, ,

Recall that the

p

-rank of a finite group

G

r k_{p} (G)

, is the largest rank of an elementary abelian

p

-subgroup of

G

and that the rank of a finite group

G

r k (G)

, is the maximum of

r k_{p} (G)

taken over all primes

p

. In addition we can define the homotopy rank of a finite group

G

h (G)

, to be the minimal integer

k

such that

G

acts freely on a finite CW-complex

Y ≃ S^{n_{1}} \times S^{n_{2}} \times \dots \times S^{n_{k}}

. Benson and Carlson [7] state a conjecture that for any finite group

G

r k (G) = h (G)

. The case of this conjecture when

G

is a rank one group is a direct result of Swan's theorem [24] . Benson and Carlson's conjecture has also been verified by Adem and Smith [1, 2] in the case when

G

is a rank two

p

-group. In addition, recall that Heller [15] has shown that for a finite group

G

, if

h (G) = 2

, then

r k (G) = 2

. In this paper we will verify this same conjecture for all rank two groups that do not contain a particular type of subgroup.

Before we state our main theorem, we define the subgroups that we exclude: for each odd prime

p

, let

T_{p}

be the semidirect product

(Z_{p} \times Z_{p}) ∖_{θ} S L_{2} (F_{p})

where the action

θ

is given by the obvious inclusion

S L_{2} (F_{p}) \to G L_{2} (F_{p}) \sim = A u t (Z_{p} \times Z_{p})

We can now state our main theorem.

Theorem 1. Let

G

be a finite group such that

r k (G) = 2

G

acts freely on a finite CW-complex

Y ≃ S^{n_{1}} \times S^{n_{2}}

unless for some odd prime p,

G

contains a subgroup

H

such that

H / O_{p'} (H) \sim = T_{p}

The following corollary, which can also be shown using more direct methods, follows immediately from Theorem 1 :

Corollary 2. If

G

is a finite group of odd order with

r k (G) = 2

, then

G

acts on a finite CW-complex

Y ≃ S^{n_{1}} \times S^{n_{2}}

I am grateful to the referee and to Antonio Viruel and Antonio Diaz for helpful comments in fixing Section 4 .

2 Preliminaries

Recall that Adem and Smith have shown that if

G

is a rank two

p

-group then

h (G) = 2

[1, 2] . In their proof they use the following theorem:

Theorem 3 (Adem and Smith [2] ). Let

G

be a finite group and let

X

be a finitely dominated, simply connected

G

-CW complex such that every isotropy subgroup has rank one. Then for some integer

N > 0

there exists a finite CW-complex

Y ≃ S^{N} \times X

and a free action of

G

Y

such that the projection

Y \to X

G

-equivariant.

In the present work, we apply Theorem 3 for a rank two group by finding an action of the group on a CW-complex

X ≃ S^{n}

such that the isotropy subgroups are of rank one. Adem and Smith in [2] also developed a sufficient condition for this type of action to exist.

To explore this sufficient condition, we start by letting

G

be a finite group. Given a

G

-CW complex

X ≃ S^{N - 1}

, there is a fibration

X \to X \times_{G} E G \to B G .

This fibration gives an Euler class

β

, which is a cohomology class in

H^{N} (G)

. This cohomology class is the transgression in the Serre spectral sequence of the fundamental class of the fiber.

Regarding this class Adem and Smith made the following definition:

Definition 4 (Adem and Smith [2] ). Let

G

be a finite group. The Euler class

β

of a

G

-CW complex

X ≃ S^{N - 1}

is called effective if the Krull dimension of

H^{*} (X \times_{G} E G, F_{p})

is less than

r k (G)

for all

p | | G |

Adem and Smith have proven the following lemma characterizing effective Euler classes:

Lemma 5 (Adem and Smith [2] ). Let

β \in H^{N} (G)

be the Euler class of an action on a finite dimensional

X ≃ S^{N - 1}

β

is effective if and only if every maximal rank elementary abelian subgroup of

G

acts without stationary points.

Remark 6. By applying Lemma 5 and Theorem 3 , we see that any finite group of rank two having an effective Euler class of an action on a finite dimensional

X ≃ S^{N - 1}

acts freely on a finite CW-complex

Y ≃ S^{m} \times S^{N - 1}

Recall that for a group

G

B G

is the classifying space of

G

. Also

[B G, B U (n)]

represents the homotopy classes of maps from

B G

B U (n)

, which may also be written as

π_{0} H o m (B G, B U (n))

(see [12] ). In addition, recall that

[B G, B U (n)]

is isomorphic to the set of isomorphism classes of rank

n

complex vector bundles over

B G

. In this discussion, complex vector bundles over

B G

will be thought of as elements of

[B G, B U (n)]

Recall that the Euler class of the universal bundle over

B U (n)

is the top Chern class, which lies in

H^{2 n} (B U (n))

. We will call this class

ξ

. A map

φ : B G \to B U (n)

induces a map

φ^{*} : H^{2 n} (B U (n)) \to H^{2 n} (B G) = H^{2 n} (G)

. Thus the image of

ξ

under

φ^{*}

is the Euler class of the vector bundle over

B G

that corresponds to

φ

. Let

β_{φ} = φ^{*} (ξ)

be this class, which is the Euler class of a

G

-CW complex that is the unit sphere bundle of the vector bundle. This

G

-CW complex is the fiber as discussed at the outset. Recall that if

φ

and

μ

are homotopic maps from

B G

B U (n)

, then

β_{φ} = β_{μ}

; therefore, the Euler class of a homotopy class of maps is well defined.

Recall that Dwyer and Zabrodsky [12] have shown that for any

p

-subgroup

P

C h a r_{n} (P) \sim = R e p (P, U (n)) \overset{\sim =}{\to} [B P, B U (n)]

where

R e p (P, U (n)) = H o m (P, U (n)) / I n n (U (n))

and the map is

ρ \mapsto B ρ

. So if

φ

is a map

B G \to B U (n)

and if

P

is a

p

-subgroup of

G

for some prime

p

φ |_{B P}

is induced by a unitary representation of

P

. The following proposition gives the characterization of effectiveness for the Euler class

β_{φ}

Proposition 7 ([1,section4] ). An Euler class

β_{φ}

is effective if and only if for each elementary abelian subgroup

E

G

with

r k (E) = r k (G)

, there is a unitary representation

λ : E \to U (n)

such that

φ_{E} = B λ

and that

λ

does not have the trivial representation as one of its irreducible constituents.

As we proceed we will want to work separately with each prime. Definition 8 will be useful in this endeavor.

Definition 8. Let

G

be a finite group. The Euler class

β_{φ}

is called p-effective if for each elementary abelian

p

-subgroup

E

G

with

r k (E) = r k (G)

, there is a unitary representation

λ : E \to U (n)

such that

φ_{E} = B λ

and that

λ

does not have the trivial representation as one of its irreducible constituents.

It is clear then that

β_{φ}

is effective if and only if it is

p

-effective for each

p | | G |

. Also if

r k_{p} (G) < r k (G)

, then any

β_{φ}

p

-effective since it trivially satisfies the above definition; therefore, we can restate the above proposition:

β_{φ}

is effective if and only if it is

p

-effective for each prime

p

such that

r k_{p} (G) = r k (G)

Recall that

G

is a finite group,

p

is a prime dividing the order of

G

, and

G_{p}

is a Sylow

p

-subgroup of

G

. In addition we will use

C h a r_{n} (G_{p})

to represent the set of degree

n

complex characters of

G_{p}

and will let

C h a r_{n}^{G} (G_{p})

represent the subset consisting of those degree

n

complex characters of

G_{p}

that are the restrictions of class functions of

G

Definition 9 concerns the characters of subgroups of

G

Definition 9. Let

H \subseteq G

be a subgroup. Let

χ

be a character of

H

. We say that

χ

respects fusion in

G

if for each pair of elements

h

k \in H

such that

\exists g \in G

with

h = g k g^{- 1}

χ (h) = χ (k)

Remark 10. Let

H \subseteq G

. A sufficient condition for a character of

H

to respect fusion in

G

is for the character to be constant on elements of the same order.

Notice that for

χ \in C h a r_{n} (G_{p})

χ

respects fusion in

G

if and only if

χ \in C h a r_{n}^{G} (G_{p})

. In a previous paper [20] , the author has shown the following theorem:

Theorem 11 (Jackson [20,Theorem1.3] ). If

G

is a finite group that does not contain a rank three elementary abelian subgroup, then the natural mapping

ψ_{G} : [B G, B U (n)] \to \prod_{p | | G |} C h a r_{n}^{G} (G_{p})

is a surjection.

Combining Theorem 11 and Definition 8 , we get the following theorem.

Theorem 12. Let

G

be a finite group,

p

a prime dividing

| G |

, and

G_{p}

a Sylow

p

-subgroup of

G

. If there is a character

χ

G_{p}

that respects fusion in

G

and has the property that

[χ |_{E}, 1_{E}] = 0

for each

E \subseteq G_{p}

elementary abelian with

r k (E) = r k (G)

, then there is a complex vector bundle over

B G

whose Euler class is

p

-effective.

Remark 13. Let

p

be a prime,

G_{p}

p

-group of rank

n

, and

χ

a character of

G_{p}

A sufficient condition to guarantee that

[χ |_{E}, 1_{E}] = 0

for each

E \subseteq G_{p}

elementary abelian with

r k (E) = r k (G_{p})

, is that

χ (1) = c (p^{n} - 1)

and

χ (g) = - c

for each element of

G_{p}

of order

p

. In this case,

χ |_{E} = c (ρ - 1)

where

ρ

is the regular character of the group

E

Theorem 12 leads to definition 14 :

Definition 14. Let

G

be a finite group,

p

a prime dividing

| G |

, and

G_{p}

a Sylow

p

-subgroup of

G

. A character

χ

G_{p}

is called a

p

-effective character of

G

under two conditions:

χ

respects fusion in

G

, and for each

E \subseteq G_{p}

elementary abelian with

r k (E) = r k (G)

[χ |_{E}, 1_{E}] = 0

Remark 15. Let

G

be a finite group,

p

a prime dividing

| G |

, and

G_{p}

a Sylow

p

-subgroup of

G

. A character

χ

G_{p}

p

-effective if

χ

is both constant on elements of the same order and

[χ |_{E}, 1_{E}] = 0

for each

E \subseteq G_{p}

elementary abelian with

r k (E) = r k (G)

. (This follows from Remark 10 .)

Applying Definition 14 , Theorem 12 can be restated as follows: If

G

has a

p

-effective character, then there is a complex vector bundle over

B G

whose Euler class is

p

-effective.

We have now proved Theorem 16 , which is a modified form of a theorem by Adem and Smith.

Theorem 16 (See [2,Theorem7.2] ). Let

G

be a finite group with

r k (G) = 2

. If for each prime

p

dividing the order of

G

there exists a

p

-effective character of

G

, then

G

acts freely on a finite CW-complex

Y ≃ S^{N_{1}} \times S^{N_{2}}

Proof. For each prime $p$ dividing the order of $G$ , let $χ_{p}$ be a $p$ -effective character of $G$ . There exists an integer $n$ such that $χ_{p} (1)$ divides $n$ for each $p$ dividing the order of $G$ . $(\frac{n}{χ_{p} (1)} \cdot χ_{p})_{p | | G |} \in \prod_{p | | G |} C h a r_{n}^{G} (G_{p})$ . By Theorem 11 , there exists an element $φ \in [B G, B U (n)]$ such that $φ_{G} (φ) = (\frac{n}{χ_{p} (1)} \cdot χ_{p})_{p | | G |}$ . The Euler class of the homotopy class $φ$ is then an effective Euler class of $G$ . Theorem 16 then follows from Remark 6 . □

Theorem 16 reduces the problem of showing that a rank two finite group

G

acts freely on a finite CW-complex

Y ≃ S^{N_{1}} \times S^{N_{2}}

to the problem of demonstrating that for each prime

p

there is a

p

-effective character of

G

. It is important to state that not all rank two finite groups have a

p

-effective character for each prime

p

. An example of a finite group that does not contain a

p

-effective character for a particular prime is given in the following corollary, which will be proven in Section 6 .

Corollary 17. Let

p

be an odd prime. The group

P S L_{3} (F_{p})

does not have a

p

-effective character.

We notice that the groups

P S L_{3} (F_{p})

, for odd primes

p

, are the only finite simple groups of rank 2 that were not shown by Adem and Smith to act freely on a finite CW-complex

Y ≃ S^{N_{1}} \times S^{N_{2}}

. [1, 2] .

3 Sufficient conditions for $p$ -effective characters

In Section 3 we will show the existence of the most important sufficient condition for the existence of a

p

-effective character. We start by recalling the following definition:

Definition 18. Let

G

be a finite group and

H

and

K

subgroups such that

H \subset K

. We say that

H

is strongly closed in

K

with respect to

G

if for each

g \in G

g H g^{- 1} \cap K \subseteq H

Remark 19. Let

G

be a finite group and

H

and

K

subgroups such that

H \subset K

We say that

H

is weakly closed in

K

with respect to

G

if for each

g \in G

with

g H g^{- 1} \subseteq K

g H g^{- 1} H

. For a subgroup

H

of prime order,

H

is strongly closed in

K

with respect to

G

if and only

H

weakly closed in

K

with respect to

G

Lemma 20. If

E

is an elementary abelian subgroup of a

p

-group

P

, then the product

A Ω_{1} (Z (P))

is an elementary abelian subgroup of

P

. In particular, if

E

is a maximal elementary abelian subgroup of

P

under inclusion, then

Ω_{1} (Z (P)) \subseteq A

Proof. Let $Z = Ω_{1} (Z (P))$ . Since $Z$ is central, the product $A Ω_{1} (Z (P))$ is abelian. The product is generated by elements of order $p$ and so must be elementary abelian. □

Corollary 21. Let

P

be a

p

-group with

r k (P) = r k (Z (P))

Ω_{1} (Z (P))

is the unique elementary abelian subgroup of

P

that is maximal under inclusion. In addition,

Ω_{1} (P) = Ω_{1} (Z (P))

, and

Ω_{1} (Z (P))

is strongly closed in

P

with respect to

G

Proof. If $E$ is a maximal elementary abelian subgroup of $P$ under inclusion, then $Ω_{1} (Z (P)) \subseteq E$ by Lemma 20 . Equality holds since $r k (E) \leq r k (P) = r k (Ω_{1} (Z (P)))$ . □

Again we let

p

be a prime dividing the order of a finite group

G

and let

G_{p} \in S y l_{p} (G)

. We show that if there is a subgroup of

Z (G_{p})

that is strongly closed in

G_{p}

with respect to

G

, then

G

has a

p

-effective character in the following proposition:

Proposition 22. Let

G

be a finite group,

n = r k (G)

p

a prime divisor of

| G |

with

r k_{p} (G) = n

, and

G_{p}

a Sylow

p

-subgroup of

G

. If there exists

H \subseteq Z (G_{p})

such that

H

is non-empty and strongly closed in

G_{p}

with respect to

G

, then

G

has a

p

-effective character.

Before we prove Proposition 22 , we state the following standard lemma, which we will need in the proof of Proposition 22 .

Lemma 23. Let

G

be a finite group and

H \subseteq G

a subgroup. If

χ

is a character of

G

such that

[χ, 1_{G}] > 0

, then

[χ |_{H}, 1_{H}] > 0

Proof. Let $χ$ be a character of $G$ such that $[χ, 1_{G}] > 0$ . Let $1_{G}, λ_{2}, . . ., λ_{k}$ be the irreducible characters for $G$ . There exist $a_{i} \geq 0$ , for each $i = 1, 2, . . ., k$ , with $a_{1} = [χ, 1_{G}] > 0$ suchthat $χ = a_{1} 1_{G} + \sum_{i = 2}^{k} a_{i} λ_{i}$ ; therefore, $χ |_{H} = a_{1} (1_{G}) |_{H} + \sum_{i = 2}^{k} a_{i} λ_{i} |_{H}$ . $(1_{G}) |_{H} = 1_{H}$ , so $[χ |_{H}, 1_{H}] = a_{1} + \sum_{i = 2}^{k} a_{i} [λ_{i} |_{H}, 1_{H}]$ . Lemma 23 holds since $a_{i} \geq 0$ and $a_{1} > 0$ . □

We will now prove Proposition 22 .

Proof. Let $H \subseteq Z (G_{p})$ such that $H$ is non-empty and strongly closed in $G_{p}$ with respect to $G$ . Let $K = Ω_{1} (H)$ (the subgroup of $H$ generated by the elements of $H$ which are of order $p$ ). So there is an $i \leq n$ with $K \sim = (Z / p Z)^{i}$ . Let $χ = \sum_{ψ \in I r r (K) \ {1_{K}}} ψ$ ; therefore, $[χ, 1_{K}] = 0$ .
Let $φ = I n d_{K}^{G_{p}} χ$ .
Notice that for each $y \in K \ {e}$ , $χ (y) = - 1$ [13,4.2.7.ii] . In addition since $K \subseteq Z (G_{p})$ , for each $y \in K \ {e}$ , $φ (y) = - [G_{p} : K]$ [13,4.4.3.i] . Since $H$ is strongly closed in $G_{p}$ with respect to $G$ , $K$ is strongly closed in $G_{p}$ with respect to $G$ . $φ$ respects fusion in $G$ ; therefore, for each $z \in G_{p} \ K$ , $φ (z) = 0$ .
Let $E$ be a subgroup of $G_{p}$ such that $E \sim = (Z / p Z)^{n}$ . By Lemma 20 , $K \subseteq E$ since $K \subseteq Ω_{1} (Z (G_{p}))$ . Now suppose $\exists E \subseteq G_{p}$ such that $E \sim = (Z / p Z)^{n}$ and $[φ |_{E}, 1_{E}] > 0$ .
$K \subseteq Ω_{1} (Z (G_{p}))$ implies that $K \subseteq E$ ; therefore, we may apply the previous Lemma 23 to see that $[(φ |_{E}) |_{K}, 1_{K}] > 0$ . Note that $(φ |_{E}) |_{K} = φ |_{K}$ and that $φ |_{K} = [G_{p} : K] χ$ . Then $[φ |_{K}, 1_{K}] = [G_{p} : K] [χ, 1_{K}] = 0,$ which is a contradiction; thus, if $E$ is an elementary abelian $p$ -subgroup of $G_{p}$ of rank $n$ , $[φ |_{E}, 1_{E}] = 0$ . $φ$ is therefore a $p$ -effective character of $G$ . □

We also prove a Corollary to Proposition 22 .

Corollary 24. Let

G

be a finite group,

p

a prime divisor of

| G |

with

r k (G) = r k_{p} (G) = n

, and

G_{p}

a Sylow

p

-subgroup of

G

. If

r k (Z (G_{p})) = n

, then

G

has a

p

-effective character.

Proof. By Corollary 21 , $Ω_{1} (Z (G_{p}))$ is strongly closed in $G_{p}$ with respect to $G$ . Corollary 24 then follows from Proposition 22 . □

In Section 4 and 5 , we will look at the possible types of rank two groups where this sufficient condition hold. In Section 4 we will look at the case where

p

is an odd prime, while in Section 5 we will examine the case where

p = 2

4 Odd primes

When

p

is an odd prime and

r k_{p} (G) = 2

, we show that if the sufficiency condition of Proposition 22 does not hold, then

G

must contain a specific type of subgroup. Let

p > 2

be a prime and let

T_{p}

be the semidirect product

(Z /_{p} \times Z /_{p}) ∖_{θ} S L_{2} (F_{p})

where the action

θ

is given by the obvious inclusion

S L_{2} (F_{p}) \to G L_{2} (F_{p}) \sim = A u t (Z /_{p} \times Z /_{p})

Before we list the next theorem, we need two definitions and a lemma.

Definition 25. Let

P \subseteq G

be a

p

-subgroup.

P

is said to be p-centric if

Z (P)

is a Sylow

p

-subgroup of

C_{G} (P) = Z (P) \times C_{G}^{'} (P)

where

C_{G}^{'} (P)

is a

p'

-subgroup.

Definition 26 (see [14] ). Let

P \subseteq G

be a

p

-centric subgroup.

P

is said to be principal p-radical if

O_{p} (N_{G} (P) / P C_{G} (P)) = {1}

Lemma 27 (Diaz, Ruiz, Viruel [10,Section3] ). Let

G

be a finite group and

p > 2

a prime with

r k_{p} (G) = 2

. Let

G_{p} \in S y l_{p} (G)

and

P

be a principal

p

-radical subgroup of

G

. If

P \neq G_{p}

and

P

is metacyclic, then

P \sim = Z /_{p^{n}} \times Z /_{p^{n}}

with either

p = 3

n = 1

The lemma above follows from Corollaries 3.7 and 3.8 as well as Proposition 3.12 of [10] .

The argument in [10] is written in the language of fusion systems, but can be applied to finite groups giving Lemma 27

Lemma 28 (Ruiz, Viruel [18,Lemma4.1] (see also [10] )). Let

G

be a subgroup of

{GL}_{2} (p)

. If

O_{p} (G) = {1}

and

p

divides the order of

G

, then

{SL}_{2} (p) \subseteq G

Theorem 29. Let

G

be a finite group and

p > 2

a prime with

r k_{p} (G) = 2

. Let

G_{p} \in S y l_{p} (G)

. If

Ω_{1} (Z (G_{p}))

is not strongly closed in

G_{p}

with respect to

G

, then

G

has a subgroup

H

such that

H / O_{p^{'}} (H) \sim = T_{p}

Proof. Notice that $r k (Z (G_{p})) = 1$ by Corollary 21 . Let $Z = Ω_{1} (Z (G_{p}))$ . By hypothesis, $Z$ is not strongly closed in $G_{p}$ with respect to $G$ . Recall that Alperin's Fusion Theory [3] states that given a weak conjugation family $ℱ$ , there exists an $(S, T) \in ℱ$ with $Z \subseteq S \subseteq G_{p}$ and with $Z$ not strongly closed in $S$ with respect to $T$ . The author has shown in a previous paper [20] that the collection of all pairs $(P, N_{G} (P))$ , where $P \subseteq G_{p}$ is a principal $p$ -radical subgroup of $G$ , is a weak conjugation family. So there exists $P \subseteq G_{p}$ that is a principal $p$ -radical subgroup of $G$ with $Z \subseteq P$ and with $Z$ not strongly closed in $P$ with respect to $N_{G} (P)$ . Notice that $Z \neq P \neq G_{p}$ and that $r k_{p} (Z (P)) = r k_{p} (P) = 2$ (in particular $p | [N_{G} (P) : P])$ . Using the classification of rank two $p$ -groups for odd primes $p$ by Blackburn [8] (see also [10, 11] ), we notice that $P$ must be metacyclic (since $r k_{p} (Z (P)) = 2$ ). By Lemma 27 , $P \sim = Z /_{p^{n}} \times Z /_{p^{n}}$ with either $p = 3$ or $n = 1$ . We may consider the group of outer automorphisms $O u t (Z /_{p^{n}} \times Z /_{p^{n}})$ as two by two matrices with coefficients in $Z /_{p^{n}}$ and with $p$ not dividing the determinant.
Reduction modulo $p$ gives the following short exact sequence of groups: $1 \to Q \to O u t (Z /_{p^{n}} \times Z /_{p^{n}}) \overset{ρ}{\to} {GL}_{2} (p) \to 1$ with $Q$ a $p$ -group. Notice that $N_{G} (P) / (P \times C_{G}^{'} (P)) \subseteq O u t (Z /_{p^{n}} \times Z /_{p^{n}})$ . Since $P$ is a principal $p$ -radical subgroup of $G$ , $Q \cap N_{G} (P) / (P \times C_{G}^{'} (P)) = {1}$ . And so $ρ$ is an injection when restricted to $N_{G} (P) / (P \times C_{G}^{'} (P))$ . Notice that $ρ (N_{G} (P) / (P \times C_{G}^{'} (P)))$ has order divisible by $p$ and $0_{p} (ρ (N_{G} (P) / (P \times C_{G}^{'} (P)))) = {1}$ ; therefore, by Lemma 28 , ${SL}_{2} (p) \subseteq ρ (N_{G} (P) / (P \times C_{G}^{'} (P)))$ . So we see that there exists $H \subseteq N_{G} (P)$ with $P \subset H$ . Clearly $H / (C_{G}^{'} (P)) \sim = T_{p}$ . □

We point out here some connections between the results of Theorem 29 and previous work by Adem and Smith. Notice that Theorem 29 implies that any rank 2 group

G

that does not contain a subgroup isomorphic to

T_{p}

has a

p

-effective character. A

p

-group

P

is called a Swan group if for any group

G

containing

P

as a Sylow

p

-subgroup, the mod-p cohomology ring

H^{*} (G)

is equal to

H^{*} (N_{G} (P)) \sim = H^{*} (P)^{N_{G} (p)}

. A group with a Sylow

p

-subgroup that is a Swan group has been shown to have a

p

-effective character by Adem and Smith [2,Section6] . Dietz and Glauberman (see [22] ) have shown that any odd order metacyclic group is a swan group. In addition, Diaz, Ruiz and Viruel [10] have shown that if

P

is a rank 2

p

-group for

p > 3

P

is Swan unless it is a Sylow

p

-subgroup of

T_{p}

. They also demonstrate that some rank two 3-groups of maximal class are Swan.

5 Prime two

The argument in this section will be done in two parts. First we will show that for the prime two, if the sufficiency condition of Proposition 22 does not hold for

G

and for prime

2

, then

G_{2} \in S y l_{2} (G)

is either dihedral, semi-dihedral, or wreathed. In the second part we will deal with these exeptional cases. We start with two definitions:

Definition 30 ([13,p.191] ). Recall that a 2-group is semi-dihedral (sometimes called quasi-dihedral) if it is generated by two elements

x

and

y

subject to the relations that

y^{2} = x^{2^{n}} = 1

and

y x y^{- 1} = x^{- 1 + 2^{n - 1}}

for some

n \geq 3

Definition 31 ([13,p.486] ). A 2-group is called wreathed if it is generated by three elements

x

y

, and

z

subject to the relations that

x^{2^{n}} = y^{2^{n}} = z^{2} = 1

x y = y x

, and

z x z^{- 1} = y

with

n \geq 2

Notice that the dihedral group of order 8 is wreathed. Now that we have these definitions we will state Proposition 32 , which is crucial in our discussion of rank two 2-subgroups. The proof of Proposition 32 follows for the proof of a theorem by Alperin, Brauer, and Gorenstein [5,Proposition7.1] .

Proposition 32 (See Proposition 7.1 of [5] and its proof.). Let

G

be a finite group and

G_{2} \in S y l_{2} (G)

. Suppose that

r k (G_{2}) = 2

. If

Ω_{1} (Z (G_{2}))

is not strongly closed in

G_{2}

with respect to

G

, then

G_{2}

is either dihedral, semi-dihedral, or wreathed.

Since this proof follows from the proof of Proposition 7.1 of [5] , we will only state a brief outline, including the changes needed to prove our Proposition 32 .

Again notice that

r k (Z (G_{2})) = 1

by Corollary 21 . If

G_{2}

does not contain a normal rank 2 elementary abelian subgroup, then

G_{2}

must be either dihedral or semi-dihedral (see [13,Theorem5.4.10] ). We may assume that

G_{2}

has a normal rank 2 elementary abelian subgroup, which we will call

V

. Letting

T = C_{G_{2}} (V)

, we notice that

[G_{2} : T] \leq 2

. Also

G_{2} \neq T

since

r k (Z (G_{2})) = 1

; therefore,

[G_{2} : T] = 2

Letting

A_{G} (V) = N_{G} (V) / C_{G} (V)

, we notice that

A_{G} (V)

is isomorphic to a subgroup of

A u t (V) \sim = Σ_{3}

and that

2 | | A_{G} (V) |

. Thus either

A_{G} (V) \sim = Σ_{3}

2 = | A_{G} (V) |

From the argument in [5] we get the following results: if

A_{G} (V) \sim = Σ_{n}

, then

G_{2}

is wreathed (including the case of

D_{8}

); and if

| A_{G} (V) | = 2

, then

G_{2} \sim = D_{8}

. This finishes the discussion of Proposition 32 .

We have seen that if

G

is a rank two finite group, it has a 2-effective character unless a Sylow 2-subgroup is dihedral, semi-dihedral, or wreathed. In the rest of this section we will show that in each of these three cases

G

also has a 2-effective character, which will complete the proof that a rank two finite group has a 2-effective character. The author proved the following two lemmas in his doctoral thesis [21] ; the proofs are included here for completeness.

We will start with a lemma that will be used in the proof of Proposition 36 .

Lemma 33. Let

G

be a finite group, let

n = r k (G)

, let

p

a prime divisor of

| G |

with

r k_{p} (G) = n

, and let

G_{p} \in S y l_{p} (G)

. If

G_{p}

is abelian,

G_{p} ⊲ G

, or

G

p

-nilpotent, then

G

has a

p

-effective character.

Proof. In each case we will show that $G$ has a $p$ -effective character by showing that $Z (G_{p})$ is strongly closed in $G_{p}$ with respect to $G$ using Proposition 22 . If $G_{p}$ is abelian, $Z (G_{p}) = G_{p}$ is obviously strongly closed in $G_{p}$ with respect to $G$ (see Corollary 21 ). In the case where $G_{p} ⊲ G$ , $Z (G_{p}) ⊲ G$ because $Z (G_{p})$ is a characteristic subgroup of $G_{p}$ ; therefore, $Z (G_{p})$ is strongly closed in $G_{p}$ with respect to $G$ . If $G$ is $p$ -nilpotent then two elements of $G_{p}$ are conjugate in $G$ if and only if they are conjugate in $G_{p}$ . Thus, in this case also, $Z (G_{p})$ is strongly closed in $G_{p}$ with respect to $G$ . □

Lemma 34. If

P

is a dihedral or semi-dihedral 2-group such that

| P | = 2^{n}

with

n \geq 3

, then there is a character

χ

P

such that

χ (g) = {\begin{matrix} 3 * 2^{n - 3} & if g = 1 \\ - 2^{n - 3} & if g is an involution \\ 2^{n - 3} & otherwise. \end{matrix}

Proof. Let $N$ be the commutator subgroup $[P, P]$ and notice that $P / N \sim = Z / 2 \times Z / 2$ . There is a non-trivial irreducible character $λ$ of $P / N$ with $λ (g N) = 1$ for $g \in Z (P)$ (recall $Z (P) \sim = Z / 2$ ). If $g \in P \ Z (P)$ is an involution, then $λ (g N) = - 1$ . Let $ρ (g) = λ (g N)$ , which is a character of $P$ .
We will define another character of $P$ by first inductively defining a character of each $D_{2^{m}}$ .
We start by letting $ψ_{3}$ be the character of $D_{8}$ such that
$ψ_{3} (g) = {\begin{matrix} 2 & if g = 1 \\ - 2 & if g \in Z (D_{8}) \ {1} \\ 0 & otherwise. \end{matrix}$
For the induction, let $ψ_{k} = I n d_{D_{2^{k - 1}}}^{D_{2^{k}}} ψ_{k - 1}$ . $ψ_{k}$ is a character of $D_{2^{k}}$ for each $k \geq 3$ . We see that if $P$ is dihedral, then $χ = ρ + ψ_{n}$ , and Lemma 34 holds for dihedral 2-groups. On the other hand, if $P$ is semidihedral, $χ = ρ + I n d_{D_{2^{n - 1}}}^{P} ψ_{n - 1}$ . □

Lemma 35. Let

G

be a finite group with

G_{2} \in S y l_{2} (G)

a wreathed 2-group.

r k (G) = 2

and

G

does not have a normal subgroup of index 2, then

G

has a 2-effective character.

Proof. Let $G_{2}$ be generated by $x$ , $y$ , and $z$ as in Definition 31 . Let $n$ be the positive integer such that $| G_{2} | = 2^{2 n + 1}$ . Recall that $n \geq 2$ . Let $S_{0}$ be the unique abelian subgroup of $G_{2}$ of index 2, which is generated by $x$ and $y$ . Let $ρ$ be the character of $S_{2} = Ω_{2} (S_{0}) = 〈 x^{2^{n - 2}} y^{2^{n - 2}} 〉$ such that for $0 \leq k, l < 4$
$ρ (x^{k 2^{n - 2}} y^{l 2^{n - 2}}) = {\begin{matrix} 4 & if k = l = 0 \\ - 4 & if k = l = 2 \\ - 2 & if k = 0 and l is odd \\ - 2 & if l = 0 and k is odd \\ 2 & if k and l are nonzero with opposite parity \\ 0 & otherwise. \end{matrix}$
Next let $H = G_{2} / 〈 x^{2^{n - 2}} y^{- 2^{n - 2}} 〉$ . Let $\bar{x}$ , $\bar{y}$ , and $\bar{z}$ be the images of $x$ , $y$ , and $z$ , respectively.
We notice that ${\bar{x}}^{2^{n - 2}} = {\bar{y}}^{2^{n - 2}}$ . Let $K = 〈 {\bar{x}}^{2^{n - 1}}, z 〉 \sim = Z_{2} \times Z_{2}$ . Let $ψ$ be the character of $K$ such that $ψ (\bar{1}) = 1$ , $ψ ({\bar{x}}^{2^{n - 1}}) = ψ (\bar{z}) = - 1$ , and $ψ ({\bar{x}}^{2^{n - 1}} \bar{z}) = 1$ . Let $χ (g) = I n d_{S_{2}}^{G_{2}} ρ (g) + 2 I n d_{K}^{H} ψ (g 〈 x^{2^{n - 2}} y^{- 2^{n - 2}} 〉)$ . We will show that $χ$ is a 2-effective character of $G$ . By definition $χ$ is a character of $G_{2}$ . Since $χ (1) = 3 (2^{2 n - 2})$ and $χ (g) = - 2^{2 n - 2}$ for each involution $g \in G_{2}$ , we see that $[χ |_{E}, 1_{E}] = 0$ for each rank 2 elementary abelian subgroup $E \subset G_{2}$ (see Remark 13 ).
It is only left to show that $χ$ respects fusion in $G$ . The fusion pattern in this situation has been described by Brauer and Wong [9,Lemma4] . First we will use is the fact that each element of $G_{2} \ S_{0}$ is conjugate in $G_{2}$ to some element of the form $x^{- h} y^{- h} z$ for $0 \leq h < 2^{n}$ . Also $x^{- h} y^{- h} z$ is conjugate in $G$ to $x^{2^{n - 1} - h} y^{- h}$ . Notice that $χ (x^{- h} y^{- h} z) = χ (x^{2^{n - 1} - h} y^{- h}) = 0$ unless $h = 0$ , $h = 2^{n - 2}$ , or $h = 2^{n - 1}$ . If $h = 0$ , then $x^{- h} y^{- h} z = z$ and $x^{2^{n - 1} - h} y^{- h} = x^{2^{n - 1}}$ . On the other hand if $h = 2^{n - 1}$ , then $x^{- h} y^{- h} z = x^{2^{n - 1}} y^{2^{n - 1}} z$ and $x^{2^{n - 1} - h} y^{- h} = y^{2^{n - 1}}$ . Since each of these four elements are involutions, $χ$ will take the value of $- 2^{2 n - 2}$ on each. Now consider the case when $h = 2^{n - 2}$ ; in this case $x^{- h} y^{- h} z = x^{- 2^{n - 2}} y^{- 2^{n - 2}} z$ and $x^{2^{n - 1} - h} y^{- h} = x^{2^{n - 2}} y^{- 2^{n - 2}}$ . We see that $χ (x^{- 2^{n - 2}} y^{- 2^{n - 2}} z) = χ (x^{2^{n - 1}} z) = 2^{2 n - 2}$ and that $χ (x^{2^{n - 2}} y^{- 2^{n - 2}}) = 2^{2 n - 2} - 2^{2 n - 1} = 2^{2 n - 2}$ .
We now check fusion inside of $S_{0}$ . Again by the work of Brauer and Wong [9,Lemma4] , we know that all such fusion occurs inside of $N_{G} (S_{0})$ and that all such fusion is generated by two elements: $z$ , which is in $G_{2}$ , and $η$ where $x^{η} = y$ and $y^{η} = x^{- 1} y^{- 1}$ . Since $z$ is in $G_{2}$ , $χ$ already is known to respect the fusion given by $z$ . Now we finish the proof of Lemma 35 by checking the fusion given by $η$ . Consider $x^{k} y^{l} \in S_{0}$ with $0 \leq k, l < 2^{n}$ . $(x^{k} y^{l})^{η} = x^{- l} y^{k - l}$ . Notice that if either $k$ or $l$ is not conjugate to $0$ , $2^{n - 2}$ , $2^{n - 1}$ , or $3 \cdot 2^{n - 2}$ modulo $2^{n}$ , then $χ (x^{k} y^{l}) = 0$ . The remaining 12 cases where $x^{k} y^{l}$ has order of at least 4 are easily verified. □

Proposition 36 (Jackson [21] ). If

G

is a finite group with a dihedral, semi-dihedral, or wreathed Sylow 2-subgroup such that

r k (G) = 2

, then

G

has a 2-effective character.

Proof. Let G 2 ∈ S y l 2 ( G ) and let n be the integer with | G 2 | = 2 n . If G 2 is either dihedral or semi-dihedral, then there is a character of G 2 such that
$χ (g) = {\begin{matrix} 3 * 2^{n - 3} & if g = 1 \\ - 2^{n - 3} & if g is an involution \\ 2^{n - 3} & otherwise . \end{matrix}$
Since χ is constant on elements of the same order, it must respect fusion in G as in Remark 15 . It is also clear that if E ⊆ G 2 is a rank 2 elementary abelian subgroup, [ χ | E , 1 E ] = 0 (see Remark 13 ). So χ is a 2-effective character of G
.
We may now assume that G 2 is wreathed and is generated by x , y , and z as in Definition 31 . Since G 2 is wreathed, one of the following holds (see [4,Proposition2,p.11] ):
- (1) $G$ has no normal subgroups of index 2,
- (2) $Z (G_{2})$ is strongly closed in $G_{2}$ with respect to $G$ ,
- (3) $G$ has a normal subgroup $K$ with Sylow 2-subgroup $S_{0}$ generated by $x$ and $y$ , or
- (4) $G$ is 2-nilpotent.
Case 1 was treated in Lemma 35 . Case 2 has been treated in Proposition 22 .
We now treat case 3. Suppose $G$ has a normal subgroup $K$ with Sylow 2-subgroup $S_{0}$ generated by $x$ and $y$ . Notice that $G_{2} \cap K = S_{0}$ . We see that $S_{0}$ is weakly closed in $G_{2}$ with respect to $G$ . By a standard result of Burnside (see [6,37.6] ), $N_{G} (S_{0})$ controls fusion in $C_{G_{2}} (S_{0}) = S_{0}$ . Also notice that any element of $G_{2} \ S_{0}$ is conjugate in $G$ to only those elements to which it is conjugate in $G_{2}$ . Also notice that $N_{G} (S_{0}) / C_{G} (S_{0}) \sim = Σ_{3}$ , and that this symmetric group permutes the elements $x$ , $y$ , and $x y$ ; furthermore, the character $χ$ of $G_{2}$ given in Lemma 35 is again a 2-effective character in this case.
Case 4 was treated in Lemma 33 . □

To conclude Section 5 , we state the following theorem whose proof has now been completed.

Theorem 37. For any rank two finite group

G

G

has a 2-effective character.

6 Counterexample

Looking at the last section, one may wonder if a similar argument can be made for odd primes; however, this is not the case. We will show the following proposition, which will show that there are rank two finite groups that do not have a

p

-effective character for each odd prime

p

Proposition 38. Let

p

be an odd prime and

T_{p}

be the semidirect product

(Z_{p} \times Z_{p}) ∖_{θ} S L_{2} (F_{p})

where the action

θ

is given by the obvious inclusion

S L_{2} (F_{p}) \to G L_{2} (F_{p}) \sim = A u t (Z_{p} \times Z_{p})

. If

G

is a finite group such that

G / O_{p'} (G) \sim = T_{p}

, then

G

does not have a

p

-effective character.

Proof. Let $G_{p} \in S y l_{p} (G)$ . Notice that $G_{p}$ , which is a Sylow $p$ -subgroup of $T_{p}$ , is an extra-special $p$ -group of size $p^{3}$ and exponent $p$ . Let $χ_{1}, . . ., χ_{n}$ be the irreducible characters of $G_{p}$ and assume that $χ_{1}$ is the trivial character. Notice from the structure of $G_{p}$ that if $χ_{i} (1) \neq 1$ , then for $g \in G_{p}$ , $χ_{i} (g) \neq 0$ if and only if $g \in Z (G_{p})$ . Also notice that $Z (G_{p})$ is cyclic of order $p$ and $Z (G_{p})$ is not strongly closed in $G_{p}$ with respect to $G$ . Let $x \in Z (G_{p})$ and $y \in G_{p} \ Z (G_{p})$ such that there is a $g \in G$ with $y^{g} = x$ . Suppose $χ$ is a character of $G_{p}$ that is a $p$ -effective character of $G$ . There exists $a_{1}, . . ., a_{n} \in Z_{\geq 0}$ such that $χ = \sum_{i = 1}^{n} a_{i} χ_{i}$ . Since $χ$ is a $p$ -effective character of $G$ , it respects fusion in $G$ , which implies $χ (x) = χ (y)$ and $a_{1} = 0$ . Suppose that $a_{i} > 0$ for some $i > 1$ with $χ_{i} (1) = 1$ . Fixing this $i$ , let $E_{i} = {g \in G_{p} | χ_{i} (g) = 1}$ . Notice that for such an $i$ , $E_{i}$ is a rank two elementary abelian subgroup of $G_{p}$ . This implies that $[χ_{i} |_{E_{i}}, 1_{E_{i}}] = 1$ , so $[χ |_{E_{i}}, 1_{E_{i}}] > 0$ , which contradicts the assumption that $χ$ is a $p$ -effective character of $G$ ; therefore, for each $i$ such that $χ_{i} (1) = 1$ , $a_{i} = 0$ . Now $χ (x) = \sum_{i such that χ_{i} (1) \neq 1} a_{i} χ_{i} (x) + \sum_{i such that χ_{i} (1) = 1} a_{i}$ $= \sum_{i such that χ_{i} (1) = 1} a_{i} χ_{i} (y) = χ (y) .$ So $χ (y) = 0$ ; therefore, $χ (x) = 0$ . Since $Z (G_{p})$ is a cyclic group of order $p$ , $χ (z) = 0$ for all $z \in Z (G_{p}) \ {1}$ . Notice that in showing $χ (z) = 0$ for each $z \in Z (G_{p}) \ {1}$ , we know that $χ (g) = 0$ for all $g \in G_{p} \ Z (G_{p})$ by the structure of $G$ ; thus, $χ (g) = 0$ for all $g \in G_{p} \ {1}$ . So we see that $χ$ must be identically zero, which contradicts the definition of $p$ -effective character. □

Lemma 39. Let

G

be a finite group with

p

a prime dividing

| G |

and

H \subseteq G

Suppose that

p

divides

| H |

and that

r k_{p} (G) = r k_{p} (H)

. If

G

has a

p

-effective character, so does

H

Proof. We may assume that $r k_{p} (H) = r k (G)$ ; otherwise the result is obvious. Let $G_{p} \in S y l_{p} (G)$ such that $H_{p} = G_{p} \cap H \in S y l_{p} (H)$ . Let $χ$ be a character of $G_{p}$ that is a $p$ -effective character of $G$ . Obviously $χ |_{H_{p}}$ is a character of $H_{p}$ , which is not identically zero. Since $χ$ respects fusion in $G$ , $χ |_{H_{p}}$ respects fusion in $H$ . Any maximal rank elementary abeliansubgroup of $H_{p}$ is also a maximal rank elementary abelian subgroup of $G_{p}$ . The lemma follows easily from this. □

Combining Lemma 39 with Theorems 29 and 37 , we get the following theorem from which Corollary 17 follows:

Theorem 40. Let

G

be a finite group of rank two and let

p

be a prime dividing

| G |

G

has

p

-effective character if and only if either

p = 2

or both

p > 2

and

G

does not contain a subgroup

H

such that

H / O_{p'} (H) \sim = T_{p}

In showing that a finite group

G

of rank two acts freely on a finite complex

Y ≃ S^{n} \times S^{m}

, we actually showed that

G

acts on a finite complex

X ≃ S^{m}

with isotropy groups of rank one and then applied Theorem 3 . Ozgun Unlu [25] has shown that for each odd prime

p

T_{p}

cannot act on any finite complex homotopy equivalent to a sphere with rank one isotropy groups.

From Unlu's proof it is clear that for any odd prime p and group

H

, such that for a normal

p^{'}

-subgroup

N ⊲ H

H / N \sim = T_{p}

H

cannot act on any finite complex homotopy equivalent to a sphere with rank one isotropy groups.

Combining the result of Unlu with Theorem 40 and the discussion in Section 2 , we conclude with the following proposition:

Proposition 41. Let

G

be a finite group of rank two.

G

acts on some finite complex homotopy equivalent to a sphere with rank one isotropy groups if and only if for each prime

p

G

does not contain a subgroup

H

such that

H / O_{p'} (H) \sim = T_{p}

References

A. Adem, J. Smith, On spaces with periodic cohomology, Electron. Res. Announc. Amer. Math. Soc. 6 (2000) 1-6.
A. Adem, J. Smith, Periodic complexes and group actions, Ann. of Math. (2) 154 (2001) 407-435.
J.L. Alperin, Sylow Intersections and Fusion, J. Algebra 6 (1967) 222-241.
J.L. Alperin, R. Brauer, D. Gorenstein, Finite groups with quasi-dihedral and wreathed Sylow 2-subgroups, Trans. Amer. Math Soc. 151 (1970) 1-261.
J.L. Alperin, R. Brauer, D. Gorenstein, Finite simple groups of 2-rank two, Scripta Math. 29 (1973) 191-214.
M. Aschbacher, Finite Group Theory, Cambridge studies in advanced mathematics, vol. 10, Cambridge Press, New York, 1986.
D. Benson, J. Carlson, Complexity and Multiple Complexes, Math. Zeit. 195, (1987) 221-238.
N. Blackburn, Generalizations of Certain Elementary Theorem on $p$ -Groups, Proc. London Math. Soc. 11, (1961) 1-22.
R. Brauer, W.J. Wong, Some properties of finite groups with wreathed Sylow 2-subgroup, J. Alg. 19 (1971) 263-273.
A. Diaz, A. Ruiz, A Viruel, All $p$ -local grops of rank two for odd prime $p$ , (in preparation).
J. Dietz, S. Priddy, The Stable Homotopy Type of Rank Two $p$ -Groups, in Homotopy Theory and its Applications (Cocoyoc, 1993), Contemp. Math. 188, Amer. Math. Soc. (1995) 93-103.
W. Dwyer, A. Zabrodsky, Maps between classifying spaces, In: Algebraic Topology, Barcelona 1986, Lecture Notes in Mathematics, vol. 1298, Springer Verlag, New York, 1987, pp. 106-119.
D. Gorenstein, Finite Groups, Harper & Row, New York, 1968.
J. Grodal, Higher limits via subgroup complexes, Ann. of Math. 155 (2002) 405-457.
A. Heller, A Note on Spaces with Operators, Illinois J. Math. 3 (1959) 98-100.
J. Higman, Suzuki 2-groups, Ill. J. Math. 7 (1962) 79-96.
I.M. Isaacs, Character Theory of Finite Groups, Academic Press, New York, 1976.
A. Ruiz, A. Viruel, The classification of $p$ -local finite groups over the extraspecial group of order $p^{3}$ and exponent $p$ , Math. Ziet. (to appear).
A. Ish-Shalom, On Sylow intersections, Bull. Austral. Math. Soc. 16 (1977) 237-246.
M. Jackson, A quotient of the set $[B G, B U (n)]$ for a finite group $G$ of small rank, J. Pure and Applied Alg. 188 (2004) 161-174.
M. Jackson, Vector bundles over $B G$ whose Euler classes are effective, Ph.D. Thesis Purdue Univ. 2001.
J. Martino, S. Priddy, On the Cohomology and Homotopy of Swan Groups, Math. Z. 225 (1997) 277-288.
M. Suzuki, Group Theory II, Springer-Verlag, New York, 1986.
R.G. Swan, Periodic Resolutions for Finite Groups, Ann. of Math. 72 (1960) 267-291.
O. Unlu, Ph.D. Thesis University of Wisconsin at Madison 2004.

Department of Mathematics, Hylan Building, University of Rochester, Rochester, NY 14627, USA, mjackson@math.rochester.edu

Michael A. Jackson