Generalized derivations on unital algebras determined by action on zero products. (Q2442298)

The authors divide the paper into four sections, where they discuss some matters in each section. In section one the authors present the preliminaries of the subject as definitions with some relation between the concepts as generalized derivations and derivations, also they mention the history of the subject. In this section the authors use \(\mathcal A\) to be a unital algebra having a nontrivial idempotent and \(\mathcal M\) be a unitary \(\mathcal A\)-bimodule, and consider linear maps \(F,G\colon\mathcal A\to\mathcal M\) satisfying \[ F(x)y+xG(y)=0,\text{ whenever \(x,y\in\mathcal A\) are such that }xy=0.\tag{1} \] And pose the question: what is the standard solution of our generalized problem (1)? If \(\mathcal A\) is a zero product determined algebra (e.g. an algebra generated by idempotents) then \(F\) and \(G\) are generalized derivations of the form \[ F(x)=F(1)x+D(x)\text{ and }G(x)=xG(1)+D(x)\tag{2} \] for all \(x\in\mathcal A\), where \(D\colon\mathcal A\to\mathcal M\) is a derivation. In the second section the authors examine the conditions under which a unital algebra containing a nontrivial idempotent is zero product determined. They say that an algebra \(\mathcal A\) is zero product determined if for every \(\mathcal A\)-module \(\mathcal M\) and every bilinear map \(B\colon\mathcal A\times\mathcal A\to\mathcal M\) the following conditions are fulfilled: if for all \(x,y\in\mathcal A\) \(xy=0\) implies \(B(x,y)=0\), then there is a linear map \(T\colon\mathcal A^2\to\mathcal M\) such that \(B(x,y)=T(xy)\) for all \(x,y\in\mathcal A\). In the third section the authors study generalized derivations determined by action on zero product elements. In Proposition 3.2, they show that if \(\mathcal A\) is a zero product determined algebra the only solution of problem (1) is of the form (2). They also give several examples with non standard solutions. Let \(R(\mathcal A)\) denote the subalgebra of \(\mathcal A\) generated by all its idempotents. We define the so-called special left (right) multipliers determined by action on zero product elements as a map \(\widetilde F\colon\mathcal A\to\mathcal M\) (\(\widetilde G\colon\mathcal A\to\mathcal M\)) that vanishes on \(R(\mathcal A)\) and satisfies condition \[ x,y\in\mathcal A,\;xy=0\implies\widetilde F(x)y=0\quad (x\widetilde G(y)=0).\tag{3} \] In the fourth section the main result of the paper is stated. In Theorem 4.3 they show that for a triangular algebra \(\mathcal A\) and under some conditions on bimodule \(\mathcal M\) the maps \(F,G\colon\mathcal A\to\mathcal M\) satisfying (1) are of the form \(F(x)=F(1)x+D(x)+\widetilde F(x)\) and \(G(x)=xG(1)+D(x)+\widetilde G(x)\) for all \(x\in\mathcal A\). Here \(D\colon\mathcal A\to\mathcal M\) is again a derivation and \(\widetilde F,\widetilde G\colon\mathcal A\to\mathcal M\) are the above mentioned special maps satisfying (3). They also give an example showing that this characterization does not hold true for general unital algebra containing a nontrivial idempotent. -- The authors give a good work, the paper contains strong results beside some examples for illustrating some problems that request that.