Colocal pairs in perfect rings (Q1976570)

A module \(L_A\) over an associative ring with identity, \(A\), is said to be colocal if it has simple essential socle, and a bimodule \(_RL_A\) is a colocal bimodule if both \(_RL\) and \(L_A\) are colocal modules. The authors consider a colocal module over a left or right perfect ring \(A\) and look for conditions for \(L_A\) to be injective. For a subset \(X\) of \(L_A\), and a subset \(M\) of \(A\), let \(l_X(M)=\{x\in X\mid xM=0\}\), and \(r_M(X)=\{a\in M\mid aM=0\}\). Let \(H=\text{End}_A(L_A)\) and \(f\in A\) be a local idempotent with \(\text{soc}(L_A)\cong fA/fJ\) (where \(\text{soc}\) denotes the socle and \(J\) is the Jacobson radical of \(A\)). If \(Af/r_{Af}(L)_{fAf}\) has finite composition length, then the authors show that the following are equivalent: (1) \(L_A\) is injective; (2) \(_HLf_{fAf}\) is a colocal bimodule and \(r_{Af}(L)=0\); (3) \(_HLf_{fAf}\) is a colocal bimodule and \(M=r_{Af}(l_L(M))\) for every submodule \(M\) of \(Af_{fAf}\). The paper also contains other similar results with injectivity replaced by more general conditions.