Blocks with a Klein Four Defect Group

David A. Craven

May 2008

The following is joint work with Charles Eaton, Radha Kessar and Marcus Linkelmann.
For the purposes of this talk, G is a finite group, K is an algebraically closed field of
characteristic p, where p | |G|, and all modules are finite-dimensional.

1 Facts from Block Theory

KG = B1 ⊕ B2 ⊕ · · · ⊕ Br ,

where the Bi are two-sided ideals that cannot be written as the sum of two smaller two-sided
ideals. These are called blocks.
We also have a 1 in KG. Write

1 = e1 + e2 + · · · + er ,

where the ei are central idempotents such that if ei = e + e0 , where e and e0 are also central
idempotents, then e = 0 or e0 = 0. If we label things correctly, then ei ∈ Bi , and in fact
Bi = KGei . The ei are also called blocks. Some people call the Bi the block ideals and
the ei the blocks, and some people call the Bi the blocks and the ei the block idempotents.
Confusion reigns.
Let M be an indecomposable module. Then M · 1 = M , and so

M = M e1 ⊕ M e2 ⊕ · · · ⊕ M er .

Since M is meant to be indecomposable, there exists a unique i such that M ei = M , and

1
 Let M be an indecomposable module. It turns out that if P denotes a Sylow p-subgroup
of G, then M is a summand of (M ↓P ) ↑G . (The proof of this is the same as Maschke’s
Theorem.) Therefore there are minimal p-subgroups Q such that M is a summand of (M ↓Q
) ↑G . These are called vertices of M .
For example, a module has vertex the trivial subgroup if and only if M is a summand of
KG, the group algebra; i.e., M has trivial vertex if and only if it is projective (a summand
of a free module).

Theorem 1.1 (Green) All vertices of an indecomposable module are p-subgroups of G and
a G-conjugate.

A source of an indecomposable module M is an indecomposable module S for the vertex

Theorem 1.2 (Green) All sources of an indecomposable module M are conjugate by an

There is an invariant of a block, called a defect group, that controls the complexity of the
modules that belong to it. We will define a defect group now, using vertices. Let M denote
the collection of all indecomposable modules belonging to some block B. To each of these
modules one may associate a conjugacy class of p-subgroups P, and thus we get a collection
P of p-subgroups (up to conjugation). This takes on the structure of a poset, using the
obvious ordering. It turns out that this poset has a unique maximal element, and this is
called the defect group of a block.
The defect group is trivial if and only if B is (isomorphic to) a matrix algebra, and
contains a unique indecomposable module, which is a projective simple module. If the
defect group D of B is cyclic, then B contains only finitely many indecomposable modules,
and if D is dihedral, semidihedral or quaternion then, although there are infinitely many
indecomposable modules in B, they can be parameterized. For all other defect groups, the
indecomposable modules belonging to them are too complicated to ‘understand’.

2
2 Blocks with a Klein Four Defect Group
Consider the group algebra KV4 . This has one simple module, which has trivial source. The
projective indecomposable module for this group algebra has the structure

K
K ⊕ K.
K

Consider the group algebra KA4 . This has a single block, and has three simple modules,
K, S1 and S2 , each with trivial source. The projective indecomposable modules for this
group algebra have the structure

K S1 S2
S 1 ⊕ S2 , K ⊕ S2 , S1 ⊕ K .
K S1 S2

Finally, consider the group algebra KA5 . This has two blocks: the principal block (con-
taining the trivial module) and a block of defect zero, containing the 4-dimensional simple
module. The principal block contains three simple modules, K, S1 and S2 , with K having
trivial source and Si having 2-dimensional source. The indecomposable modules for this
group algebra have the structure

K S1 S2
S1 S2 K K
K ⊕ K, S2 , S1 .
S2 S1 K K
K S1 S2

In 1982, Karin Erdmann published a paper in which it was proved that every block with
V4 defect group had either one or three simple modules (originally proved by Brauer) with
the principal indecomposable modules looking as above. These facts can be determined from
the module category, and so Erdmann really proved that there are three types of module
category that can be associated to a block with Klein four defect group.
However, the sources of simple modules cannot be determined using the information in
the module category, the so-called Morita equivalence class of the block. We need something
sharper: this is the notion of a Puig equivalence. We won’t discuss the technical details here,
but rather mention that a Puig equivalence is a Morita equivalence that also remembers
some of the representation theory of the group. For example, two Puig equivalent blocks
have the same sources of the simple modules.

3
Conjecture 2.1 (Erdmann, 1982) Let B be a block with defect group V4 , and suppose
that it is Morita equivalent to a block b from the list {KV4 , KA4 , B0 (KA5 )}. Then the
sources of the simple modules are isomorphic.

In 1996, Linckelmann proved the following result, which gives some more context to the
conjecture.

Theorem 2.2 (Linckelmann, 1996) Erdmann’s conjecture is true if and only if there are
exactly three Puig equivalence types of block with defect group V4 .

To give yet another formulation of this conjecture, we look at the Green correspondence.
Recall that if M is an indecomposable module with vertex Q, then the restriction M ↓NG (Q)
splits up as one (indecomposable) summand with vertex Q and other summands with vertex
contained properly within Q (up to conjugacy). In 1982, Erdmann proved the following.
[Here we need a definition of the functor Ω.]

Theorem 2.3 (Erdmann, 1982) Let B be a block with defect group P , and let H =
NG (P ). Let f denote the Green correspondence from B to the corresponding block b of H.
Let S0 , S1 and S2 denote (where applicable) the one or three simple modules in B.

(i) If B is Morita equivalent to KV4 , then there is an integer i ∈ Z such that Ωi (f (S0 )) is
simple.

(ii) If B is Morita equivalent to KA4 , then there is an integer i ∈ Z such that Ωi (f (Sj )) is
simple for j = 0, 1, 2 (and the simple modules are non-isomorphic).

(iii) If B is Morita equivalent to B0 (KA5 ), then there is an integer i ∈ Z such that Ωi (f (S0 ))
is simple, and S1 and S2 are periodic.

Erdmann’s conjecture is equivalent to the statement that in all parts of the theorem
above, the integer i is 0.
There is another equivalent formulation, which is how I became interested in this con-
jecture: recall that a module is algebraic if it satisfies a polynomial with integer coefficients,
where addition and multiplication are given by direct sum and tensor product.

Proposition 2.4 (C, 2007) Erdmann’s conjecture is true if and only if all simple modules
in blocks with Klein four defect group are algebrai.

Thus Erdmann’s conjecture determines the possible sources of the simple modules, the
Green correspondence in such blocks, and the Puig equivalence classes of block.

4
3 The Proof of Erdmann’s Conjecture
The first step is what makes it possible to attack this conjecture.

Theorem 3.1 (EKL, 2006) Erdmann’s conjecture is true for all groups if and only if it
is true for all groups HP , where H is an odd central extension of a simple group, and P is
isomorphic with a Klein four group, which acts as automorphisms on H/ Z (H).

Thus we reduce the entire conjecture to checking it for each finite quasisimple group,
possibly with automorphisms on top. The idea is that P is the defect group of a block B
with defect group V4 that we are examining. This will be our situation from now on.
One of the main tools in attacking this conjecture is to notice that it is true for real
blocks. To see this, note that a source is only defined up to conjugacy; however, if it is of
the form Ωi (K), as we know it is, then it is unique up to isomorphism. The dual of Ωi (K)
is Ω−i (K); so if a block is self-dual, then a simple module in it is self-dual, and therefore its
source is self-dual. Thus the source is K, and so the conjecture is true for this block.
The first theorem is easy.

Theorem 3.2 (E (2006), and C (2007)) Suppose that H is sporadic or alternating. Then
B is real, and hence Erdmann’s conjecture is true.

The next theorem is also easy, but it requires some knowledge of the groups involved.

Theorem 3.3 (C, 2008) Suppose that H is symplectic, orthogonal or of type 3 D4 (q). Then
B is real, and hence Erdmann’s conjecture is true.

To prove this, we actually show that all 2-blocks of the quasisimple group H are real.
Charles Eaton and I sorted out the exceptional groups, with two exceptions (!).

Theorem 3.4 (CE, 2008) Let H be an exceptional group, except for E6 (q) or E7 (q) when
P ∩ H = 1. Then Erdmann’s conjecture is true.

Radha Kessar is responsible for the next two theorems, which finish off all finite qua-
sisimple groups.

Theorem 3.5 (K, 2007/8) Let H be linear or unitary. Then Erdmann’s conjecture is
true.

Theorem 3.6 (K, 2008) Let H be of type E6 (q) or E7 (q), where P ∩ H = 1. Then
Erdmann’s conjecture is true.

5
 This last result is a nice application of Deligne–Lusztig theory, and so we will briefly
describe some of the highlights, in the case where H is of type E6 (q).

• There is a unique V4 subgroup of Out(H/ Z (H)) up to conjugation, and so we may

• Since the graph automorphism inverts the centre of 3.E6 (q), and P is meant to cen-
tralize Z (H), we see that we may assume that Z (H) = 1.

• The adjoint group is actually E6 (q).3, and we can take this group instead. (This
makes the Deligne–Lusztig theory easier.) The dual group is the simply connected
group L = 3.E6 (q).

• By a theorem of Feit and Zuckerman, all semisimple elements of L are real. (Remember
that the semisimple elements of L label the Lusztig series.)

• Let B be a block of defect zero in KH, and let M denote the unique simple module.
Then M comes from a Lusztig series with (real) semisimple label s.

• Since M has trivial vertex, the centralizer of s is a torus. Hence the Lusztig series
contains a single element, namely the module M .

• Finally, as s is real, the dual of M also lies in the series defined on s, and so M is
self-dual, as required.

4 After the Proof

• The Feit conjecture for V4 vertex.

• Blocks with quaternion group Q8 .

• Studying dihedral blocks.

• Blocks with defect group C3 × C3 .

• 2-blocks with abelian defect group.