Dualities between almost completely decomposable groups and their endomorphism rings. (Q2462028): Difference between revisions

The main result of the paper is Theorem~3.4: Let \(X\) and \(X'\) be nearly isomorphic almost completely decomposable groups with common regulator \(A\) and common regulator exponent \(e=\exp(X/A)=\exp(X'/A)\) such that \(X\) and \(X'\) contain no non-zero \(p\)-divisible subgroups for every prime divisor \(p\) of \(e\). Then the additive groups \(\text{End}(X)^+\) and \(\text{End}(X')^+\) are again nearly isomorphic almost completely decomposable groups. In the main theorem the assumption of a common regulator and regulator exponent is no loss of generality but it is also assumed that the critical types are types of ring type. This may actually not be needed. [Comment: Suppose that \(\mathcal N\) is a ``near-isomorphism category for almost completely decomposable groups'' and that \(f\colon X\to X'\) is a near-isomorphism. Then \(\mathcal N(X,X)^+\ni\alpha\mapsto f\alpha f^{-1}\in\mathcal N(X',X')^+\) is a near-isomorphism. This differs from the result of the author because \(\text{End}(X)\) may be properly contained in \(\mathcal N(X,X)\).] The ``duality'' mentioned in the title is as follows. Let \(X\) be an almost completely decomposable group and \(A\) its regulator. The \(p\)-constituent of \(X\) is its subgroup \(X_p\) determined by the property that \(X_p/A\) equals the \(p\)-primary component of the finite Abelian group \(X/A\). Then \(A\) is the regulator of all the groups \(X_p\) and is fully invariant in \(X\) and in the \(X_p\). By restriction one gets embeddings of \(\text{End\,}X\) and \(\text{End\,}X_p\) in \(\text{End\,}A\). For a set \(P\) of prime divisors of \(\exp(X/A)\) the author sets \(X(P):=\sum_{p\in P}X_p\) and again obtains an embedding \(\text{End\,}X(P)\subseteq\text{End\,}A\). The ``duality'' is the assignment \(X(P)\mapsto\text{End\,}X(P)\subseteq\text{End\,}A\) whose properties are studied.