space and $T\in B\left(H\right)$  be a self adjoint operator.

By Stone's theorem there is a one-parameter group of unitaries $t\mapsto {\alpha }_t({\alpha }_t\in B(H\left)\right)$  which its infinitesimal generator is $iT$  . By Example 2.2 this is a dynamical system on Hilbert $K\left(H\right)$  -module $H$  with associated dynamical system $t\mapsto {{\alpha }_t}^{\prime }$  where ${{\alpha }_t}^{\prime }\left(V\right)={\alpha }_{t}V{\alpha }_t^{*}$  . Here ${\alpha }_t^{*}$  denotes the adjoint of ${\alpha }_t$  in $B\left(H\right)$  . If $\delta$  and ${\delta }^{{}^{\prime }}$  are infinitesimal generators of $\alpha$  and ${\alpha }^{\prime }$  respectively then $\delta =iT$  and by theorem 4.3 we have $\delta (V.x)=V.\delta \left(x\right)+{\delta }^{{}^{\prime }}\left(V\right).x$  and hence $\delta \left(V\right(x\left)\right)=V\left(\delta \right(x\left)\right)+{\delta }^{{}^{\prime }}\left(V\right)\left(x\right)$  . Thus $${\delta }^{{}^{\prime }}\left(V\right)\left(x\right)=\delta \left(V\right(x\left)\right)-V\left(\delta \right(x\left)\right)=iT\left(V\right(x\left)\right)-iV\left(T\right(x\left)\right).$$  So $${\delta }^{{}^{\prime }}\left(V\right)=i(TV-VT)=i[T,V],V\in D\left({\delta }^{{}^{\prime }}\right)$$  The density of $D\left({\delta }^{{}^{\prime }}\right)$  in $K\left(H\right)$  and the fact that ${\delta }^{{}^{\prime }}$  has no proper extension implies that ${\delta }^{{}^{\prime }}\left(V\right)=i[T,V],V\in K\left(H\right)$  . References

1. M. Amyari and A. Niknam, A note on finsler modules, Bull. Iran. Math. Soc. Vol. 29 No. 1 (2003), 77-81.
2. D. Bakič and B. Guljaš, On class of module maps of Hilbert $C^{*}$  -modules. Math. Commun.,7,(2002), 177-192.
3. E. G. Effros, Z. J. Ruan, Operator spaces. London Mathematical Society Monographs. New Series, 23. The Clarendon Press, Oxford Univ. Press, New York, 2000.
4. E. Hille and R. Phillips, Functional Analysis and Semi-groups, Proc. of Symp. in Pure Math., vol 31, Amer. Math. Soc., Rhode Island, 1957.
5. B. Hvala, Generalized derivations in rings, Comm. Algebra 26 (1998), 1147-1166.
6. K.K. Jensen and K. Tomsen, Elements of KK-Theory, Birkhauser, Boston-Basel-Berlin, 1991.
7. I. Kaplansky, Modules over operator algebras, Amer. J. Math. 75 (1953), 839-853.
8. E. C. Lance, Hilbert $C^{*}$  -Modules, LMS Lecture Note Series 210, Cambridge Univ. Press, 1995.
9. M. Mathieu, Elementary Operators and Applications, Proceeding of the International Workshop, World Scientific, Singapore, 1992.
10. M. S. Moslehian, On full Hilbert $C^{*}$  -modules. Bull. Malaysian Math. Soc. 24(2001), 45-47.
11. M. S. Moslehian, Hyers–Ulam–Rassias stability of generalized derivations, arXiv:math.FA/0501012.
12. G. J. Murphy, $C^{*}$  -algebras and Operator Theory, Academic Press, 1990.
13. A. Niknam, Closable derivations of simple $C^{*}$  -algebras. Glasgow Math. J. 24(1983)181-183.
14. W. L. Paschke, Inner product modules over $B^{*}$  -algebras, Trans Amer. Math. Soc. 182 (1973),443-468.
15. I. Raeburn and D.P. Williams, Morita equivalence and continuous-trace $C^{*}$  -algebras, Mathematical surveys and Monographs AMS 60, 1998.
16. M. A. Rieffel, Morita equivalence representations of $C^{*}$  -algebras, Adv. in Math., 13 (1974), 176-257.
17. S. Sakai, Operator Algebra in Dynamical Systems. Cambridge Univ. press, 1991.
18. S. L. Woronowicz, compact quantum groups, Symetrics quantiques(Les ouches, 1995), 845-884, North Holland, Amsterdum, 1998.

Address: Dept. of Math., Ferdowsi Univ., P. O. Box 1159, Mashhad, Iran Gholamreza Abbaspour Tabadkan, tabadkan@math.um.ac.ir Mohammad Sal Moslehian, msalm@math.um.ac.ir Assadollah Niknam, niknam@math.um.ac.ir