On a filtered multiplicative basis of group algebras. II. (Q1411047)

Let \(R\) be a finite-dimensional algebra over a field \(K\). A \(K\)-basis \(S\) of \(R\) is a filtered multiplicative \(K\)-basis (f.m. \(K\)-basis) of \(R\) if (i) For all \(a,b\in S\) either \(ab=0\) or \(ab\in S\) and (ii) \(S\cap\text{rad\,}R\) is a \(K\)-basis of \(\text{rad\,}R\). f.m. \(K\)-bases were introduced by the reviewer in connection with the study of Nakayama-algebras (= generalized uniserial algebras) [\textit{H. Kupisch}, J. Reine Angew. Math. 201, 100-112 (1959; Zbl 0094.25101)] and, more generally, of algebras of finite (representation-) type [\textit{H. Kupisch}, J. Reine Angew. Math. 219, 1-25 (1965; Zbl 0132.28002)]. For algebras of finite type over an algebraically closed field \(K\) f.m. \(K\)-bases do exist [\textit{R. Bautista}, \textit{P. Gabriel}, \textit{A. V. Rojter} and \textit{L. Salmerón}, Invent. Math. 81, 217-285 (1985; Zbl 0575.16012)]. However, finite type is not necessary for their existence. In case of group-algebras \(KG\) of finite groups \(G\) over a field \(K\) of characteristic \(p\) it is well-known that \(KG\) is of finite type iff the Sylow \(p\)-subgroups of \(G\) are cyclic. The question now is: When does \(KG\) have an f.m. \(K\)-basis in the non-cyclic case. In the paper this question is studied (as in the preceding part I [\textit{V. Bovdi}, Arch. Math. 74, No. 2, 81-88 (2000; Zbl 0986.20004)]) for \(p\)-groups \(G\). The main results are: (1) Let \(G\) be a non-Abelian \(p\)-group with a cyclic subgroup of index \(p^2\). Then \(KG\) has an f.m. \(K\)-basis iff \(p=2\) and one of the following conditions holds: (i) \(G\) is either the dihedral 2-group, or \(D_{2^m}\times C_2\), or the central product of \(D_8\) and \(C_4\). (ii) \(K\) contains a primitive cube root of unity and \(G\) is either \(Q_8\) or \(Q_8\times C_2\). (iii) \(G\) is one of 9 groups, explicitly given by generators and relations. (2) \(KG\) does not have an f.m. \(K\)-basis if (i) \(G\) is a non-Abelian powerful \(p\)-group (i.e. \(G/G^4\) resp. \(G/G^p\) is Abelian according to whether \(p=2\) or \(p>2\); or (ii) \(p\) is odd and \(G\) is a 2-generated non-Abelian \(p\)-group with a central cyclic commutator subgroup. Essential for the proof of (2) are results of \textit{S. A. Jennings} [Trans. Am. Math. Soc. 50, 175-185 (1941; Zbl 0025.24401)] and \textit{R. J. Miech} [J. Aust. Math. Soc. 20, 178-198 (1975; Zbl 0319.20024)], for the proof of (1), the list of groups in \textit{Y. Ninomiya}'s paper [Math. J. Okayama Univ. 36, 1-21 (1994; Zbl 0838.20017)].