Generalized Derivations on Modules *   *   2000 Mathematics Subject Classification. Primary 46H25; Secondary 46L57. Keywords and phrases. (generalized) derivation, (generalized) inner derivation, tiangular Banach algebra, Banach module, (generalized) first cohomology group.

Gh. Abbaspour, M. S. Moslehian and A. Niknam

Abstract
Let A   be a Banach algebra and M   be a Banach left A   -module. A linear map δ : M → M   is called a generalized derivation if there exists a derivation d : A → A   such that δ ( a x ) = a δ ( x ) + d ( a ) x ( a ∈ A , x ∈ M )   .
In this paper, we associate a triangular Banach algebra T   to Banach A   -module M   and investigate the relation between generalized derivations on M   and derivations on T   . In particular, we prove that the so-called generalized first cohomology group of M   is isomorphic to the first cohomology group of T   .

1 Introduction.

Throughout the paper, A   denotes a Banach algebra (not necessarily unital) and M   is a Banach left A   -module.
A linear mapping d : A → A   is called a derivation if d ( a b ) = a d ( b ) + d ( a ) b ( a , b ∈ A )   .
If a ∈ A   and we define d a   by d a ( x ) = a x − x a ( x ∈ A )   . Then d a   is a derivation and such derivation is called inner.
A linear mapping δ : M → M   is called a generalized derivation if there exists a derivation d : A → A   such that δ ( a x ) = a δ ( x ) + d ( a ) x ( x ∈ M , a ∈ A )   . For the convenience, we may say such δ   is a d   -derivation. In general, the derivation d : A → A   is not unique and it may happen that δ   ( d   ) is bounded but d   ( δ   ) is not bounded as the following example shows:
Example 1.1 Assume that the action of A   on M   is trivial, i.e A M = { 0 }   . Then every linear mapping δ : M → M   is indeed a d   -derivation for each derivation d   on A   .
This notion is a generalization of both concepts of generalized derivation (cf. [4 ]    and [9 ]    ) and multiplier (cf. [5 ]    ) on an algebra. For seeing this, regard the algebra as a module over itself. The authors in [1 ]    investigated the generalized derivations on Hilbert C *   -modules and showed that these maps are appeared as the infinitesimal generators of dynamical systems.
Example 1.2 Let M   be Hilbert C *   -module over a C *   -algebra A   of compact operators acting on a Hilbert space (see [12 ]    for more details on Hilbert C *   -modules). By Theorem 4 of [2 ]    , M   has an orthonormal basis and so that each element x   of M   can be expressed as x = ∑ λ < x , v λ > v λ   . If d   is a derivation on A   , then the mapping δ : M → M   defined by δ ( x ) = ∑ λ d ( < x , v λ > ) v λ   is a d   -derivation since
δ ( a x ) = δ ( ∑ λ < a x , v λ > v λ
= ∑ λ d ( a < x , v λ > ) v λ
= ∑ λ a d ( < x , v λ > ) v λ + ∑ λ d ( a ) < x , v λ > v λ
= a δ ( x ) + d ( a ) x .
The set ℬ ( M )   of all bounded module maps on M   is a Banach algebra and M   is a Banach A − ℬ ( M )   -bimodule equipped with x . T = T ( x ) ( x ∈ M , T ∈ ℬ ( M ) )   , since we have ( a x ) . T = T ( a x ) = a T ( x ) = a ( x . T )   and ∥ a x . T ∥ ≤ ∥ a ∥ ∥ x ∥ ∥ T ∥   for all a ∈ A , x ∈ M , T ∈ ℬ ( M )   .
We call δ : M → M   a generalized inner derivation if there exist a ∈ A   and T ∈ ℬ ( M )   such that δ ( x ) = a x − x . T = a x − T ( x )   . Mathieu in [13 ]    called a map δ : A → A   where δ ( x ) = a x − x b   for some a , b ∈ A   a generalized inner derivation. If we consider A   as an A   -module and take T ( x ) = x b   , then our definition covers the notion of Mathieu. In this paper we deal with the derivations on the triangular Banach algebras of the form T = ( A M 0 ℬ ( M ) )   . Such algebras were introduced by Forrest and Marcoux [6 ]    that in turn are motivated by work of Gilfeather and Smith in [8 ]    (these algebras also investigated by Y. Zhang who called them module extension Banach algebras [18 ]    ). Among some facts on generalized derivations, we investigate the relation between generalized derivations on M   and derivations on T   . In particular, we show that the generalized first cohomology group of M   is isomorphic to the first cohomology group of T   .

2 Main Results.

If we consider A   as an A   -module then we have the following lemma about generalized derivations on A   .
Lemma 2.1 A linear mapping δ : A → A   is a generalized derivation if and only if there exist a derivation d : A → A   and a module map φ : A → A   such that δ = d + φ   .
Proof. Suppose δ   be a generalized derivation on A   , then there exists a derivation d   on A   such that δ   is a d   -derivation. Put φ = δ − d   . Then for each a , x ∈ A   we have φ ( a x ) = δ ( a x ) − d ( a x ) = a δ ( x ) + d ( a ) x − ( a d ( x ) + d ( a ) x ) = a ( δ ( x ) − d ( x ) ) = a φ ( x )   Thus φ   is a module map and δ = d + φ   .
Conversely, let d   be a derivation on A   , φ   be a module map on A   and put δ = d + φ   .
Then clearly δ   is a linear map and δ ( a x ) = d ( a x ) + φ ( a x ) = a d ( x ) + d ( a ) x + a φ ( x ) = a ( d ( x ) + φ ( x ) ) + d ( a ) x = a δ ( x ) + d ( a ) x   for all a , x ∈ A   . Therefore δ   is a d   -derivation. □   The next two results concern the boundedness of a generalized derivation.
Theorem 2.2 Let A   have a bounded right approximate identity { e α } α ∈ I   and let δ   be a d   -derivation on A   . Then δ   is bounded if and only if so is d   .
Proof. First we show that every module map on A   is bounded. suppose that φ   is a module map on A   and let { a n }   is a sequence in A   converging to zero. By a consequence of Cohen Factorization Theorem (see Corollary 11.12 of [3 ]    ) there exist a sequence { b n }   and an element c   in A   such that b n → 0   and a n = b n c , n ∈ N   . Then φ ( a n ) = φ ( b n c ) = b n φ ( c ) → 0   .
Thus φ   is bounded.
Now let δ   be a d   -derivation. By Lemma 2.1, δ = d + φ   for some module map φ   on A   .
Therefore δ   is bounded if and only if d   is bounded. □  
Corollary 2.3 Every generalized derivation on a C *   -algebra is bounded.
Proof. Every derivation on a C *   -algebra is automatically continuous; cf. [10 ]    . □   Let φ : A → A   be a homomorphism (algebra morphism). A linear mapping T : M → M   is called a φ   -morphism if T ( a x ) = φ ( a ) T ( x ) ( a ∈ A , x ∈ M )   . If φ   is a isomorphism and T is a bijective mapping then we say T to be a φ   -isomorphism. An i d A   -morphism is a module map (module morphism). Here i d A   denotes the identity operator on A   .
Proposition 2.4 Suppose δ   is a bounded d   -derivation on M   and d   is bounded. Then T = exp ( δ )   is a bi-continuous exp ( d )   -isomorphism.
Proof. Using induction one can easily show that δ ( n ) ( a x ) = ∑ r = 0 n ( r n ) d ( n − r ) ( a ) δ ( r ) ( x ) .   For each a ∈ A , x ∈ M   we have
T ( a x ) = exp ( δ ) ( a x )
= ∑ n = 0 ∞ 1 n ! δ ( n ) ( a x )
= ∑ n = 0 ∞ 1 n ! ∑ r = 0 n ( r n ) d ( n − r ) ( a ) δ ( r ) ( x )
= ∑ n = 0 ∞ ∑ r = 0 n ( 1 ( n − r ) ! d ( n − r ) ( a ) ) ( 1 r ! δ ( r ) ( x ) )
= ( ∑ n = 0 ∞ 1 n ! d ( n ) ( a ) ) ( ∑ n = 0 ∞ 1 n ! δ ( n ) ( a ) )
= exp ( d ) ( a ) exp ( δ ) ( x )
The operators exp ( δ ) , exp ( d )   are invertible in the Banach algebras of bounded operators on M   and A   , respectively. Hence T   is an exp ( d )   -isomorphism. □  
Proposition 2.5 Let δ   be a bounded generalized derivation on M   . Then δ   is a generalized inner derivation if and only if there exists an inner derivation d a   on A   such that δ   is d a   -derivation.
Proof. Let δ   be a generalized inner derivation. Then there exist a ∈ A   and T ∈ ℬ ( M )   such that δ ( a x ) = a x − T ( x ) ( a ∈ A , x ∈ M )   . We have b δ ( x ) + d a ( b ) x = b ( a x − T ( x ) ) + a b x − b a x = a b x − b T ( x ) = a ( b x ) − T ( b x ) = δ ( b x )   . Hence δ   is a d a   -derivation.
Conversely, suppose δ   is a d a   -derivation for some a ∈ A   . Define T : M → M   by T ( x ) = a x − δ ( x )   . Then T is linear, bounded and T ( b x ) = a b x − δ ( b x ) = a b x − ( b δ ( x ) + d a ( b ) x ) = a b x − b δ ( x ) − a b x + b a x = b ( a x − δ ( x ) ) = b T ( x )   . It follows that T ∈ ℬ ( M )   and δ ( x ) = a x − ( a x − δ ( x ) ) = a x − T ( x )   . Therefore δ   is a generalized inner derivation. □   The linear spaces of all bounded generalized derivations and generalized inner derivations on M   are denoted by G Z 1 ( M , M )   and G N 1 ( M , M )   , respectively. We call the quotient space G H 1 ( M , M ) = G Z 1 ( M , M ) / G N 1 ( M , M )   as the generalized first cohomology group of M   .
Corollary 2.6 G H 1 ( M , M ) = 0   whenever H 1 ( A , A ) = 0  
Proof. Let δ : M → M   be a generalized derivation. Then there exists a derivation d : A → A   such that δ   is a d   -derivation. Due to H 1 ( A , A ) = 0   , we deduce that d   is inner and, by Proposition 2.5, so is δ   . Hence G H 1 ( M , M ) = 0 . □   Using some ideas of [6 ]    and [14 ]    , we give the following notion:
Definition 2.7 T = { ( a x 0 T ) ; a ∈ A , x ∈ M , T ∈ ℬ ( M ) }   equipped with the usual 2 × 2   matrix addition and formal multiplication and with the norm ∥ ( a x 0 T ) ∥ = ∥ a ∥ + ∥ x ∥ + ∥ T ∥   is a Banach algebra. We call this algebra the triangular Banach algebra associated to M.
The following two theorems give some interesting relations between generalized derivations on M   and derivations on T   .
Let δ   be a bounded d   -derivation on M   . We define Δ δ : ℬ ( M ) → ℬ ( M )   by Δ δ ( T ) = δ T − T δ   . Then Δ δ   is clearly a derivation on ℬ ( M )   .
Theorem 2.8 Let δ   be a bounded d   -derivation on M   and d   be bounded. Then the map D δ : T → T   defined by D δ ( a x 0 T ) = ( d ( a ) δ ( x ) 0 Δ δ ( T ) )   is a bounded derivation on T   .
Also δ   is a generalized inner derivation if and only if D δ   is an inner derivation.
Proof. It is clear that D δ   is linear. For any a 1 , a 2 ∈ A , x 1 , x 2 ∈ M , T 1 , T 2 ∈ ℬ ( M )   we have
D δ ( ( a 1 x 1 0 T 1 ) ( a 2 x 2 0 T 2 ) ) = D δ ( a 1 a 2 a 1 x 2 + x 1 . T 2 0 T 1 T 2 )
= ( d ( a 1 a 2 ) δ ( a 1 x 2 + x 1 . T 2 ) 0 Δ δ ( T 1 T 2 ) )
= ( a 1 d ( a 2 ) + d ( a 1 ) a 2 a 1 δ ( x 2 ) + d ( a 1 ) x 2 + δ ( T 2 ( x 1 ) ) 0 Δ δ ( T 1 T 2 ) )
= ( a 1 d ( a 2 ) + d ( a 1 ) a 2 a 1 δ ( x 2 ) + ( δ T 2 ) ( x 1 ) − T 2 ( δ ( x 1 ) ) + d ( a 1 ) x 2 + T 2 ( δ ( x 1 ) ) 0 Δ δ ( T 1 T 2 ) )
= ( a 1 d ( a 2 ) + d ( a 1 ) a 2 a 1 δ ( x 2 ) + ( δ T 2 − T 2 δ ) ( x 1 ) + d ( a 1 ) x 2 + T 2 ( δ ( x 1 ) ) 0 Δ δ ( T 1 T 2 ) )
= ( a 1 x 1 0 T 1 ) ( d ( a 2 ) δ ( x 2 ) 0 Δ δ ( T 2 ) ) + ( d ( a 1 ) δ ( x 1 ) 0 Δ δ ( T 1 ) ) ( a 2 x 2 0 T 2 )
= ( a 1 x 1 0 T 1 ) D δ ( a 2 x 2 0 T 2 ) + D δ ( ( a 1 x 1 0 T 1 ) ) ( a 2 x 2 0 T 2 )
Thus D δ   is a derivation on T   . Due to ∥ ( d ( a ) δ ( x ) 0 Δ δ ( T ) ) ∥ = ∥ d ( a ) ∥ + ∥ δ ( x ) ∥ + ∥ Δ δ ( T ) ∥ ≤ max { ∥ d ∥ , ∥ δ ∥ , ∥ Δ δ ∥ } ∥ ( a x 0 T ) ∥   , we infer that D δ   is bounded.
Now suppose that δ   is a generalized inner derivation. Then there exist a ∈ A   and T ∈ ℬ ( M )   such that δ ( x ) = a x − T ( x ) ( x ∈ M )   . For all b ∈ A   , S ∈ ℬ ( M )   and x ∈ M   we have D ( a 0 0 T ) ( b x 0 S ) = ( a 0 0 T ) ( b x 0 S ) − ( b x 0 S ) ( a 0 0 T ) = ( a b − b a a x − x . T 0 T S − S T ) = ( d a ( b ) δ ( x ) 0 Δ δ ( S ) ) = D δ ( b x 0 S )   Hence D δ = D ( a 0 0 T )   and so D δ   is inner.
Conversely, let δ   be a bounded d   -derivation such that the associated derivation D δ   be an inner derivation. It is straightforward to show that d   is inner and so, in virtue of Proposition 2.5, δ   is a generalized inner derivation. □   The converse of the above theorem is true in the unital case.
Theorem 2.9 Let A   be unital and T   be the triangular Banach algebra associated to a unital Banach left A   -module M   . Assume that D : T → T   be a bounded derivation. Then there exist m 0 ∈ M   , a bounded derivation d : A → A   and a bounded d   -derivation δ : M → M   such that D ( a x 0 T ) = ( d ( a ) δ ( x ) + a m 0 − m 0 . T 0 Δ δ ( T ) )   Moreover, D   is inner if and only if δ   is a generalized inner derivation.
Proof. We use some ideas of Proposition 2.1 of [6 ]    (see also [15 ]    ). By simple computation one can verify that (i) D ( 1 A 0 0 0 ) = ( 0 m 0 0 0 )   for some m 0 ∈ M   ; (ii) D ( a 0 0 0 ) = ( d ( a ) a m 0 0 0 )   for some bounded derivation d   on A   ; (iii) D ( 0 x 0 0 ) = ( 0 δ ( x ) 0 0 )   for some bounded d   -derivation δ   on M   ; (iv) D ( 0 0 0 T ) = ( 0 − m 0 . T 0 Δ δ ( T ) )   ; and finally D ( a x 0 T ) = ( d ( a ) δ ( x ) + a m 0 − m 0 . T 0 Δ δ ( T ) )   .
We have
( 0 δ ( a x ) 0 0 ) = D ( ( 0 a x 0 0 ) ) = D ( ( a 0 0 0 ) ( 0 x 0 0 ) )
= ( a 0 0 0 ) D ( ( 0 x 0 0 ) ) + D ( ( a 0 0 0 ) ) ( 0 x 0 0 )
= ( a 0 0 0 ) ( 0 δ ( x ) 0 0 ) + ( d ( a ) a m 0 0 0 ) ( 0 x 0 0 )
= ( 0 a δ ( x ) + d ( a ) x 0 0 )
Thus δ ( a x ) = a δ ( x ) + d ( a ) x   and so δ   is a d   -derivation.
It is clear that D   is inner if and only if d   is inner and, using Proposition 2.5, the later holds if and only if δ   is a generalized inner derivation. □  
Theorem 2.10 Let A   be a unital Banach algebra, M   be a unital Banach left A   -module and T = ( A M 0 ℬ ( M ) )   . Then H 1 ( T , T ) ∼ = G H 1 ( M , M )  
Proof. Let Ψ : G Z 1 ( M , M ) → H 1 ( T , T )   be defined by Ψ ( δ ) = [ D δ ]   where [ D δ ]   represents the equivalence class of D δ   in H 1 ( T , T )   . Clearly Ψ   is linear. We shall show that Ψ   is surjective:
Let D   be a bounded derivation on T   . Let δ   , d   , Δ δ   and m 0 ∈ M   be as in the Theorem 2.9. Then ( D − D δ ) ( a x 0 T ) = ( d ( a ) δ ( x ) + a m 0 − m 0 . T 0 Δ δ ( T ) ) − ( d ( a ) δ ( x ) 0 Δ δ ( T ) ) = ( 0 a m 0 − m 0 . T 0 0 ) = D ( 0 − m 0 0 0 ) ( a x 0 T )   So [ D ] = [ D δ ] = Ψ ( δ )   and thus Ψ   is surjective. Therefore H 1 ( T , T ) ∼ = G Z 1 ( M , M ) / K e r ( Ψ )   .
Note that δ ∈ K e r ( Ψ )   if and only if D δ   is inner derivation on T   . Hence K e r ( Ψ ) = G N 1 ( M , M )   , by Theorem 2.8. Thus H 1 ( T , T ) ∼ = G H 1 ( M , M ) . □  
Example 2.11 Suppose that A   is unital and M = A   . Then ℬ ( A ) = A   and so G H 1 ( A , A ) ∼ = H 1 ( ( A A 0 A ) , ( A A 0 A ) ) = H 1 ( A , A )   , by Proposition 4.4 of [7 ]    . In particular, every generalized derivation on a unital commutative semisimple Banach algebra [17 ]    , a unital simple C *   -algebra [16 ]    , or a von Neumann algebra [11 ]    is generalized inner.
References

  1. Gh. Abbaspour, M. S. Moslehian and A. Niknam, Dynamical systems on Hilbert C *   -modules, arXiv:math.FA/0503615.
  2. D. Bakić and B. Guljaš, Hilbert C *   -modules over C *   -algebras of compact operators, Acta Sci. Math. (Szeged) 68 (2002), no. 1-2, 249-269.
  3. F.F. Bonsall and J. Duncan, Complete Normed Algebras, Springer-Verlag, New York, 1973.
  4. M. Brešar, On the distance of the compositions of two derivations to the generalized derivations, Glasgow Math. J. 33 (1991), 89-93.
  5. H. G. Dales, Banach algebras and automatic continuity, London Mathematical Society Monographs, New Series, 24. Oxford Science Publications. The Clarendon Press, Oxford University Press, New York, 2000.
  6. B. E. Forrest and L. W. Marcoux, Derivations of triangular Banach algebras, Indiana Univ. Math. J. 45 (1996), 441-462.
  7. B. E. Forrest and L. W. Marcoux, Weak amenability of triangular Banach algebras. Trans. Amer. Math. Soc. 354 (2002), no. 4, 1435-1452.
  8. F.L. Gilfeather and R.R. Smith, Cohomology for operator algebras:joins, American Journal of Mathematics 116 (1994), 541-561.
  9. B. Hvala, Generalized derivation in rings, Comm. Algebra 26(4)(1988)1147-66.
  10. B. E. Johnson and A. M. Sinclair, Continuity of derivations and a problem of Kaplansky, Amer. J. Math. 90 (1968), 1067-1073.
  11. R. V. Kadison, Derivations of operator algebras, Ann. of Math. (2) 83 (1966), 280-293.
  12. E. C. Lance, Hilbert C *   -modules, LMS Lecture Note Series 210, Cambridge Univ. Press, 1995.
  13. M. Mathieu, Elementary operators and applications, Proceeding of the International workshop, World Scientific, Singapore 1992.
  14. M. S. Moslehian, On (co)homology of triangular Banach algbras, Topological algebras and its related topics, Banach Center Publ. 67, Polish Acad. Sci., Warsaw, 2005.
  15. M. S. Moslehian, Vanishing of cohomology groups of triangular Banach algebras, arXiv:math.FA/0502346.
  16. S. Sakai, C *   -algebras and W *   -algebras, Classics in Mathematics. Springer-Verlag, Berlin, 1998.
  17. I. M. Singer and J. Wermer, Derivations on commutative normed algebras, Math. Ann. 129, (1955), 260-264.
  18. Y. Zhang, Weak amenability of module extensions of Banach algebras, Trans. Amer. Math. Soc. 354 (2002), 4131-4151.

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