Let G be a finite p-group and let dl(G) be its derived
length, i.e., the length of the shortest possible normal series of
G such that all factors of consecutive members of the series are
abelian. dl(G) is a measure for the "height" of the group; the
larger it is, the more complex is the group. Therefore it is an
important problem to study by which parameters the derived length
can be controlled. One of the parameters stems from the
representation theory of finite groups and is the number of the
irreducible character degrees of G. These character degrees are
the dimensions of the irreducible representations of G over the
complex numbers. Denote by cd(G) the set of all the irreducible
character degrees of G. By an old result of Taketa it is known
that
dl(G) is bounded above by the cardinality of cd(G),
which is the best general bound to date. However, one conjectures
that the true bound in fact is logarithmic in the cardinality of cd(G).
While this has been proven for some families of groups (e.g. Sylow
subgroups of some classical groups), there is not a more general
class of groups for which such a bound is known, and a general
proof of the conjecture seems to be far out of the reach of
today's techniques.
The proposed project aims at investigating a special class
C of groups for which there is some hope of establishing the
desired bound. Due to the enormous complexity of p-groups in
general, such a class C has to be chosen very carefully
to be tractable at all. For this project Keller suggests looking
at the class C of normally monomial p-groups of
maximal class. "Normally monomial" is a representation theoretic
restriction while "maximal class" is a group theoretic condition.
While the difficulty of the problem even for groups in C is still considerable, by the choice of C it becomes accessible to undergraduate students mainly for the following two reasons:
Moreover such examples should give clues on how to deal with the general question of what the right bound for dl(G) in terms of |cd(G)| for groups in C is and how to prove it. Thus the analysis of the examples with respect to patterns and structural behavior would be the third part of the project. The hope is to obtain best possible bounds for small values of dl(G) and to figure out and prove a general better than linear asymptotic bound (possibly logarithmic).