Derivations on Group Algebras (Q4487609)

Let \(G\) be a locally compact group, \(L^1(G)\) the convolution algebra of integrable functions and \(M(G)\) the convolution algebra of regular Borel measures on \(G\). Convolution with \(L^1\)-functions makes \(M(G)\) an \(L^1(G)\)-bimodule. It is a natural question of whether every continuous derivation from \(L^1(G)\) into \(M(G)\) is an inner derivation, that is, whether the first cohomology space \(H^1(L^1(G), M(G))\) of \(L^1(G)\) with coefficients in \(M(G)\) is trivial. \textit{B. E. Johnson} [Cohomology in Banach algebras, Mem. Am. Math. Soc. 127 (1972; Zbl 0256.18014)] was the first to investigate this problem. He has shown that \(H^1(L^1(G), M(G))= \{0\}\) whenever \(G\) is amenable or \(G\) is a SIN-group (a group that has a neighbourhood basis of the identity consisting of conjugation invariant sets). Not a single group is known for which \(H^1(L^1(G), M(G))\neq \{0\}\). The question is even open for the semidirect product \(G= \text{SL}(2,\mathbb{Z})\ltimes\mathbb{T}^2\) where \(\text{SL}(2,\mathbb{Z})\) acts on \(\mathbb{T}^2= \mathbb{R}^2/\mathbb{Z}^2\) in the obvious manner. It seems that this group \(G\) is viewed as a test example. In the paper under review a number of non-trivial partial results are proved, but no substantial progress is achieved towards the main problem. In Theorem 2.7 it is shown that for any unimodular locally compact, \(H^1(L^1(G), L^1(G))\) is Hausdorff. Note that in general such \(H^1\)-spaces need not be Hausdorff. For derivations from \(L^1(G)\) into \(L^1(G)\) satisfying a certain condition, it is shown that they are implemented by elements of \(VN(G)\), the von Neumann algebra generated by the regular representation of \(L^1(G)\) (Theorem 3.6). According to Corollary 4.3, this conclusion holds for IN-groups (groups having at least one compact invariant neighbourhood of the identity). In particular, it applies to the example mentioned above. In Proposition 7.4, building on arguments of Johnson, the authors prove that any maximally almost periodic locally compact group satisfies \(H^1(L^1(G), M(G)) =\{0\}\). Finally, Corollary 2.4 says that \(H^1(L^1(G), M(G))= \{0\}\) for a unimodular group \(G\) whose reduced group \(C^*\)-algebra possesses a central bounded approximate identity. However, by a result of Losert, this condition forces \(G\) to be a SIN-group. Thus Corollary 2.4 just reproves one of Johnson's results.