On the number of cells of a cellular algebra (Q4700482)

Quasi-hereditary algebras, which were introduced by Cline, Parshall and Scott, are associative algebras defined in terms of a heredity chain. A heredity chain can always be refined to one which is maximal, and the length of a maximal chain is an invariant of the algebra. The authors are interested in the analogue of this question for the class of cellular algebras introduced by Graham and Lehrer; this is a natural question because there is a large overlap between the class of cellular algebras and the class of quasi-hereditary algebras. As the authors have shown in their previous work, there is a fruitful base-free treatment of cellular algebras via ``cell chains''. This paper looks at the question of whether a cell chain is maximal and whether the length of a maximal cell chain is an invariant of the algebra. It turns out that each cell chain is indeed maximal, but the length of a maximal cell chain is not an invariant of the algebra. A rather surprising counterexample is given: it is possible for an associative algebra to have two different cellular structures with respect to the same involution and yielding the same cell basis but for which the cell chains have different lengths.