This article defines a replacement theorem View a complete list of replacement theorems | View a complete list of failures of replacement
Statement
Suppose
P
{\displaystyle P}
is a group of prime power order (we'll call the prime
p
{\displaystyle p}
).
Let
A
(
P
)
{\displaystyle {\mathcal {A}}(P)}
denote the set of all abelian subgroups of maximum order in
P
{\displaystyle P}
(i.e.,
|
A
|
≥
|
B
|
{\displaystyle |A|\geq |B|}
for all abelian subgroups
B
{\displaystyle B}
of
P
{\displaystyle P}
).
Suppose
A
∈
A
(
P
)
{\displaystyle A\in {\mathcal {A}}(P)}
and
B
{\displaystyle B}
is an abelian subgroup such that
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
but
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
. Then, there exists an abelian subgroup
A
∗
{\displaystyle A^{*}}
of
A
B
{\displaystyle AB}
such that:
A
∗
∈
A
(
P
)
{\displaystyle A^{*}\in {\mathcal {A}}(P)}
.
A
∩
B
{\displaystyle A\cap B}
is a proper subgroup of
A
∗
∩
B
{\displaystyle A^{*}\cap B}
.
A
∗
{\displaystyle A^{*}}
is contained in
⟨
A
B
⟩
=
⟨
A
A
B
⟩
{\displaystyle \langle A^{B}\rangle =\langle A^{AB}\rangle }
, i.e., it is contained in the normal closure of
A
{\displaystyle A}
in
A
B
{\displaystyle AB}
. In particular, it is contained in the normal closure
⟨
A
P
⟩
{\displaystyle \langle A^{P}\rangle }
of
A
{\displaystyle A}
in
P
{\displaystyle P}
.
A
∗
{\displaystyle A^{*}}
normalizes
A
{\displaystyle A}
.
We have the following additional conclusions:
A
∗
{\displaystyle A^{*}}
normalizes
B
{\displaystyle B}
: Note that since
A
∗
{\displaystyle A^{*}}
is contained in
A
B
{\displaystyle AB}
, and
A
B
≤
N
P
(
B
)
{\displaystyle AB\leq N_{P}(B)}
by assumption, we see that
A
∗
{\displaystyle A^{*}}
normalizes
B
{\displaystyle B}
as well.
A
∗
∩
B
{\displaystyle A^{*}\cap B}
is a proper subgroup of
B
{\displaystyle B}
: This is because
A
∗
{\displaystyle A^{*}}
cannot contain
B
{\displaystyle B}
, since
A
∗
{\displaystyle A^{*}}
normalizes
A
{\displaystyle A}
but
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
.
Thus, all the conditions assumed for
A
{\displaystyle A}
also hold for
A
∗
{\displaystyle A^{*}}
, except possibly the fact that
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
. Hence, the term replacement .
Related facts
Similar replacement theorems
For a complete list of replacement theorems, refer:
Category:Replacement theorems
Applications
Facts used
Thompson's lemma on product with centralizer of commutator with abelian subgroup of maximum order
Nilpotent implies center is normality-large
Normality satisfies image condition
Proof
This proof uses a tabular format for presentation. Provide feedback on tabular proof formats in a survey (opens in new window/tab) | Learn more about tabular proof formats |View all pages on facts with proofs in tabular format
Given : A finite
p
{\displaystyle p}
-group
P
{\displaystyle P}
.
A
(
P
)
{\displaystyle {\mathcal {A}}(P)}
is the set of abelian subgroups of
P
{\displaystyle P}
of maximum order.
A
∈
A
(
P
)
{\displaystyle A\in {\mathcal {A}}(P)}
, and
B
{\displaystyle B}
is a subgroup of
A
{\displaystyle A}
such that
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
but
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
.
To prove : There exists
A
∗
∈
A
(
P
)
{\displaystyle A^{*}\in {\mathcal {A}}(P)}
such that
A
∩
B
{\displaystyle A\cap B}
is a proper subgroup of
A
∗
∩
B
{\displaystyle A^{*}\cap B}
,
A
∗
{\displaystyle A^{*}}
normalizes
A
{\displaystyle A}
and
B
{\displaystyle B}
, and
A
∗
{\displaystyle A^{*}}
is contained in
A
B
{\displaystyle AB}
.
Proof : We let
C
=
A
B
{\displaystyle C=AB}
and
N
=
N
B
(
A
)
{\displaystyle N=N_{B}(A)}
.
basic diagram of key initial constructions (steps 1-6)
More detailed diagram
Step no.
Assertion
Facts used
Given data used
Proof steps used
Explanation
1
C
=
A
B
{\displaystyle C=AB}
is a subgroup of
P
{\displaystyle P}
, and
B
{\displaystyle B}
is normal in
C
{\displaystyle C}
--
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
--
[SHOW MORE]
C
=
A
B
{\displaystyle C=AB}
is a subgroup since
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
. Since both
A
{\displaystyle A}
and
B
{\displaystyle B}
normalize
B
{\displaystyle B}
,
C
=
A
B
{\displaystyle C=AB}
also normalizes
B
{\displaystyle B}
, so
B
{\displaystyle B}
is normal in
C
{\displaystyle C}
.
2
B
{\displaystyle B}
normalizes
N
{\displaystyle N}
--
B
{\displaystyle B}
is abelian
--
[SHOW MORE] Since
N
≤
B
{\displaystyle N\leq B}
by definition, and
B
{\displaystyle B}
is abelian, so
N
{\displaystyle N}
is normal in
B
{\displaystyle B}
. Thus,
B
{\displaystyle B}
normalizes
N
{\displaystyle N}
.
3
A
{\displaystyle A}
normalizes
N
{\displaystyle N}
--
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
--
[SHOW MORE]
A
≤
N
P
(
A
)
{\displaystyle A\leq N_{P}(A)}
, so
A
{\displaystyle A}
normalizes
N
P
(
A
)
{\displaystyle N_{P}(A)}
. Also,
A
{\displaystyle A}
normalizes
B
{\displaystyle B}
. Thus,
A
{\displaystyle A}
normalizes the intersection
N
P
(
A
)
∩
B
=
N
B
(
A
)
{\displaystyle N_{P}(A)\cap B=N_{B}(A)}
.
4
N
=
N
B
(
A
)
{\displaystyle N=N_{B}(A)}
is normal in
C
{\displaystyle C}
--
--
steps (2), (3)
[SHOW MORE] Both
A
{\displaystyle A}
and
B
{\displaystyle B}
normalize
N
{\displaystyle N}
, so
C
=
A
B
{\displaystyle C=AB}
also normalizes
N
{\displaystyle N}
.
5
N
{\displaystyle N}
is a proper subgroup of
B
{\displaystyle B}
--
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
[SHOW MORE] By definition,
N
=
N
B
(
A
)
{\displaystyle N=N_{B}(A)}
normalizes
A
{\displaystyle A}
. If
N
=
B
{\displaystyle N=B}
, we'd get that
B
{\displaystyle B}
normalizes
A
{\displaystyle A}
, contrary to assumption.
6
B
/
N
{\displaystyle B/N}
is a nontrivial normal subgroup of
C
/
N
{\displaystyle C/N}
fact (3)
--
steps (1), (4), (5)
[SHOW MORE]
C
/
N
{\displaystyle C/N}
makes sense by step (4).
B
/
N
{\displaystyle B/N}
is normal in it because
B
{\displaystyle B}
is normal in
C
{\displaystyle C}
(step (1)). Finally,
B
/
N
{\displaystyle B/N}
is nontrivial because
N
{\displaystyle N}
is a proper subgroup of
B
{\displaystyle B}
(step (3)).
7
B
/
N
∩
Z
(
C
/
N
)
{\displaystyle B/N\cap Z(C/N)}
is nontrivial
fact (2)
--
step (6)
Direct combination of the facts
8
Let
x
∈
P
{\displaystyle x\in P}
be such that its image mod
N
{\displaystyle N}
is a nontrivial element in
B
/
N
∩
Z
(
C
/
N
)
{\displaystyle B/N\cap Z(C/N)}
. Then
M
=
[
x
,
A
]
{\displaystyle M=[x,A]}
is contained in
N
{\displaystyle N}
and hence in
B
{\displaystyle B}
step (7)
Note that step (7) guarantees the existence of such
x
{\displaystyle x}
. [SHOW MORE] Mod
N
{\displaystyle N}
,
x
{\displaystyle x}
is in the center of
C
/
N
{\displaystyle C/N}
, so
[
x
,
A
]
{\displaystyle [x,A]}
is trivial mod
N
{\displaystyle N}
. Thus,
M
=
[
x
,
A
]
≤
N
{\displaystyle M=[x,A]\leq N}
. Since
N
≤
B
{\displaystyle N\leq B}
,
M
{\displaystyle M}
is contained in
B
{\displaystyle B}
.
9
M
=
[
x
,
A
]
{\displaystyle M=[x,A]}
is an abelian subgroup of
P
{\displaystyle P}
B
{\displaystyle B}
is abelian
step (8)
[SHOW MORE] By the previous step,
M
≤
B
{\displaystyle M\leq B}
by assumption, with
B
{\displaystyle B}
abelian. Thus,
M
{\displaystyle M}
, being a subgroup of an abelian group, is abelian.
10
A
∗
=
M
C
A
(
M
)
{\displaystyle A^{*}=MC_{A}(M)}
is an abelian subgroup of maximum order in
P
{\displaystyle P}
fact (1)
A
{\displaystyle A}
is abelian of maximum order
step (9)
Piece together.
11
A
∩
B
{\displaystyle A\cap B}
is contained in
C
A
(
M
)
{\displaystyle C_{A}(M)}
B
{\displaystyle B}
is abelian
step (8)
[SHOW MORE] Any element of
A
∩
B
{\displaystyle A\cap B}
centralizes
B
{\displaystyle B}
(since
B
{\displaystyle B}
is abelian). By step (8),
M
{\displaystyle M}
is contained in
B
{\displaystyle B}
, so
A
∩
B
{\displaystyle A\cap B}
centralizes
M
{\displaystyle M}
. Also, this element is in
A
{\displaystyle A}
, so it is in
C
A
(
M
)
{\displaystyle C_{A}(M)}
.
12
A
∩
B
{\displaystyle A\cap B}
is contained in
A
∗
∩
B
{\displaystyle A^{*}\cap B}
--
step (11)
[SHOW MORE] By step (11),
A
∩
B
≤
C
A
(
M
)
≤
A
∗
{\displaystyle A\cap B\leq C_{A}(M)\leq A^{*}}
. Since
A
∩
B
≤
B
{\displaystyle A\cap B\leq B}
, we get
A
∩
B
≤
A
∗
∩
B
{\displaystyle A\cap B\leq A^{*}\cap B}
.
13
A
∗
{\displaystyle A^{*}}
normalizes
A
{\displaystyle A}
step (8)
[SHOW MORE] Note that
M
≤
N
=
N
B
(
A
)
{\displaystyle M\leq N=N_{B}(A)}
, so
M
{\displaystyle M}
normalizes
A
{\displaystyle A}
.
C
A
(
M
)
≤
A
{\displaystyle C_{A}(M)\leq A}
, with
A
{\displaystyle A}
abelian, so
C
A
(
M
)
{\displaystyle C_{A}(M)}
normalizes
A
{\displaystyle A}
. Thus, the product
M
C
A
(
M
)
{\displaystyle MC_{A}(M)}
normalizes
A
{\displaystyle A}
.
14
M
{\displaystyle M}
is not contained in
A
{\displaystyle A}
step (8)
[SHOW MORE] By assumption
x
∉
N
=
N
B
(
A
)
{\displaystyle x\notin N=N_{B}(A)}
,
x
{\displaystyle x}
does not normalize
A
{\displaystyle A}
. Thus,
M
=
[
x
,
A
]
{\displaystyle M=[x,A]}
is not contained in
A
{\displaystyle A}
.
15
A
∗
∩
B
{\displaystyle A^{*}\cap B}
is not contained in
A
∩
B
{\displaystyle A\cap B}
steps (8), (14)
[SHOW MORE] By step (8),
[
x
,
A
]
=
M
≤
N
≤
B
{\displaystyle [x,A]=M\leq N\leq B}
, and also
M
≤
A
∗
{\displaystyle M\leq A^{*}}
, so
M
≤
A
∗
∩
B
{\displaystyle M\leq A^{*}\cap B}
. By step (14),
M
{\displaystyle M}
is not contained in
A
{\displaystyle A}
, and hence,
M
{\displaystyle M}
is not contained in
A
∩
B
{\displaystyle A\cap B}
. Thus,
A
∗
∩
B
{\displaystyle A^{*}\cap B}
is not contained in
A
∩
B
{\displaystyle A\cap B}
.
16
A
∗
{\displaystyle A^{*}}
is contained in
C
=
A
B
{\displaystyle C=AB}
steps (1), (8), (10)
[SHOW MORE]
A
∗
=
M
C
A
(
M
)
{\displaystyle A^{*}=MC_{A}(M)}
. By step (8),
M
{\displaystyle M}
is in
B
{\displaystyle B}
and hence in
A
B
{\displaystyle AB}
. By definition,
C
A
(
M
)
{\displaystyle C_{A}(M)}
is in
A
{\displaystyle A}
and hence in
A
B
{\displaystyle AB}
. Thus, the product is in
A
B
{\displaystyle AB}
.
17
A
∗
{\displaystyle A^{*}}
is contained in
⟨
A
C
⟩
{\displaystyle \langle A^{C}\rangle }
.
steps (8), (9), (10)
[SHOW MORE] A^* = MCA_(M)</math>.
M
{\displaystyle M}
is contained in
⟨
A
C
⟩
{\displaystyle \langle A^{C}\rangle }
because it's defined as
[
A
,
x
]
{\displaystyle [A,x]}
and
x
∈
C
{\displaystyle x\in C}
.
C
A
(
M
)
≤
A
≤
⟨
A
C
⟩
{\displaystyle C_{A}(M)\leq A\leq \langle A^{C}\rangle }
. Thus, the product is also contained in
⟨
A
C
⟩
{\displaystyle \langle A^{C}\rangle }
.
18
A
∗
{\displaystyle A^{*}}
normalizes
B
{\displaystyle B}
step (1), (16)
[SHOW MORE] By step (1),
B
{\displaystyle B}
is normal in
C
{\displaystyle C}
. Since
A
∗
≤
C
{\displaystyle A^{*}\leq C}
by step (16),
A
∗
{\displaystyle A^{*}}
thus also normalizes
B
{\displaystyle B}
.
The conclusions of steps (10), (12), (13), (15), (16), (17), and (18) complete the proof.
Conceptual summary of proof
The key idea of the proof is to use fact (1) somehow, and this requires finding a suitable
x
{\displaystyle x}
such that
[
x
,
A
]
{\displaystyle [x,A]}
is abelian. and the resulting subgroup
A
∗
{\displaystyle A^{*}}
that we get is nice enough.
Working backward from this, we see that we would like that
[
x
,
A
]
{\displaystyle [x,A]}
not be a subgroup of
A
{\displaystyle A}
, which in turn means that we want
x
{\displaystyle x}
outside
N
P
(
A
)
{\displaystyle N_{P}(A)}
. On the other hand, we do want
A
∗
{\displaystyle A^{*}}
to normalize
A
{\displaystyle A}
, which means that we want
[
x
,
A
]
{\displaystyle [x,A]}
to lie inside
N
P
(
A
)
{\displaystyle N_{P}(A)}
. Finally, we would like
[
x
,
A
]
⊆
B
{\displaystyle [x,A]\subseteq B}
so that the subsequent group constructed is in
A
B
{\displaystyle AB}
. With all these considerations, we should aim to find
x
{\displaystyle x}
outside
N
B
(
A
)
{\displaystyle N_{B}(A)}
but inside
B
{\displaystyle B}
, but such that
[
x
,
A
]
⊆
N
B
(
A
)
{\displaystyle [x,A]\subseteq N_{B}(A)}
.
This motivation now allows us to work out the proof details given above. Here is a rough outline of the proof:
Steps (1)-(6) set things up. The fact that
B
{\displaystyle B}
does not normalize
A
{\displaystyle A}
is used to show that
B
/
N
{\displaystyle B/N}
is a nontrivial normal subgroup of
P
/
N
{\displaystyle P/N}
.
The crucial steps for the construction are (7) and (8), that involve the choice of
M
{\displaystyle M}
. Steps (9) and (10) now use fact (1) to get a hold on
A
∗
{\displaystyle A^{*}}
.
The remaining steps help us get a clearer idea of the subgroup inclusions and prove the remaining desired properties for
A
∗
{\displaystyle A^{*}}
.
GAP implementation
Here is a GAP implementation of the constructive approach used in the proof:
ThompsonsReplacementAbelian := function(P,A,B)
local C,N,x,M,CAM,Astar;
C := Group(Union(A,B));
N := Normalizer(B,A);
x := Filtered(Set(B),y -> (not(y in N)) and IsSubgroup(N,CommutatorSubgroup(Group(y),C)))[1];
M := Group(List(Set(A),y -> x^(-1) * y^(-1) * x * y));
CAM := Centralizer(A,M);
Astar := Group(Union(M,CAM));
return Astar;
end;;
References
Textbook references
Journal references