Radicals and bands of associative algebras (Q916755)

An algebra means an F-algebra, where F is a fixed commutative associative ring. A band is an idempotent semigroup, a semilattice is a commutative band. Let I be a band, and let r be an algebra radical (in the sense of Kurosh); r commutes with I if and only if \(r(R)=\sum r(R_ i)\) for any algebra \(R=\sum R_ i\) (i\(\in I)\) graded by I. The problem discussed in the paper is the following one: given a class D of bands, characterize radicals commuting with all \(I\in D\). This problem is completely solved. To formulate the author's results we need some notions. Let R be an algebra; then \(R^ 0=(R,+,\circ)\) is an algebra with the same F-module structure as that of R, and with the zero multiplication \(\circ\). A radical r is said to be: (i) an A-radical if and only if \((r(R))^ 0=r(R^ 0)\), (ii) a strict radical if and only if any r-semisimple algebra contains no r-radical nonzero subalgebra, (iii) a weakly hereditary radical if and only if, for any algebra R and for any ideal I from R, one has \(r(R)\supseteq I.r(R)+r(R).I\) (clearly, every hereditary radical is weakly hereditary). Main results: if D contains a noncommutative band then a radical r commutes with all \(I\in D\) if and only if r is an A-radical (Theorem 1); if D is a class of semilattices then a radical r commutes with all \(I\in D\) if and only if r is strict and weakly hereditary (Theorem 2). The paper contains also an example which shows that a radical commuting with all semilattices need not to be neither strict hereditary nor an A- radical, a negative answer to a conjecture of \textit{B. J. Gardner} [Pac. J. Math. 58, 387-392 (1975; Zbl 0281.16007)].