0
, then
n_{2}\neq n_{3}
and
N_{\tilde{G}}(\tilde{C})=N_{\tilde{G}_{1}}(\tilde{C}_{1})\times\tilde{G}_{2}\times\tilde{G}_{3}
.
(f)
In all cases,
|\tilde{D}|=|D|p^{a}
,
|t|=p^{a+a_{1}}
and
|\bar{D}|=p^{a+a_{1}-c^{\prime}}
.
(g)
We have
C_{\tilde{G}}(\tilde{D})\leq C_{\tilde{G}}(D)
and
N_{\tilde{G}}(\tilde{D})\leq N_{\tilde{G}}(D)
, with equality in either
instance, if the minimal polynomial of
t
is reducible.
Proof.
(a) Lemma
3.15
(a) implies that
|t_{1}|\mid p^{a+a_{1}}
with
equality if
a_{1}>0
, and
|t_{j}|\mid p^{a+a_{1}}
for
1\leq j\leq h
.
(b) The first inequality
of (
1
) follows from Lemma
3.11
(b), the
second one from
D=\langle t,O_{p}(Z)\rangle
and
|O_{p}(Z)|=p^{c}\leq p^{a}
,
and the last one from (a).
The last assertions are clear from (
1
).
(c) Here,
D=G\cap O_{p}(Z(\tilde{C}_{1}))
. If
a_{1}=0
, then
\tilde{C}_{1}=\tilde{G}
and
D\leq Z
, a case we have excluded. Thus
a_{1}>0
and
D
is cyclic of order
p^{a+a_{1}}
by (a) and
Lemma
3.8
(f). If
\tilde{D}_{1}
is not cyclic, then
p^{a+a_{1}}=|D|\geq|\tilde{D}|/p^{a}\geq p^{a+2a_{1}}
, where the latter
estimate follows from Lemma
3.5
, applied to
\tilde{C}_{1}
.
This contradiction shows that
\tilde{D}
is cyclic. In particular,
|\tilde{D}|=p^{a+a_{1}+b_{1}}
, and
\tilde{D}
is the Sylow
p
-subgroup of a
maximal torus of
\tilde{G}
; see
[
HL25
, Corollary
3.6.2
]
.
Lemma
3.11
(d) implies that
p^{a+a_{1}}=|D|=|\tilde{D}|/p^{a}=p^{a_{1}+b_{1}}
, and thus
b_{1}=a
as
claimed.
(d) Recall that
3\geq r(\tilde{D})=r(\tilde{D}_{1})+r(\tilde{D}_{2})
.
If
\tilde{D}_{j}
is non-cyclic, then
|\tilde{D}_{j}|\geq p^{2(a+a_{j})}
by Lemma
3.5
, applied to
\tilde{C}_{j}
for
j=1,2
.
Suppose first that
\tilde{D}_{1}
is non-cyclic. Then
\tilde{D}_{2}
is cyclic,
and, using (
1
) and
[
HL25
, Corollary
3.6.2
]
, we get
p^{3a+a_{1}}\geq|\tilde{D}|\geq p^{2(a+a_{1})+a+a_{2}+b_{2}}=p^{3a+2a_{1}+a_{2}+b_{2}},
which implies
a_{1}=a_{2}=b_{2}=0
and also that
|\tilde{D}|=p^{3a}
.
Now
D=G\cap O_{p}(Z(\tilde{C}))
. Since
a_{1}=a_{2}=0
, we have
|O_{p}(Z(\tilde{C}))|=p^{2a}
. Clearly,
O_{p}(Z(\tilde{C}))\not\leq G
since
b_{2}=0
, i.e.
p\nmid n_{2}
. Hence
|D|0
in the following. By (b) and Lemma
3.8
(e), we obtain
p^{a}\mid n
. We claim that
b_{2}=0
, i.e.
p\nmid n_{2}^{\prime}
. Indeed,
|\det(t_{1})|=p^{a}
by Lemma
3.15
(d), and
\det(t_{2})=\xi_{2}^{m_{2}m_{2}^{\prime}p^{b_{2}}}=\xi_{2}^{n_{2}^{\prime}m_{2}^{\prime}}
, as
a_{2}=0
. From
\det(t_{1})\det(t_{2})=\det(t)=1
, we conclude
p^{a}=|\det(t_{1})|=|\det(t_{2})|
, and thus
p\nmid n_{2}^{\prime}
, which is our claim. Now
n=n_{1}^{\prime}p^{a_{1}}+n_{2}^{\prime}
and
p\nmid n_{2}^{\prime}
. As
a_{1}>0
, we conclude that
p\nmid n
, a contradiction.
We next show that
b_{1}=b_{2}=0
, assuming first that
a_{1}>0
. Then
|t|=p^{a+a_{1}}
by (a). Now
D=\langle t,O_{p}(Z)\rangle
and
|O_{p}(Z)|=p^{c}
, so that
|D|\leq|t||O_{p}(Z)|=p^{a+a_{1}+c}
.
Since
\tilde{D}_{1}
and
\tilde{D}_{2}
are cyclic, we get
|\tilde{D}|=p^{2a+a_{1}+b_{1}+a_{2}+b_{2}}
. As
\tilde{D}
is a Sylow
p
-subgroup
of
\tilde{G}
by
[
HL25
, Corollary
3.6.2
]
,
Lemma
3.11
(d) implies that
(2)
|D|=p^{a+a_{1}+b_{1}+a_{2}+b_{2}},
and thus
(3)
c\geq b_{1}+a_{2}+b_{2}.
Since
a\geq c\geq b_{1}+a_{2}+b_{2}
, we obtain
a\geq b_{j}
for
j=1,2
.
If also
a_{2}>0
, we have
|\det(t_{j})|=q^{a-b_{j}}
for
j=1,2
by
Lemma
3.15
(d). From
|\det(t_{1})|=|\det(t_{2})|
, we obtain
b_{1}=b_{2}
, if
a_{1},a_{2}>0
.
Assume now that
a_{1}>a_{2}>0
. Then
n=m_{1}p^{a_{1}+b_{1}}+m_{2}p^{a_{2}+b_{2}}
. As
a_{1}+b_{1}>a_{2}+b_{1}=a_{2}+b_{2}
, we obtain
a_{2}+b_{2}=b\geq c\geq b_{1}+a_{2}+b_{2}
. It follows
that
b_{2}=b_{1}=0
.
Next assume that
a_{1}>a_{2}=0
. Then
|\det(t_{1})|=p^{a-b_{1}}
by
Lemma
3.15
(d). Since
\det(t_{2})=\xi_{2}^{m_{2}m_{2}^{\prime}p^{b_{2}}}
and
|\xi_{2}|\mid p^{a}
, it follows that
b_{2}=0
if
b_{1}=0
. Now
n=m_{1}p^{a_{1}+b_{1}}+m_{2}p^{b_{2}}
. If
a_{1}+b_{1}>b_{2}
, then
b_{2}=b\geq c\geq b_{1}+b_{2}
and hence
b_{1}=0
. As noticed above, this
implies
b_{2}=0
. Otherwise,
a_{1}+b_{1}\leq b_{2}
. If
a=b_{1}
, we obtain
b_{2}=0
from
c\geq b_{1}+b_{2}=a+b_{2}\geq a\geq c
. But then
p\nmid n
, contradicting
c\geq b_{1}+b_{2}=a>0
. Hence
a\neq b_{1}
,
i.e.
|\det(t_{1})|=p^{a-b_{1}}>1
. Writing
|\xi_{2}|=p^{e}
, we obtain
a-b_{1}=e-b_{2}
from
|\det(t_{1})|=|\det(t_{2})|
.
It follows that
a-b_{1}=e-b_{2}\leq e-a_{1}-b_{1}
, and so
e\geq a+a_{1}>a
, a contradiction.
Finally assume that
a_{1}=a_{2}>0
. As
D=G\cap O_{p}(Z(\tilde{C}))
and
O_{p}(Z(\tilde{C}))=\langle t_{1}\rangle\times\langle t_{2}\rangle
we get
D\cong\{(t_{1}^{i},t_{2}^{j})\mid\det(t_{1}^{j})=\det(t_{2}^{i})^{-1}\}
.
Thus
|D|=p^{a+a_{1}+a_{2}+b_{1}}
, as the kernel of the map
\det\colon\langle t_{2}\rangle\rightarrow\mathbb{F}^{*}
has order
p^{a_{2}+b_{2}}
. From
p^{a+a_{1}+a_{2}+b_{2}}=|D|\geq|\tilde{D}|/p^{a}=p^{a+a_{1}+b_{1}+a_{2}+b_{2}}
we obtain
b_{1}=0
, and thus
b_{2}=0
.
Let us now assume that
a_{1}=a_{2}=0
. Simultaneously to the proof
hat of
b_{1}=b_{2}=0
, we also prove statement (iii).
Choose the notation such that
b_{1}\geq b_{2}
. Here,
O_{p}(Z(\tilde{C}))=\langle t_{1}^{\prime}\rangle\times\langle t_{2}^{\prime}\rangle
with
|t_{j}^{\prime}|=p^{a}
for
j=1,2
. Since
\det(t_{1}^{\prime})
and
\det(t_{2}^{\prime})
lie in the same subgroup of
\mathbb{F}^{*}
and since
b_{1}\geq b_{2}
, we get
|D|=|G\cap O_{p}(Z(\tilde{C}))|=p^{a+b_{2}}
. On the other hand
|D|\geq|\tilde{D}|/p^{a}=p^{a+b_{1}+b_{2}}
, which implies that
b_{1}=b_{2}=0
.
From
p\nmid n_{1}n_{2}
, we find
D=G\cap O_{p}(Z(\tilde{C}))\cong\langle t_{1}^{\prime}\rangle
. This yields (iii).
Let us finally prove the statements (i)β(vii).
We have already proved (iii). Clearly, (v) follows from (i) and (ii). By (iii),
it suffices to prove (iv) in the
situation of (i). In that case,
c=a_{2}=0
, so that
p\nmid|Z|
. Hence
Y
is trivial and so
D=\bar{D}
is cyclic. Let us now prove (i), (ii).
In this situation,
|t|=p^{a+a_{1}}
and
|D|=p^{a+a_{1}+a_{2}}
by (
2
). Moreover,
a_{2}\leq c
by (
3
).
Assume now that
a_{1}>a_{2}
. As
n=m_{1}p^{a_{1}}+m_{2}p^{a_{2}}
with
p\nmid m_{1}m_{2}
and
a_{1}>a_{2}
, we
get
b=a_{2}
. Since
c\leq b=a_{2}\leq c
, we find
a_{2}=c
.
It follows that
p^{a+a_{1}-c^{\prime}}=|\bar{D}|=|D/Y|\geq|D/O_{p}(Z)|=p^{a+a_{1}}\geq p^{a+a_{1}-c^{\prime}}.
Hence
c^{\prime}=0
,
Y=O_{p}(Z)
and
\langle t\rangle\cap O_{p}(Z)=\{1\}
.
This proves (i). Now assume that
a_{1}=a_{2}>0
. Here,
|D|=p^{a+2a_{1}}
by (
2
). Since
|t|=p^{a+a_{1}}
and
|\langle t\rangle\cap Y|=p^{c^{\prime}}
by definition, we get
p^{a+a_{1}-c^{\prime}}=|\bar{D}|=|D/Y|.
In particular,
|Y|=p^{a_{1}+c^{\prime}}
. Moreover,
|\langle t\rangle\cap O_{p}(Z)|=p^{c-a_{1}}
, since
D=\langle t,O_{p}(Z)\rangle
. Assume that
|Y|\leq p^{c-a_{1}}
. Then
Y\leq\langle t\rangle\cap O_{p}(Z)
, which
implies
Y=\langle t\rangle\cap Y
. It follows that
p^{a_{1}+c^{\prime}}=|Y|=|\langle t\rangle\cap Y|=p^{c^{\prime}},
which implies
a_{1}=0
, a contradiction. Hence
|Y|>p^{c-a_{1}}
. Then
\langle t\rangle\cap O_{p}(Z)\leq Y
, which implies
\langle t\rangle\cap Y=\langle t\rangle\cap O_{p}(Z)
.
It follows
that
p^{c^{\prime}}=p^{c-a_{1}}
, and so
c^{\prime}=c-a_{1}
. Moreover,
|Y|=p^{a_{1}+c^{\prime}}=p^{a_{1}+c-a_{1}}=p^{c},
and thus
Y=O_{p}(Z)
. This completes the proof of (ii).
To prove (vi), assume that
n_{1}=n_{2}
. Then
n=2n_{1}
and
p\nmid n_{1}
. Hence
c=0
, contradicting
0\leq c^{\prime}0
for
j=1,2
, then
|t_{j}|=p^{a+a_{j}}
by
Lemma
3.15
(a). If
a_{1}>a_{2}=0
, then
|\det(t_{1})|=p^{a}
by
Lemma
3.15
(d) and thus
|t_{2}|=p^{a}
, since
\det(t_{1}t_{2})=1
.
If
a_{1}=a_{2}=0
, then
t=t_{1}t_{2}=t_{2}t_{1}
,
|t|=p^{a}
and
\det(t_{1}t_{2})=1
imply
|t_{1}|=|t_{2}|=p^{a}
. This proves (vii).
(e) Since
r(\tilde{D})\leq 3
by Lemma
3.11
(b), the factors
\tilde{D}_{j}
are cyclic for
1\leq j\leq 3
. By
[
HL25
, Corollary
3.6.2
]
, we have
|\tilde{D}_{j}|=p^{a+a_{j}+b_{j}}
for
1\leq j\leq 3
. The first two claims follow from (
1
).
For the consequences of these consult (b) and Lemma
3.8
(d) and (e).
Let us now prove the final claim. We have
n=m_{1}p^{a_{1}}+n_{2}+n_{3}
and
p\nmid n_{2}n_{3}
. Since
p^{a}\mid n
and
a>0
, we also have
p\mid n_{2}+n_{3}
since
a_{1}>0
. This implies
n_{2}\neq n_{3}
since
p
is odd.
Now
\tilde{C}=C_{\tilde{G}}(t)
, and thus
N_{\tilde{G}}(\tilde{C})
permutes the spaces
V_{1}
,
V_{2}
and
V_{3}
. Since these have pairwise distinct
dimensions,
N_{\tilde{G}}(\tilde{C})
fixes all of these, which gives our
claim.
(f) This is contained in (c)β(e).
(g) Suppose first that the minimal polynomial of
t
is irreducible.
Then
\tilde{D}
is cyclic by (c), and our claims hold.
Suppose next that the minimal polynomial of
t
has two irreducible
factors. Then,
b_{1}=b_{2}=0
by (d). Thus
|\tilde{D}|=p^{2a+a_{1}+a_{2}}=|O_{p}(Z(\tilde{C}))|
and so
\tilde{D}=O_{p}(Z(\tilde{C}))\leq Z(\tilde{C})
. Hence
\tilde{C}\leq C_{\tilde{G}}(\tilde{D})\leq C_{\tilde{G}}(D)=\tilde{C}
. Since
D=G\cap\tilde{D}
, we obtain
N_{\tilde{G}}(\tilde{D})\leq N_{\tilde{G}}(D)\leq N_{\tilde{G}}(C_{\tilde{G}}(D))=N_{\tilde{G}}(\tilde{C})\leq N_{\tilde{G}}(O_{p}(Z(\tilde{C})))=N_{\tilde{G}}(\tilde{D})
.
The case when the minimal polynomial of
t
has three irreducible
factors is treated analogously.
β
3.17.
The intermediate configuration
Let us write
h\in\{1,2,3\}
for the number of irreducible factors of the
minimal polynomial of
t
. If
h=1
, let
\bar{E}
denote the trivial subgroup
of
\bar{D}
; otherwise, let
\bar{E}
be the unique subgroup of
\bar{D}
of
order
p
.
Let
\bar{{\mathbf{d}}}
denote a Brauer correspondent of
\bar{{\mathbf{B}}}
in
C_{\bar{G}}(\bar{E})
. Then
W(\bar{{\mathbf{B}}})=W(\bar{{\mathbf{d}}})
. The latter
will be computed by pulling back
(\bar{E},\bar{{\mathbf{d}}})
to
G
.
Notation 3.18
.
Assume the terminology of Notation
3.7
.
(i) Put
E:=\langle t^{\prime},Y\rangle
with
t^{\prime}=1
, if
h=1
,
and
t^{\prime}:=t^{p^{a+a_{1}-c^{\prime}-1}}
, otherwise. (As we have assumed that
\bar{D}
is non-trivial and
|\bar{D}|=p^{a+a_{1}-c^{\prime}}
by
Lemma
3.16
(f), we have
a+a_{1}-c^{\prime}\geq 1
.)
(ii) Let
\tilde{{\mathbf{H}}}:=C_{\tilde{{\mathbf{G}}}}(E)=C_{\tilde{{\mathbf{G}}}}(t^{\prime})
and
{\mathbf{H}}:=\tilde{{\mathbf{H}}}\cap{\mathbf{G}}=C_{{\mathbf{G}}}(E)=C_{{\mathbf{G}}}(t^{\prime})
.
β
Notice that
\tilde{{\mathbf{H}}}=\tilde{{\mathbf{G}}}
and
{\mathbf{H}}={\mathbf{G}}
if the minimal
polynomial of
t
is irreducible. Notice also that the inclusion
{\mathbf{H}}\rightarrow\tilde{{\mathbf{H}}}
is a regular embedding, and that
\tilde{{\mathbf{H}}}
and
{\mathbf{H}}
are
F
-stable.
We collect a few properties of the objects introduced above, assuming the
terminology of
Notations
3.7
,
3.10
,
3.14
,
3.18
and
Lemma
3.15
(e). Also, recall that
D=\langle t,O_{p}(Z)\rangle
by Lemma
3.8
(c),
and that
|t|=p^{a+a_{1}}
by Lemma
3.16
(f).
Lemma 3.19
.
Assume that
h\geq 2
. For Parts
(e)
and
(f)
assume in addition
that
|\{n_{1},n_{2},n_{3}\}|\geq 2
if
h=3
and
a_{1}=0
. Then the
following statements hold.
(a)
We have
|t^{\prime}|=p^{c^{\prime}+1}
.
(b)
We have
N_{\tilde{G}}(D)\leq N_{\tilde{G}}(E)
and
N_{G}(D)\leq N_{G}(E)
.
(c)
Suppose that
h=2
. Then
\tilde{H}=\tilde{G}_{1}\times\tilde{G}_{2}
unless
a_{1}=a_{2}=0
and
c^{\prime}0
, then
\tilde{H}=\tilde{G}_{1}\times\tilde{G}_{2,3}
with
\tilde{G}_{2,3}:=I({V}_{2}\oplus{V}_{3},\kappa_{2,3})
, where
\kappa_{2,3}
denotes the restriction of
\kappa
to
{V}_{2}\oplus{V}_{3}
.
(e)
We have
[N_{G}(E)\colon\!H]=[N_{\tilde{G}}(E)\colon\!\tilde{H}]\leq 2
.
(f)
We have
\bar{H}=C_{\bar{G}}(\bar{E})
.
Proof.
An element
\xi\in\mathbb{F}^{*}
is called rational, if
\xi\in\mathbb{F}_{q^{\delta}}
. If
\xi
is a
p
-element, this is the case
if and only if
|\xi|\leq p^{a}
.
(a) This follows from
|t|=p^{a+a_{1}}
.
(b) Notice that
Y
is a central subgroup of
\tilde{G}
so that
Y\leq N_{\tilde{G}}(D)
and
Y\leq N_{\tilde{G}}(E)
. Hence
N_{\tilde{G}}(D)/Y=N_{\tilde{G}/Y}(D/Y)\leq N_{\tilde{G}/Y}(E/Y)=N_{\tilde{G}}(E)/Y,
where the equalities arise from
[
HL24
, Lemma
2.2
(a)]
, and the inclusion
is due to the fact that
D/Y
is cyclic. Our assertions follow from this.
(c) Suppose first that
a_{1}>a_{2}
, so that Lemma
3.16
(d)(i)
applies. As
c^{\prime}=0
, we have
|t^{\prime}|=p\leq p^{a}
, so that the eigenvalues
of
t^{\prime}
on
V_{j}
are rational for
j=1,2
. As
\langle t^{\prime}\rangle\cap O_{p}(Z)=\{1\}
, the
eigenvalues of
t^{\prime}
on
V_{1}
and
V_{2}
are distinct.
Now suppose that
a_{1}=a_{2}
, so that
Lemma
3.16
(d)(ii)(iii) applies. Hence
|D|=p^{a+2a_{1}}
.
From
D=\langle t,O_{p}(Z)\rangle
and
|t|=p^{a+a_{1}}
we conclude that
|\langle t\rangle\cap O_{p}(Z)|=p^{c-a_{1}}
. By (a), we have
t^{\prime}\in Z\leq Z(\tilde{G})
, if and only if
c^{\prime}0
.
Now suppose that
c^{\prime}=c-a_{1}
. If
a_{1}>0
, then
c^{\prime}+1=c-a_{1}+1\leq c\leq a
, so that the eigenvalues of
t^{\prime}
on
V_{1}
and
V_{2}
are
rational. If
a_{1}=0
, then
|t|=p^{a}
and thus the
eigenvalues of
t^{\prime}
on on
V_{1}
and
V_{2}
are rational.
(d) Apply Lemma
3.16
(e). Suppose first that
a_{1}>0
.
Then
|t_{1}|=p^{a+a_{1}}
and
p^{a}\mid|t_{j}|
for
j=2,3
. As
|t^{\prime}|=p
and
a+a_{1}-1\geq a
, the eigenvalues of
t^{\prime}
on
V_{1}
have
order
p
, and
V_{2}\oplus V_{3}
is the fixed space of
t^{\prime}
. This proves our
assertion.
Suppose now that
a_{1}=0
. Then
|t|=p^{a}
and all the eigenvalues of
t
are
rational. Moreover, as
Y=O_{p}(Z)
and
|O_{p}(Z)|=p^{a}
, we may assume that
t
acts as the identity on
V_{3}
, i.e.
\xi_{3}=1
. From
\det(t)=1
and
p\nmid n_{1}n_{2}
we conclude
|\xi_{1}|=|\xi_{1}^{n_{1}}|=|\xi_{2}^{n_{2}}|=|\xi_{2}|
. Thus
|\xi_{1}|=|\xi_{2}|=p^{a}
.
The eigenvalues of
t^{\prime}
are
\xi_{1}^{p^{a-1}}
,
\xi_{2}^{p^{a-1}}
and
1
.
Assume that
\xi_{1}^{p^{a-1}}=\xi_{2}^{p^{a-1}}
. Then
1=\det(t^{\prime})=\xi_{1}^{p^{a-1}n_{1}}\xi_{2}^{p^{a-1}n_{2}}=\xi_{1}^{p^{a-1}(n_{1}+n_{2})}
. As
p
divides
n=n_{1}+n_{2}+n_{3}
and
p\nmid n_{1}n_{2}n_{3}
,
the sum
n_{1}+n_{2}
is prime to
p
. This contradicts the fact that
|\xi_{1}|=p^{a}
and thus implies that
t^{\prime}
has three pairwise distinct
eigenvalues. This proves our assertion.
(e) As
[\tilde{H}\colon\!H]=q-\varepsilon
by
Lemma
3.4
, we obtain
\tilde{H}G=\tilde{G}
.
It follows that
N_{\tilde{G}}(E)G=\tilde{G}
and thus
[N_{\tilde{G}}(E)\colon\!N_{G}(E)]=q-\varepsilon
.
Hence
[N_{G}(E)\colon\!H]=[N_{\tilde{G}}(E)\colon\!\tilde{H}]
.
The structure of
\tilde{H}
determined in (c) and (d) implies that
[N_{\tilde{G}}(E)\colon\!\tilde{H}]\leq 2
, as
N_{\tilde{G}}(E)
permutes the eigenspaces of
t^{\prime}
on
V
.
(f) This follows from (e) and Lemma
2.2
.
β
The following example shows that the extra hypothesis for the statements (e)
and (f) in Lemma
3.19
is necessary.
Example 3.20
.
Let
p=n=3
, and consider
G=\text{\rm SL}_{3}(7)
. Then
\bar{G}:=G/Z=\text{\rm PSL}_{3}(7)
has a cyclic
3
-block
\bar{{\mathbf{B}}}
of defect
1
.
Now
G
has a unique conjugacy class of non-central elements of order
3
. If we
let
t:=\left(\begin{array}[]{ccc}\xi&0&0\\
0&\xi^{-1}&0\\
0&0&1\end{array}\right),
where
\xi\in\mathbb{F}_{7}
has order
3
, we may assume that
D=\langle t,Z\rangle
is a defect group of the block
{\mathbf{B}}
of
G
dominating
\bar{{\mathbf{B}}}
.
We are thus in the situation that
h=3
,
a_{1}=a_{2}=a_{3}=0
and
n_{1}=n_{2}=n_{3}=1
. Moreover,
D=E
. However,
|C_{G}(E)|=36=|C_{\bar{G}}(\bar{E})|
, so that
\overline{C_{G}(E)}
has index
3
in
C_{\bar{G}}(\bar{E})
.
β
3.21.
The intermediate blocks
Keep the notation of Subsection
3.17
. In
particular,
h
denotes the number of irreducible factors of the minimal
polynomial of
t
. Recall that
\bar{H}=\overline{C_{G}(E)}=C_{\bar{G}}(\bar{E})
by
Lemma
3.19
(f), unless
h=3
,
a_{1}=0
and
n_{1}=n_{2}=n_{3}
. In any case,
\bar{H}\leq C_{\bar{G}}(\bar{E})
. Notice
that
C_{G}(D)=C_{H}(D)
and
C_{\tilde{G}}(\tilde{D})=C_{\tilde{H}}(\tilde{D})
.
Notation 3.22
.
(i) Let
{\mathbf{d}}
denote the block of
H
such that
(E,{\mathbf{d}})\leq(D,{\mathbf{c}})
as Brauer pairs of
H
.
(ii) Let
\tilde{{\mathbf{d}}}
denote the block of
\tilde{H}
such that
(\tilde{D},\tilde{{\mathbf{c}}})
is a
\tilde{{\mathbf{d}}}
-Brauer pair of
\tilde{H}
.
(iii) Let
\bar{{\mathbf{d}}}
denote the block of
\bar{H}
dominated by
{\mathbf{d}}
.
Lemma 3.23
.
If
h=1
, assume that
D\not\leq Z
. Then
\tilde{{\mathbf{d}}}
covers
{\mathbf{d}}
and
\bar{{\mathbf{d}}}
is a Brauer correspondent of
\bar{{\mathbf{B}}}
.
Proof.
Observe that
C_{\tilde{H}}(D)=C_{\tilde{G}}(D)=\tilde{C}
. The
block
\tilde{{\mathbf{c}}}
of
\tilde{C}
has defect group
\tilde{D}
, and
so
\tilde{{\mathbf{d}}}
has defect group
\tilde{D}
by Lemma
2.3
.
Since
C_{G}(D)=C_{H}(D)
,
the inclusion
(E,{\mathbf{d}})\leq(D,{\mathbf{c}})
of Brauer pairs of
H
is also an
inclusion of Brauer pairs of
G
. Hence
(E,{\mathbf{d}})
is a
{\mathbf{B}}
-Brauer pair by
[
Lin18
, Proposition
6.3.6
]
, and so
{\mathbf{d}}
has defect group
D
, once
more by Lemma
2.3
.
Notice that
N_{\tilde{H}}(\tilde{D})\leq N_{\tilde{H}}(D)
by
Lemma
3.16
(g). Let
\tilde{{\mathbf{b}}}
and
{\mathbf{b}}
denote the
Brauer correspondent of
\tilde{{\mathbf{d}}}
, respectively
{\mathbf{d}}
,
in
N_{\tilde{H}}(D)
, respectively
N_{H}(D)
.
By
the Harris-KnΓΆrr correspondence
[
HK85
, Theorem]
, it suffices to show
that
\tilde{{\mathbf{b}}}
covers
{\mathbf{b}}
. By definition of the inclusion of Brauer
pairs,
\tilde{{\mathbf{b}}}
covers
\tilde{{\mathbf{c}}}
, and
{\mathbf{b}}
covers
{\mathbf{c}}
.
As
\tilde{{\mathbf{c}}}
covers
{\mathbf{c}}
by definition,
\tilde{{\mathbf{b}}}
covers some block
of
N_{H}(D)
covering
{\mathbf{c}}
. We will show that
N_{H}(D)/C
is a
p
-group.
Then
{\mathbf{b}}
is the unique block of
N_{H}(D)
covering
{\mathbf{c}}
and
hence
\tilde{{\mathbf{b}}}
covers
{\mathbf{b}}
.
Suppose that
h=1
, so that
\tilde{H}=\tilde{G}
. By
Lemma
3.16
(c) we have
C_{\tilde{G}}(D)=\tilde{C}\cong\text{\rm GL}_{n_{1}}^{\varepsilon}(q^{p^{a_{1}}})
and thus
|N_{\tilde{G}}(\tilde{C})/\tilde{C}|=p^{a_{1}}
. Now consider the chain of
maps
N_{G}(D)\hookrightarrow N_{\tilde{G}}(D)\hookrightarrow N_{\tilde{G}}(C_{\tilde{G}}(D))=N_{\tilde{G}}(\tilde{C})\rightarrow N_{\tilde{G}}(\tilde{C})/\tilde{C},
whose kernel equals
N_{G}(D)\cap\tilde{C}=C
. It follows that
N_{G}(D)/C
is a
p
-group.
Suppose now that
h\geq 2
. First,
\overline{N_{H}(D)}=N_{\bar{H}}(\bar{D})
by
[
HL24
, Lemma
2.2
(a)]
.
Thus
N_{H}(D)/C\cong\overline{N_{H}(D)}/\bar{C}\cong N_{\bar{H}}(\bar{D})/\bar{C}.
Now
N_{\bar{H}}(\bar{D})/C_{\bar{H}}(\bar{D})
is a
p
-group,
since
\bar{D}
is cyclic and
N_{\bar{H}}(\bar{D})
acts trivially
on
\bar{E}
. Also,
C_{\bar{H}}(\bar{D})/\bar{C}
is a
p
-group by
Lemma
2.2
.
Since
|N_{\bar{H}}(\bar{D})/\bar{C}|=|N_{\bar{H}}(\bar{D})/C_{\bar{H}}(\bar{D})|\cdot|C_{\bar{H}}(\bar{D})/\bar{C}|,
it follows that
N_{H}(D)/C
is a
p
-group.
We now show that
\bar{{\mathbf{d}}}
is a Brauer correspondent of
\bar{{\mathbf{B}}}
. This is
trivial if
h=1
. Assume then that
h\geq 2
. As
\bar{H}=C_{\bar{G}}(\bar{D})
, this statement makes sense. By
Lemma
3.19
(b), we have
N_{G}(D)\leq N_{G}(E)
.
Let
{\mathbf{b}}^{\prime}
and
\bar{{\mathbf{b}}^{\prime}}
denote the Brauer correspondents of
{\mathbf{B}}
in
N_{G}(E)
, respectively of
\bar{{\mathbf{B}}}
in
N_{\bar{G}}(\bar{E})
.
Then
\overline{N_{G}(E)}=N_{\bar{G}}(\bar{E})
by
[
HL24
, Lemma
2.2
(a)]
, and
{\mathbf{b}}^{\prime}
dominates
\bar{{\mathbf{b}}}^{\prime}
by
[
HL25
, Lemma
2.4.1
]
.
Now
(E,{\mathbf{d}})
is a
{\mathbf{B}}
-Brauer pair
and so
{\mathbf{b}}^{\prime}
covers
{\mathbf{d}}
. This easily implies that
\bar{{\mathbf{b}}^{\prime}}
covers
\bar{{\mathbf{d}}}
, and so
\bar{{\mathbf{d}}}
is a Brauer correspondent
of
\bar{{\mathbf{B}}}
.
β
Figure 2.
Some subgroups and blocks of
\text{\rm GL}^{\varepsilon}_{n}(q)
, II
\textstyle{\tilde{H}=C_{\tilde{G}}(E),\tilde{{\mathbf{d}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}
\textstyle{H=C_{G}(E),{{\mathbf{d}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}
\textstyle{\tilde{C}=C_{\tilde{G}}(D),\tilde{{\mathbf{c}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}
\textstyle{\bar{H},\bar{{\mathbf{d}}}}
\textstyle{C=C_{G}(D),{{\mathbf{c}}}}
A diagram of the relevant groups and blocks is displayed in
Figure
2
.
The maximal tori of
\tilde{C}
and
\tilde{H}
are also maximal tori
of
\tilde{G}
. The
\tilde{G}
-conjugacy classes of maximal tori of
\tilde{G}
are labelled by partitions of
n
. If the
\tilde{G}
-conjugacy class of a
maximal torus
\tilde{T}
is labelled by the partition
\pi=(f_{1},f_{2},\ldots,f_{m})
, we call
\pi
the type of
\tilde{T}
.
Then
(4)
\tilde{T}\cong\tilde{T}_{f_{1}}\times\cdots\times\tilde{T}_{f_{m}},
and
N_{\tilde{G}}(\tilde{T})
fixes the sub-products of (
4
)
corresponding to the same
f_{j}
.
Lemma 3.24
.
If
h=3
, assume that
a_{1}>0
. Then
\tilde{{\mathbf{d}}}
is strictly regular with
respect to a maximal torus
\tilde{T}
of
\tilde{H}
whose type has the parts
n_{1},\ldots,n_{h}
.
In particular, there is
\theta\in\Irr(\tilde{T})
of
p^{\prime}
-order and
in general position with respect to
\tilde{H}
, such that
\tilde{\chi}:=\varepsilon_{\tilde{{\mathbf{T}}}}\varepsilon_{\tilde{{\mathbf{H}}}}R_{\tilde{{\mathbf{T}}}}^{\tilde{{\mathbf{H}}}}(\theta)
is an irreducible character of
\tilde{{\mathbf{d}}}
.
Proof.
For
j\in\{1,\ldots,h\}
, the block
\tilde{{\mathbf{c}}}_{j}
of
\tilde{C}_{j}
is
strictly regular with respect to a cyclic maximal torus
\tilde{T_{j}}
of
\tilde{C}_{j}
, of type
(n_{j})
when viewed as a torus of
\tilde{G}_{j}
,
and
\tilde{D}_{j}
is a Sylow
p
-subgroup of
\tilde{T_{j}}
; see
[
HL25
, Corollary
3.6.2
]
. Thus
\tilde{T}=\tilde{T}_{1}\times\cdots\times\tilde{T}_{h}
contains
\tilde{D}
as a
Sylow
p
-subgroup and
\tilde{{\mathbf{c}}}
is a strictly regular block of
\tilde{C}=\tilde{C}_{1}\times\cdots\times\tilde{C}_{h}
with respect
to
\tilde{T}
.
We claim that
N_{\tilde{H}}(\tilde{T})
fixes each of the factors
\tilde{T}_{j}
for
1\leq j\leq h
. This is trivial if
h=1
.
It is also clear if
h=2
and
\tilde{H}=\tilde{G}_{1}\times\tilde{G}_{2}
.
If
h=2
and
\tilde{H}=\tilde{G}
, then
a_{1}=a_{2}=0
and
c^{\prime}0
by assumption, and
then
n_{1}=m_{1}p^{a_{1}},n_{2},n_{3}
are pairwise distinct. Indeed,
n_{2}=n_{3}
would
give
n=m_{1}p^{a_{1}}+2n_{2}
and thus
p\mid n_{2}
, a contradiction. This yields
our claim.
Now
\tilde{{\mathbf{c}}}
contains an irreducible character
\tilde{\psi}
of the form
\tilde{\psi}:=\varepsilon_{\tilde{{\mathbf{T}}}}\varepsilon_{\tilde{{\mathbf{C}}}}R_{\tilde{{\mathbf{T}}}}^{\tilde{{\mathbf{C}}}}(\theta),
for some irreducible character
\theta
of
\tilde{T}
of
p^{\prime}
-order and
in general position with
respect to
\tilde{C}
. By the claim,
\theta
is also in general position with
respect to
\tilde{H}
, and thus
\varepsilon_{\tilde{{\mathbf{T}}}}\varepsilon_{\tilde{{\mathbf{H}}}}R_{\tilde{{\mathbf{T}}}}^{\tilde{{\mathbf{H}}}}(\theta)=\varepsilon_{\tilde{{\mathbf{C}}}}\varepsilon_{\tilde{{\mathbf{H}}}}R_{\tilde{{\mathbf{C}}}}^{\tilde{{\mathbf{H}}}}(\tilde{\psi})
is an irreducible character of
\tilde{H}
. Moreover,
\tilde{\chi}
lies in
\tilde{{\mathbf{d}}}
by
[
CaEn99
, Theorem
2.5
]
. This proves our
assertions.
β
If
h=1
, we have
\bar{{\mathbf{B}}}=\bar{{\mathbf{d}}}
,
{\mathbf{B}}={\mathbf{d}}
, and
\tilde{{\mathbf{B}}}=\tilde{{\mathbf{d}}}
. In this case, we compute
W(\bar{{\mathbf{B}}})
from
W({\mathbf{B}})
using
[
HL25
, Lemma
2.4.2
(c)]
. If
h\geq 2
, then, by
definition,
\bar{H}\leq C_{\bar{G}}(\bar{E})
, where
\bar{E}
denotes the
unique subgroup of
\bar{D}
of order
p
. In this case,
[
HL25
, Remark
2.3.3
]
shows that
W(\bar{{\mathbf{d}}})
can be
computed from the sign sequence
\sigma_{\bar{\chi}}^{[l]}(\bar{t})
,
where
\bar{\chi}
denotes the non-exceptional character of
\bar{{\mathbf{d}}}
, and
l
is defined by
|\bar{D}|=p^{l}
. Namely, if
\Lambda=\{0,\ldots,l-1\}
and
A\subseteq\Lambda\setminus\{0\}
is such that
\sigma_{\bar{\chi}}^{[l]}(\bar{t})=\omega_{\Lambda}(\mathbf{1}_{A})
, then
W(\bar{{\mathbf{d}}})=W_{\bar{D}}(A)
; for the notation see
[
HL25
, Definitions
2.1.1
,
2.1.2
]
.
Lemma 3.25
.
If
h=3
, assume that
a_{1}>0
.
Let
\bar{\chi}
denote the non-exceptional character of
\Irr(\bar{{\mathbf{d}}})
,
and let
\chi
denote the inflation of
\bar{\chi}
to
H
. Moreover, let
\tilde{\chi}\in\Irr(\tilde{{\mathbf{d}}})
be as in
Lemma
3.24
.
Then
(5)
\sigma_{\chi}^{[a+a_{1}]}(t)=\omega_{\tilde{{\mathbf{H}}}}^{[a+a_{1}]}(t)=\sigma_{\tilde{\chi}}^{[a+a_{1}]}(t)
and
(6)
\sigma_{\bar{\chi}}^{[a+a_{1}-c^{\prime}]}(\bar{t})=\omega_{\tilde{{\mathbf{H}}}}^{[a+a_{1}-c^{\prime}]}(t).
(For the notation
\omega_{\tilde{{\mathbf{H}}}}
see
[
HL25
, Definition
2.5.6
]
).
Proof.
By definition,
{\mathbf{d}}
dominates the nilpotent cyclic block
\bar{{\mathbf{d}}}
. By
[
HL25
, Lemma
2.3.4
]
,
the character
\bar{\chi}
is the unique
p
-rational character in
\bar{{\mathbf{d}}}
.
As the irreducible characters of
{\mathbf{d}}
which do not have
Y
in their kernels
are not
p
-rational,
\Irr({\mathbf{d}})
has a unique
p
-rational character,
namely
\chi
, and
\chi
lifts the unique irreducible Brauer character
of
{\mathbf{d}}
. Thus the hypotheses of
[
HL25
, Lemma
2.5.18
]
are satisfied, and
so
\tilde{\chi}
lies above
\chi
.
As
\bar{\chi}
is the non-exceptional character of
\Irr(\bar{{\mathbf{d}}})
, the
values
\bar{\chi}(\bar{u})
are non-zero integers for
\bar{u}\in\bar{D}
;
see
[
HL24
, Lemma
3.3
]
. The same is then true for the values
\chi(u)
for
u\in\langle t\rangle
.
By
[
HL25
, Lemma
2.5.18
]
we have
\sigma_{\tilde{\chi}}(u)=\sigma_{\chi}(u)
for all
u\in\tilde{D}
, which
implies
\sigma_{\tilde{\chi}}^{[a+a_{1}]}(t)=\sigma_{\chi}^{[a+a_{1}]}(t).
By
[
HL25
, Lemma
2.5.7
]
,
\sigma_{\tilde{\chi}}^{[a+a_{1}]}(t)=\omega_{\tilde{{\mathbf{H}}}}^{[a+a_{1}]}(t),
which yields (
5
). This trivially implies
\sigma_{\chi}^{[a+a_{1}-c^{\prime}]}(t)=\omega_{\tilde{{\mathbf{H}}}}^{[a+a_{1}-c^{\prime}]}(t).
Since
\chi
is the inflation of
\bar{\chi}
to
H
, and as the kernel
of the epimorphism
\langle t\rangle\rightarrow\langle\bar{t}\rangle=\bar{D}
has order
p^{c^{\prime}}
, we obtain
\sigma_{\bar{\chi}}^{[a+a_{1}-c^{\prime}]}(\bar{t})=\sigma_{\chi}^{[a+a_{1}-c^{\prime}]}(t),
which gives (
6
).
β
3.26.
Computing the invariants in case
h=1
Assume that the minimal polynomial of
t
is irreducible. Recall the results of
Lemmas
3.16
(c) and
3.23
in this case.
If
D\not\leq Z
, then
n=m_{1}p^{a+a_{1}}
with
p\nmid m_{1}
and
a_{1}>0
.
The defect group
D
is cyclic of order
p^{a+a_{1}}
, and
{\mathbf{B}}
is covered by a
cyclic block
\tilde{{\mathbf{B}}}
of
\tilde{G}
with defect group
\tilde{D}
of
order
p^{2a+a_{1}}
. Moreover,
|\bar{D}|=p^{a+a_{1}-c^{\prime}}
.
Proposition 3.27
.
Suppose that the minimal polynomial of
t
is irreducible. If
D\leq Z
, then
W(\bar{{\mathbf{B}}})\cong k
.
Assume in the following that
D\not\leq Z
, so that
a_{1}\geq 1
by
Lema
3.16
(c)
. Then
W(\bar{{\mathbf{B}}})\cong k
if
\varepsilon=1
, or if
p\equiv 1\,\,(\mbox{\rm mod}\,\,4)
, or if
m_{1}^{\prime}
is even, or if
c^{\prime}=a
and
a_{1}=1
. Otherwise,
W(\bar{{\mathbf{B}}})=W_{\bar{D}}([a-c^{\prime},a+a_{1}-c^{\prime}-1])
, if
c^{\prime}1
.
Proof.
If
D\leq Z
, then
\bar{D}\leq Z(\bar{G})
, and so
W(\bar{{\mathbf{B}}})\cong k
by
[
HL24
, Lemma
3.6
(b)]
. Assume then that
D\not\leq Z
.
Put
l:=a+a_{1}
and
\Lambda:=\{0,\ldots,l-1\}
. Since
t^{p^{l-1}}
has order
p
, we have
t^{p^{l-1}}\in Z(\tilde{H})
. Let
\tilde{\chi}
be
as in Lemma
3.24
. Then
[
HL25
, Lemma
4.1.3
]
yields
\sigma_{\tilde{\chi}}^{[a+a_{1}]}(t)=\omega_{\Lambda}(I)
with
I=\emptyset
, unless
\varepsilon=-1
,
n
is odd, and
p\equiv-1\,\,(\mbox{\rm mod}\,\,4)
, in which case
I=[a,l-1]
. In view
of
[
HL25
, Remark
2.3.3
]
and
Equation (
5
) in Lemma
3.25
, this gives
W({\mathbf{B}})\cong W_{D}(I)
.
Our assertion follows from this and
[
HL25
, Lemma
2.4.2
(c)]
.
β
3.28.
Computing the invariants in case
h=2
We now assume that the minimal polynomial of
t
has exactly two irreducible
factors. Recall the following facts from Lemma
3.16
(d). We
have
n=n_{1}+n_{2}
with
n_{1}=m_{1}p^{a_{1}}
,
n_{2}=m_{2}p^{a_{2}}
for some
non-negative integers
a_{1}\geq a_{2}
, and
p\nmid m_{1}m_{2}
. Also,
c^{\prime}=0
if
a_{1}>a_{2}
, and
c^{\prime}=c-a_{1}
if
a_{1}=a_{2}>0
. Moreover,
|\bar{D}|=p^{l}
with
l=a+a_{1}-c^{\prime}
.
Recall that
V=V_{1}\oplus V_{2}
is the primary decomposition of
V
with
respect to
t
, where
\dim(V_{j})=n_{j}=m_{j}p^{a_{j}}
for
j=1,2
. Recall also
from
Notation
3.2
(ii) that
{\mathbf{V}}=\mathbb{F}^{n}
is the natural vector space
for
\tilde{{\mathbf{G}}}
. For
i=1,2
, let
{\mathbf{V}}_{i}
denote the
\mathbb{F}
-span
of
V_{i}
, and let
\tilde{{\mathbf{G}}}_{i}
denote the subgroup of
\tilde{{\mathbf{G}}}
induced
on
{\mathbf{V}}_{i}
. Then
\tilde{{\mathbf{G}}}_{i}
is
F
-stable and
\tilde{{\mathbf{G}}}_{i}^{F}=\tilde{G}_{i}
for
i=1,2
.
Proposition 3.29
.
Suppose that the minimal polynomial of
t
has exactly two irreducible factors.
Then
W(\bar{{\mathbf{B}}})=W_{\bar{D}}(I)
, with
I\subseteq\{1,\ldots,l-1\}
an interval.
If
l=1
, or if
\varepsilon=1
, or if
a_{1}=a_{2}=0
and
c^{\prime}=c
, then
I=\emptyset
, i.e.
W(\bar{{\mathbf{B}}})\cong k
.
Suppose in the following that
l\geq 2
, that
\varepsilon=-1
, and that
c^{\prime}a_{2}
,
p\equiv-1\,\,(\mbox{\rm mod}\,\,4)
and at
least one of
n_{1}
,
n_{2}
is odd. Then
I=\begin{cases}[a,l-1],&\text{\ if\ }n_{1}\text{\ odd and\ }(n_{2}\text{\ even or\ }a_{2}=0);\\
[a+a_{1}-a_{2},l-1],&\text{\ if\ }n_{1}\text{\ even,\ }n_{2}\text{\ odd, and\ }a_{2}>0;\\
[a,l-a_{2}-1],&\text{\ if\ }n_{1}\text{\ and\ }n_{2}\text{\ odd}.\end{cases}
(b)
We have
a_{1}=a_{2}>0
,
p\equiv-1\,\,(\mbox{\rm mod}\,\,4)
, and
n
is odd. Then
I=[a-c^{\prime},l-1].
(c)
We have
a_{1}=a_{2}=0
,
c^{\prime}a_{2}
or if
a_{1}=a_{2}
and
c^{\prime}=c-a_{1}
, we have
\tilde{{\mathbf{H}}}=\tilde{{\mathbf{G}}}_{1}\times\tilde{{\mathbf{G}}}_{2}
by
Lemma
3.19
(c), and thus
(7)
\omega_{\tilde{{\mathbf{H}}}}^{[l]}(t)=\omega_{\tilde{{\mathbf{G}}}_{1}}^{[l]}(t_{1})\omega_{\tilde{{\mathbf{G}}}_{2}}^{[l]}(t_{2}),
where the two
l
-tuples on the right hand side are multiplied component-wise.
We continue the proof under the assumption that (
7
)
holds. Let
I:=I_{1}\diamond I_{2}
denote the symmetric difference of
I_{1}
and
I_{2}
.
Then
\omega_{\tilde{{\mathbf{G}}}_{1}}^{[l]}(t_{1})\omega_{\tilde{{\mathbf{G}}}_{2}}^{[l]}(t_{2})=\omega_{\Lambda}(I)
; see
[
HL25
, Subsection
2.2
]
.
This yields the assertions in (a) and (b).
Now assume that (
7
) is not satisfied. Then
a_{1}=a_{2}=0
and
c^{\prime}0
.
Recall that
V=V_{1}\oplus V_{2}\oplus V_{3}
is the primary decomposition of
V
with respect to
t
, where
\dim(V_{j})=n_{j}
for
j=1,2,3
. Recall also from
Notation
3.2
(ii) that
{\mathbf{V}}=\mathbb{F}^{n}
is the natural vector space
for
\tilde{{\mathbf{G}}}
. For
i=1,2,3
, let
{\mathbf{V}}_{i}
denote the
\mathbb{F}
-span
of
V_{i}
, and let
\tilde{{\mathbf{G}}}_{i}
denote the subgroup of
\tilde{{\mathbf{G}}}
induced
on
{\mathbf{V}}_{i}
. Put
{\mathbf{V}}_{2,3}:={\mathbf{V}}_{2}\oplus{\mathbf{V}}_{3}
, and let
\tilde{{\mathbf{G}}}_{2,3}
denote the subgroup of
\tilde{{\mathbf{G}}}
induced on
{\mathbf{V}}_{2,3}
.
Then
\tilde{{\mathbf{G}}}_{i}
is
F
-stable, and
\tilde{{\mathbf{G}}}_{i}^{F}=\tilde{G}_{i}
for
i=1,2,3
. Similarly,
\tilde{{\mathbf{G}}}_{2,3}
is
F
-stable and
\tilde{{\mathbf{G}}}_{2,3}^{F}=\tilde{G}_{2,3}
in the notation of
Lemma
3.19
(c)(iii).
Proposition 3.31
.
Suppose that the minimal polynomial of
t
has exactly three irreducible
factors. Then
W(\bar{{\mathbf{B}}})=W_{D}(A)
, with
A\subseteq\{1,\ldots,l-1\}
a union of two intervals. If
a_{1}=0
or
if
\varepsilon=1
, then
A=\emptyset
, i.e.
W(\bar{{\mathbf{B}}})\cong k
.
Suppose in the following that
a_{1}>0
and
\varepsilon=-1
. Then
A
is
non-empty exactly in the following cases.
(a)
At least one of
n_{2}
,
n_{3}
is even,
n_{1}
is odd and
p\equiv-1\,\,(\mbox{\rm mod}\,\,4)
. Then
A=[a,l-1].
(b)
Each of
n_{2}
and
n_{3}
is odd and
a_{1}\geq a
. Then
A=\begin{cases}\{a_{1}\},&\text{if\ }n_{1}\text{\ is even or\ }p\equiv 1\,\,(\mbox{\rm mod}\,\,4);\\
[a,l-1]\setminus\{a_{1}\},&\text{if\ }n_{1}\text{\ is odd and\ }p\equiv-1\,\,(\mbox{\rm mod}\,\,4).\end{cases}
(c)
Each of
n_{2}
and
n_{3}
is odd and
a_{1}0
. Set
t_{2,3}:=t_{2}t_{3}
and
\Lambda:=\{0,1,\ldots,l-1\}
. Since
C_{\tilde{{\mathbf{H}}}}(t)=C_{\tilde{{\mathbf{G}}}_{1}}(t_{1})\times C_{\tilde{{\mathbf{G}}}_{2,3}}(t_{2,3})
by
Lemma
3.19
(d), we obtain
\omega_{\tilde{{\mathbf{H}}}}^{[l]}(t)=\omega_{\tilde{{\mathbf{G}}}_{1}}^{[l]}(t_{1})\omega_{\tilde{{\mathbf{G}}}_{2,3}}^{[l]}(t_{2,3}),
where the multiplication of the
l
-tuples on the right hand side is defined
component-wise.
By
[
HL25
, Corollary
4.1.4
]
we have
\omega_{\tilde{{\mathbf{G}}}_{1}}^{[l]}(t_{1})=\omega_{\Lambda}(\mathbf{1}_{I_{1}})
with
I_{1}=\emptyset
unless
n_{1}
is odd,
\varepsilon=-1
and
p\equiv-1\,\,(\mbox{\rm mod}\,\,4)
, in which case
I_{1}=[a,l-1]
.
Let
1\leq j\leq l
and put
u:=t_{2,3}^{p^{l-j}}
.
Lemma
3.16
(e) implies that
C_{\tilde{{\mathbf{G}}}_{2,3}}(u)=\tilde{{\mathbf{G}}}_{2,3}
if
j\leq a_{1}
, and
C_{\tilde{{\mathbf{G}}}_{2,3}}(u)=\tilde{{\mathbf{G}}}_{2}\times\tilde{{\mathbf{G}}}_{3}
if
a_{1}+1\leq j\leq a+a_{1}
. In the former case,
\omega_{\tilde{{\mathbf{G}}}_{2,3}}(u)=1
. In the latter case,
\omega_{\tilde{{\mathbf{G}}}_{2,3}}(u)=\varepsilon_{\tilde{{\mathbf{G}}}_{2,3}}\varepsilon_{\tilde{{\mathbf{G}}}_{2}}\varepsilon_{\tilde{{\mathbf{G}}}_{3}}.
By
[
HL25
, Example
2.5.5
]
, we get
\varepsilon_{\tilde{{\mathbf{G}}}_{2,3}}\varepsilon_{\tilde{{\mathbf{G}}}_{2}}\varepsilon_{\tilde{{\mathbf{G}}}_{3}}=\begin{cases}-1,&\text{\rm if}\ \varepsilon=-1\text{\rm\ and\ }n_{2},n_{3}\text{\rm\ odd};\\
1,&\text{\rm otherwise}.\end{cases}
It follows that
\omega_{\tilde{{\mathbf{G}}}_{2,3}}^{[l]}(t_{2,3})=\omega_{\Lambda}(\mathbf{1}_{I_{2,3}})
with
I_{2,3}=\{a_{1}\}
,
if
n_{2}
and
n_{3}
are odd and
\varepsilon=-1
, and
I_{2,3}=\emptyset
,
otherwise.
We conclude from the considerations in
[
HL25
, Subsection
2.2
]
that
\omega_{\tilde{{\mathbf{G}}}_{1}}^{[l]}(t_{1})\omega_{\tilde{{\mathbf{G}}}_{2,3}}^{[l]}(t_{2,3})=\omega_{\Lambda}^{[l]}(\mathbf{1}_{A})
with
A=I_{1}\diamond I_{2,3}
,
the symmetric difference of
I_{1}
and
I_{2,3}
. This yields our assertions.
β
We record a specific corollary of the above results.
If
\tilde{T}
is a maximal torus of
\tilde{G}
of type
\pi
,
we write
T_{\pi}:=\tilde{T}\cap G
. As every maximal torus of
\text{\rm SL}_{n}^{\varepsilon}(q)
is of this form, this yields a labelling of
the maximal tori of
\text{\rm SL}_{n}^{\varepsilon}(q)
(up to conjugation) by
partitions of
n
.
Corollary 3.32
.
Suppose that
p=3
and
n\in\{3,6,9\}
or that
p=5
and
n=5
.
Let
{\mathbf{B}}
be a
p
-block of
G=\text{\rm SL}_{n}^{\varepsilon}(q)
with a non-trivial cyclic defect group
D
.
If
n=9
assume that
D\not\leq Z
and that
a\geq 2
.
Then the following statements hold.
(a)
We have
|D|=p
or
|D|=p^{a}
. Also,
W({\mathbf{B}})\cong k
or
W({\mathbf{B}})\cong W_{D}(\{1\})
. The latter occurs
exactly if
\varepsilon=-1
,
|D|=3^{a}
with
a\geq 2
, and
n=6
In this case, we have
h=2
and
\{n_{1},n_{2}\}=\{5,1\}
in the notation of
Proposition
3.29
.
(b)
Suppose that
p=3
and
n=6
. If
W({\mathbf{B}})\not\cong k
,
then
{\mathbf{B}}
is strictly regular with respect to
T_{(5,1)}
.
Proof.
(a) In the setup of Notation
3.7
we have
Y=\{1\}
, hence
\bar{{\mathbf{B}}}={\mathbf{B}}
and
c^{\prime}=0
. Also, the group
D
of
Notation
3.7
(ii) is cyclic, a defect group of
{\mathbf{B}}
, and
|D|=|t|=p^{a+a_{1}}
, the latter by Lemma
3.16
(f).
In particular, we cannot have
h=3
, since in that case
Y=O_{p}(Z)\neq\{1\}
.
If
D\leq Z
, then
W({\mathbf{B}})\cong k
by
[
HL24
, Lemma
3.6
(b)]
.
Moreover,
n\neq 9
by hypothesis, and hence
|D|=p
. Assume then
that
D
is non-central in the following. Suppose that
h=1
. Then
p^{a+a_{1}}\mid n
with
a_{1}\geq 1
by Lemma
3.16
(c).
As we have assumed that
a\geq 2
if
n=9
, this case cannot occur.
Suppose finally that
h=2
. As
D
is cyclic,
c>0
, and
|D|=p^{a+a_{1}}
,
we must have
a_{1}=a_{2}=0
by Lemma
3.16
(d)(i)(ii). In
particular,
|D|=p^{a}
. Proposition
3.29
shows that
W({\mathbf{B}})\cong k
, unless
\varepsilon=-1
,
a\geq 2
, and
n_{1}
and
n_{2}
are odd. If all these latter conditions are satisfied,
n=6
and
\{n_{1},n_{2}\}=\{5,1\}
since
3\nmid n_{1}n_{2}
. Moreover,
W({\mathbf{B}})\cong W_{D}(\{c-c^{\prime}\})=W_{D}(\{1\})
by
Proposition
3.29
(c).
(b) Suppose that
p=3
,
n=6
and
W({\mathbf{B}})\not\cong k
. Without loss of
generality we may assume that
n_{1}=5
and
n_{2}=1
. In the notation of
Lemma
3.15
(e), we have
\tilde{C}_{1}=\tilde{G}_{1}=\text{\rm GU}_{5}(q)
and
\tilde{C}_{2}=\tilde{G}_{2}=\text{\rm GU}_{1}(q)
. Moreover,
D\leq\tilde{D}=\tilde{D}_{1}\times\tilde{D}_{2}
, where
\tilde{D}_{j}
is a cyclic defect group of
a block of
\tilde{C}_{j}
,
j=1,2
. In
particular,
\tilde{D}
is a Sylow
3
-subgroup of a maximal torus
\tilde{T}\leq\tilde{G}
of type
(5,1)
; see
[
HL25
, Corollary
3.6.2
]
.
Let
\chi\in\Irr({\mathbf{B}})
. Then
\chi
has height
0
, and thus
\chi(1)_{3}=3^{4a+2}
, as
|G|_{3}=3^{5a+2}
. The character degrees of
G
given
in
[
Lue21a
]
show that
\chi(1)=[G\colon\!T]_{r^{\prime}}/f
with
T=T_{(4,1,1)}
and
f\in\{1,2\}
, or
T=T_{(5,1)}
and
f=1
. Hence
{\mathbf{B}}
is regular
with respect to one of these tori. The former case cannot occur, as the torus
T_{(4,1,1)}
does not have a cyclic Sylow
3
-subgroup. Hence
{\mathbf{B}}
is strictly
regular with respect to
T_{(5,1)}
.
β
Remark 3.33
.
Suppose that
p=3
and
n=9
, and that
a\geq 2
, so that
c=2
.
Let
Y\leq Z
with
|Y|=3
, and let
\bar{{\mathbf{B}}}
be a
3
-block of
\bar{G}=G/Y
with a non-central, cyclic defect group
\bar{D}=D/Y
. Then
h=2
and
a_{1}=a_{2}=0
by Lemma
3.16
(c)(d)(e).
Let
{\mathbf{B}}
be a block of
G
with defect group
D
dominating
\bar{{\mathbf{B}}}
. Then
D\not\leq Z
by hypothesis on
\bar{D}
. In particular,
{\mathbf{B}}
is as in
Corollary
3.32
, so that
D
is cyclic of order
3^{a}
. Thus
c^{\prime}=1
and
|\bar{D}|=3^{a-1}
. By Corollary
3.32
(a) and
[
HL25
, Lemma
2.4.2
(c)]
, we have
W(\bar{{\mathbf{B}}})\cong k
.
4.
Synthesis
We now investigate which of the parameter sets exhibited in
Section
3
correspond to blocks.
Throughout this section, we fix an odd prime
p
.
Recall that we have fixed a sign
\varepsilon\in\{-1,1\}
. As in
Notation
3.2
(iv), we let
\delta=1
, if
\varepsilon=1
, and
\delta=2
, if
\varepsilon=-1
. To construct specific examples, we will
vary the parameters,
r
,
q
and
n
, keeping their principal significance.
Thus
q
is a power of the prime
r
, where
r\neq p
, and
n
is a positive
integer. Whenever we have chosen
r
,
q
and
n
, we adopt the corresponding
notation introduced in Section
3
.
In particular,
\mathbb{F}
denotes an
algebraic closure of the finite field with
r
elements. Moreover,
\tilde{{\mathbf{G}}}=\text{\rm GL}_{n}(\mathbb{F})
,
\tilde{G}=\text{\rm GL}_{n}^{\varepsilon}(q)=\tilde{{\mathbf{G}}}^{F}
,
G=\text{\rm SL}_{n}^{\varepsilon}(q)=\tilde{{\mathbf{G}}}^{F}
and
Z=Z(G)
, where
F
is as in Notation
3.2
(iii). These definitions
are in accordance with Notation
3.2
, except that we also allow
n=1
here. Furthermore, we put
{\mathbf{G}}^{*}:=\tilde{{\mathbf{G}}}/Z(\tilde{{\mathbf{G}}})=\text{\rm PGL}_{n}(\mathbb{F})
,
with the Steinberg morphism induced from the one on
\tilde{{\mathbf{G}}}
. Then
G^{*}=\tilde{G}/Z(\tilde{G})=\text{\rm PGL}_{n}(q)
. Note that the inclusion
i:{\mathbf{G}}\rightarrow\tilde{{\mathbf{G}}}
is a regular embedding and that the dual
epimorphism
i^{*}:\tilde{{\mathbf{G}}}^{*}\rightarrow{\mathbf{G}}^{*}
is just the canonical
map, if we identify
\tilde{{\mathbf{G}}}
with its dual group
\tilde{{\mathbf{G}}}^{*}
.
Our principal aim is to prove Theorem
1.2
.
We begin by constructing suitable prime powers
q
.
Lemma 4.1
.
Let
a
be a positive integer. Then there is a prime
r
and a power
q
of
r
such that
p^{a}\mid q-\varepsilon
and
p^{a+1}\nmid q-\varepsilon
. If
p^{a}=3
, there is such a
q
with
q>2
.
Proof.
Let
r
be a prime such that
p\mid r-\varepsilon
but
p^{2}\nmid r-\varepsilon
. Then put
q:=r^{p^{a-1}}
. The last statement
is trivial.
β
The following result considers the situation of
Lemma
3.16
(c).
Proposition 4.2
.
Let
a,a_{1},c^{\prime}
be integers with
a,a_{1}
positive
and
0\leq c^{\prime}\leq a
. Let
q
satisfy the conclusion of
Lemma
4.1
with respect to
p
and
a
, and put
n:=p^{a+a_{1}}
. Then there is a
block
{\mathbf{B}}
of
G=\text{\rm SL}^{\varepsilon}_{n}(q)
such that the following
statements hold.
(a)
The defect group
D
of
{\mathbf{B}}
is cyclic of order
p^{a+a_{1}}
.
(b)
There is a cyclic block of
\text{\rm GL}^{\varepsilon}_{n}(q)
covering
{\mathbf{B}}
.
(c)
There is a subgroup
Y\leq Z
with
|Y|=p^{c^{\prime}}
, and a
block
\bar{{\mathbf{B}}}
of
G/Y
dominated by
{\mathbf{B}}
and with defect group
D/Y
.
Proof.
Let
\tilde{{\mathbf{T}}}
denote a Coxeter torus of
\tilde{{\mathbf{G}}}
; thus
\tilde{T}
is a
cyclic group of order
q^{n}-\varepsilon
(notice that
n
is odd and
n\geq 3
). Put
{\mathbf{T}}:=\tilde{{\mathbf{T}}}\cap{\mathbf{G}}
. By
[
HL25
, Lemma
3.6.3
]
there
exists
\tilde{s}\in\tilde{T}
, whose image in
G^{*}
is strictly regular.
Let
\tilde{\chi}
denote the corresponding irreducible Deligne-Lusztig
character of
\tilde{G}
and let
\tilde{{\mathbf{B}}}
denote the
p
-block
of
\tilde{G}
containing
\tilde{\chi}
. By
[
HL25
, Lemma
2.5.16
]
,
a Sylow
p
-subgroup
\tilde{D}
of
\tilde{T}
is a defect group of
\tilde{{\mathbf{B}}}
. In particular,
\tilde{D}
is cyclic of order
p^{2a+a_{1}}
.
As
s\in G^{*}
is strictly regular,
[
HL25
, Corollary
2.5.17
]
implies that
\chi:=\Res^{\tilde{G}}_{G}(\tilde{\chi})
is
irreducible,
\tilde{{\mathbf{B}}}
covers a unique block
{\mathbf{B}}
of
G
and that
D:=\tilde{D}\cap G
is a defect group of
{\mathbf{B}}
. Clearly,
D
is cyclic of
order
p^{a+a_{1}}
. Let
Y\leq G
with
|Y|=p^{c^{\prime}}
. The
p
-block
\bar{{\mathbf{B}}}
of
G/Y
dominated by
{\mathbf{B}}
has defect group
D/Y
. This concludes our proof.
β
Proposition
4.2
shows that all possible parameters determined
in Lemma
3.16
(c) arise from blocks.
We next investigate the situation of Lemma
3.16
(d), where
we restrict to the cases of Proposition
3.29
(a) with
n_{1}
,
n_{2}
odd, and Proposition
3.29
(b) with
n_{1}
even and
n_{2}
odd.
Proposition 4.3
.
Let
a,a_{1},a_{2}
be integers with
a,a_{1}
positive and
0\leq a_{2}\leq a,a_{1}
. If
a_{1}>a_{2}
put
c:=a_{2}
. If
a_{1}=a_{2}
, choose
an integer
c
with
a_{1}\leq c\leq a
. If
a_{1}=a_{2}=c
, let
p>3
.
Let
q
satisfy the conclusion of
Lemma
4.1
with respect
to
p
and
a
. If
p^{a}=3
, assume that
q>2
. Put
m_{2}:=1
. If
a_{1}>a_{2}
, let
m_{1}:=1
. If
a_{1}=a_{2}
, let
m_{1}:=p^{c-a_{1}}-1
unless
c=a_{1}
; in the latter case, let
m_{1}:=2
if
p\neq 3
, and
m_{2}:=4
, if
p=3
. Finally, let
n:=n_{1}+n_{2}
with
n_{j}:=m_{j}p^{a_{j}}
for
j=1,2
.
Let
Y:=O_{p}(Z)
. Then
|Y|=p^{c}
and there is a block
\bar{{\mathbf{B}}}
of
\bar{G}=G/Y
with cyclic defect group of order
p^{a+a_{1}+a_{2}-c}
.
Moreover, the block
{\mathbf{B}}
of
G
dominating
\bar{{\mathbf{B}}}
has abelian defect
groups, which are direct products of two cyclic groups of order
p^{a+a_{1}}
and
p^{a_{2}}
.
Proof.
Notice that
n=p^{a_{1}}+p^{a_{2}}
if
a_{1}>a_{2}
, and
n=p^{c}
if
a_{1}=a_{2}3
, and
n=5p^{a_{2}}
if
p=3
. In any case,
|O_{p}(Z)|=p^{c}
.
Let
V
denote the natural vector space of
\tilde{G}
. Write
V=V_{1}\oplus V_{2}
, an orthogonal decomposition into non-degenerate subspaces
if
\varepsilon=-1
, with
\dim V_{j}=n_{j}
for
j=1,2
. Fix
j\in\{1,2\}
. Let
\tilde{G}_{j}
denote the subgroup of
\tilde{G}
induced
on
V_{j}
, so that
\tilde{G}_{j}\cong\text{\rm GL}_{n_{j}}^{\varepsilon}(q)
. Choose a
maximal torus
\tilde{T}_{j}\leq\tilde{G}_{j}
such that
\tilde{T}_{j}
is
cyclic of order
q^{n_{j}}-\varepsilon^{n_{j}}
. Let
\tilde{D}_{j}
denote the
Sylow
p
-subgroup of
\tilde{T}_{j}
; then
\tilde{D}_{j}
is cyclic of
order
p^{a+a_{j}}
.
Put
\tilde{D}=\tilde{D}_{1}\times\tilde{D}_{2}
and
\tilde{T}:=\tilde{T_{1}}\times\tilde{T}_{2}\leq\tilde{G}_{1}\times\tilde{G}_{2}\leq\tilde{G}
. Then
\tilde{T}
is a maximal
torus of
\tilde{G}
corresponding to the partition of
n
with parts
n_{1},n_{2}
. If
n_{2}=1
, put
f_{2}=1
and
\tilde{s}_{2}=1
. Otherwise,
n_{j}\geq 3
for
j=1,2
, and the torus
\tilde{T}_{j}
contains a
p^{\prime}
-element
\tilde{s}_{j}
of prime order
f_{j}
, such that
f_{j}\nmid q^{i}-\varepsilon^{i}
for all
1\leq ia_{2}
, then
\langle u_{1}u_{2}^{e}\rangle\cap Y=\{1\}
,
since the non-trivial eigenvalues of
u_{1}^{p^{a_{1}}}
and
u_{2}^{ep^{a_{1}}}
have
distinct orders. As
|D|=p^{a+a_{1}+a_{2}}
and
a_{2}=c
, we get
D=\langle u_{1}u_{2}^{e}\rangle\times Y
. In particular,
\bar{D}
is cyclic.
If
a_{1}=a_{2}
, we have
Y=\langle(u_{1}u_{2})^{p^{a+a_{1}-c}}\rangle
by our
choice of
u_{1}
and
u_{2}
. The Frattini subgroup
\Phi(D)
of
D
is generated
by
(u_{1}u_{2}^{e})^{p}
and
u_{2}^{p^{a+1}}
as a direct product. An elementary
calculation shows that
(u_{1}u_{2})^{p^{a+a_{1}-c}}
is not contained in
\Phi(D)
.
Indeed,
u_{1}u_{2}=(u_{1}u_{2}^{e})u_{2}^{1-e}
, and thus
(u_{1}u_{2})^{p^{a+a_{1}-c}}\in\langle(u_{1}u_{2}^{e})^{p}\rangle\times\langle u_{2}^{p^{a+1}}\rangle
would imply
that
u_{2}^{fp^{a+1}}=u_{2}^{(1-e)p^{a+a_{1}-c}}=u_{2}^{(1+m_{1})p^{a+a_{1}-c}}
for some integer
f
. However, the
p
-part of
1+m_{1}
equals
p^{c-a_{1}}
.
Hence
|u_{2}^{(1+m_{1})p^{a+a_{1}-c}}|=|u_{2}^{p^{a}}|>|u_{2}^{fp^{a+1}}|,
a contradiction.
As
D
is a
2
-generator group, this implies that
\bar{D}
is cyclic.
Let
\bar{{\mathbf{B}}}
denote
the block of
\bar{G}
dominated by
{\mathbf{B}}
. The defect group
of
\bar{{\mathbf{B}}}
equals
\bar{D}
, which proves our assertions.
β
We next investigate the situation of Lemma
3.16
(d)(iii).
Proposition 4.4
.
Let
p
be an odd prime and let
c^{\prime},c,a
be integers with
0\leq c^{\prime}\leq c0
in the
following. Put
n_{1}:=n-1
and
n_{2}:=1
. Let
G_{1}:=\text{\rm GL}_{n_{1}}^{\varepsilon}(q)
, naturally embedded into
G
. Let
D:=O_{p}(Z(G_{1}))
. Then
D
is cyclic of order
p^{a}
. As
c2
, a cyclic maximal torus of
G_{1}
of order
q^{n_{1}}-\varepsilon^{n_{1}}
contains
p^{\prime}
-elements which are regular with
respect to
G_{1}
; see
[
HL25
, Lemma
3.6.3
]
.
In this case,
[
HL25
, Lemma
2.5.16
]
guarantees the existence of a block
{\mathbf{b}}
of
G_{1}
with defect group
D
. The same conclusion clearly also holds for
n_{1}=2
. Now
N_{G}(G_{1})=G_{1}
, and thus the Brauer correspondent
{\mathbf{B}}
of
{\mathbf{b}}
satisfies (a).
The proof of (b) is analogous to the proof of (c) of
Proposition
4.2
.
β
We finally investigate the situation of Lemma
3.16
(e),
where we restrict to the cases of Proposition
3.31
(b)(c), also
assuming that
a\neq a_{1}
. (If
a=a_{1}
in Proposition
3.31
(b),
the resulting set
A
is an interval.)
Proposition 4.5
.
Let
p
be an odd prime and let
a,a_{1}
be positive integers with
a\neq a_{1}
.
Let
q
satisfy the conclusion of
Lemma
4.1
with respect
to
p
and
a
. Put
n_{1}:=p^{a_{1}}
and
n_{3}:=1
. If
a_{1}>a
, put
n_{2}:=2p^{a}-1
, and if
a_{1}0
and
c=\min\{a,b\}
;
\bullet
{\mathbf{V}}:=\mathbb{F}^{n}
is the natural vector space for
\tilde{{\mathbf{G}}}
,
and
V:=\mathbb{F}_{q^{\delta}}^{n}\subseteq{\mathbf{V}}
;
\bullet
{\mathbf{G}}^{*}:=\tilde{{\mathbf{G}}}/Z(\tilde{{\mathbf{G}}})=\text{\rm PGL}_{n}(\mathbb{F})
(with the Steinberg morphism induced from the one on
\tilde{{\mathbf{G}}}
);
\bullet
G^{*}:=\tilde{G}/Z(\tilde{G})=\text{\rm PGL}_{n}(q)
;
\bullet
Y\leq Z
is a
p
-subgroup and
\bar{G}:=G/Y
is a central quotient
of
G
.
The blocks and the associated local configuration (Section
3
)
:
\bullet
\bar{{\mathbf{B}}}
is a cyclic
p
-block of
\bar{G}
and
{\mathbf{B}}
is the unique
block of
G
dominating
\bar{{\mathbf{B}}}
;
\bullet
D
is a defect group of
{\mathbf{B}}
such that
\bar{D}:=D/Y
is
a defect group of
\bar{{\mathbf{B}}}
;
\bullet
t\in D
is such that
\bar{t}:=tY\in\bar{G}
generates
\bar{D}
;
\bullet
h\in\{1,2,3\}
is the number of irreducible factors of the minimal
polynomial of
t
acting on
V
;
\bullet
c^{\prime}
is the non-negative integer such that
|\bar{D}|=|t|/p^{c^{\prime}}
.
\bullet
C:=C_{G}(D)
and
\tilde{C}:=C_{\tilde{G}}(D)
;
\bullet
{\mathbf{c}}
is a Brauer correspondent of
{\mathbf{B}}
in
C
and
\tilde{{\mathbf{c}}}
is a block of
\tilde{C}
covering
{\mathbf{c}}
; see Figure
1
;
\bullet
\tilde{D}
is a defect group of
\tilde{{\mathbf{c}}}
with
D=C\cap\tilde{D}=G\cap\tilde{D}
.