The term direct sum of Abelian groups generalizes the concept of the direct sum of vector spaces . It is of great importance for the theory of Abelian groups . If a group can be broken down into a direct sum, the structure of the group is reduced to simpler groups. New groups can be formed from the direct summands. Most structure theorems make a statement about a direct decomposition of groups.
Definitions
The Abelian group is said to be a direct sum of two subgroups , if the following two conditions are met.
A.
{\ displaystyle A}
A.
1
{\ displaystyle A_ {1}}
A.
2
{\ displaystyle A_ {2}}
A.
1
+
A.
2
=
A.
{\ displaystyle A_ {1} + A_ {2} = A}
.
A.
1
∩
A.
2
=
0
{\ displaystyle A_ {1} \ cap A_ {2} = 0}
.
In this case it is written . Here referred to the subgroup that only the identity element contains.
A.
=
A.
1
⊕
A.
2
{\ displaystyle A = A_ {1} \ oplus A_ {2}}
0
{\ displaystyle 0}
0
{\ displaystyle 0}
A subgroup is a direct summand if there is a subgroup are with: . In this case the complement of .
A.
1
↪
A.
{\ displaystyle A_ {1} \ hookrightarrow A}
A.
2
{\ displaystyle A_ {2}}
A.
=
A.
1
⊕
A.
2
{\ displaystyle A = A_ {1} \ oplus A_ {2}}
A.
2
{\ displaystyle A_ {2}}
A.
1
{\ displaystyle A_ {1}}
A.
{\ displaystyle A}
means directly indecomposable if and are the only direct summands of .
A.
{\ displaystyle A}
0
{\ displaystyle 0}
A.
{\ displaystyle A}
Be a family of subgroups of . The group is called the direct sum of when the following conditions are met.
(
A.
i
|
i
∈
I.
)
{\ displaystyle (A_ {i} | i \ in I)}
A.
{\ displaystyle A}
A.
{\ displaystyle A}
(
A.
i
|
i
∈
I.
)
{\ displaystyle (A_ {i} | i \ in I)}
∑
i
∈
I.
A.
i
=
A.
{\ displaystyle \ sum _ {i \ in I} A_ {i} = A}
. The family creates .
(
A.
i
|
i
∈
I.
)
{\ displaystyle (A_ {i} | i \ in I)}
A.
{\ displaystyle A}
For each applies: .
i
∈
I.
{\ displaystyle i \ in I}
A.
i
∩
∑
j
≠
i
A.
j
=
0
{\ displaystyle A_ {i} \ cap \ sum _ {j \ neq i} A_ {j} = 0}
It is written: . Or if .
A.
=
⨁
i
∈
I.
A.
i
{\ displaystyle A = \ bigoplus \ limits _ {i \ in I} A_ {i}}
A.
=
A.
1
⊕
⋯
⊕
A.
n
{\ displaystyle A = A_ {1} \ oplus \ dots \ oplus A_ {n}}
I.
=
{
1
,
...
,
n
}
{\ displaystyle I = \ {1, \ dots, n \}}
Explanations, simple sentences
They are subgroups of the Abelian group . Then the following statements are equivalent:
B.
,
C.
{\ displaystyle B, C}
A.
{\ displaystyle A}
It is .
A.
=
B.
⊕
C.
{\ displaystyle A = B \ oplus C}
Each can be clearly written as with .
a
∈
A.
{\ displaystyle a \ in A}
a
=
b
+
c
{\ displaystyle a = b + c}
b
∈
B.
,
c
∈
C.
{\ displaystyle b \ in B, c \ in C}
It is and follows from with .
A.
=
B.
+
C.
{\ displaystyle A = B + C}
0
=
b
+
c
{\ displaystyle 0 = b + c}
b
∈
B.
,
c
∈
C.
{\ displaystyle b \ in B, c \ in C}
b
=
0
=
c
{\ displaystyle b = 0 = c}
If , then the two endomorphisms and have the following property: and . It is the identity .
A.
=
B.
⊕
C.
{\ displaystyle A = B \ oplus C}
π
B.
:
A.
∋
b
+
c
↦
b
∈
A.
{\ displaystyle \ pi _ {B} \ colon A \ ni b + c \ mapsto b \ in A}
π
C.
:
a
∋
b
+
c
↦
c
∈
A.
{\ displaystyle \ pi _ {C} \ colon a \ ni b + c \ mapsto c \ in A}
π
B.
2
=
π
B.
,
π
C.
2
=
π
C.
,
π
C.
π
B.
=
π
B.
π
C.
=
0
{\ displaystyle \ pi _ {B} ^ {2} = \ pi _ {B}, \ pi _ {C} ^ {2} = \ pi _ {C}, \ pi _ {C} \ pi _ {B } = \ pi _ {B} \ pi _ {C} = 0}
1
A.
=
π
B.
+
π
C.
{\ displaystyle 1_ {A} = \ pi _ {B} + \ pi _ {C}}
1
A.
{\ displaystyle 1_ {A}}
A.
{\ displaystyle A}
Homomorphisms provide a way of identifying and recognizing direct summands:
Be homomorphisms. Then:
A.
⟶
f
B.
⟶
G
C.
{\ displaystyle A {\ overset {f} {\ longrightarrow}} B {\ overset {g} {\ longrightarrow}} C}
G
f
{\ displaystyle gf}
is a monomorphism and is a monomorphism.
⟺
image
(
f
)
∩
core
(
G
)
=
0
{\ displaystyle \ iff \ operatorname {image} (f) \ cap \ operatorname {core} (g) = 0}
f
{\ displaystyle f}
If there is an epimorphism , then is .
G
f
{\ displaystyle gf}
image
(
f
)
+
core
(
G
)
=
B.
{\ displaystyle \ operatorname {image} (f) + \ operatorname {core} (g) = B}
If there is an isomorphism , then is .
G
f
{\ displaystyle gf}
image
(
f
)
⊕
core
(
G
)
=
B.
{\ displaystyle \ operatorname {image} (f) \ oplus \ operatorname {core} (g) = B}
The following statements are equivalent for a subgroup :
B.
↪
A.
{\ displaystyle B \ hookrightarrow A}
B.
{\ displaystyle B}
is a direct summand in .
A.
{\ displaystyle A}
There is an endomorphism with: and .
π
:
A.
→
A.
{\ displaystyle \ pi \ colon A \ to A}
π
2
=
π
{\ displaystyle \ pi ^ {2} = \ pi}
π
(
A.
)
=
B.
{\ displaystyle \ pi (A) = B}
If the inclusion map is, there is a homomorphism with .
ι
:
B.
→
A.
{\ displaystyle \ iota \ colon B \ to A}
π
:
A.
→
B.
{\ displaystyle \ pi \ colon A \ to B}
π
ι
=
1
B.
{\ displaystyle \ pi \ iota = 1_ {B}}
Examples
0
{\ displaystyle 0}
is a direct summand in every group.
Let it be the cyclic group with the associated addition. Be it . Then is . There are and subsets of . Their average is and their sum is . It is for example .
A.
{\ displaystyle A}
A.
=
Z
/
6th
Z
=
{
0
,
1
,
2
,
3
,
4th
,
5
}
{\ displaystyle A = \ mathbb {Z} / 6 \ mathbb {Z} = \ {0,1,2,3,4,5 \}}
U
=
{
0
,
3
}
,
V
=
{
0
,
2
,
4th
}
{\ displaystyle U = \ {0.3 \}, V = \ {0.2.4 \}}
A.
=
U
⊕
V
{\ displaystyle A = U \ oplus V}
U
{\ displaystyle U}
V
{\ displaystyle V}
A.
{\ displaystyle A}
0
{\ displaystyle 0}
A.
{\ displaystyle A}
3
+
4th
=
1
mod
6th
{\ displaystyle 3 + 4 = 1 {\ bmod {6}}}
The groups of whole numbers and rational numbers are indecomposable. If a prime number , it is directly indecomposable.
Z
{\ displaystyle \ mathbb {Z}}
Q
{\ displaystyle \ mathbb {Q}}
p
{\ displaystyle p}
Z
/
p
Z
{\ displaystyle \ mathbb {Z} / p \ mathbb {Z}}
If the Abelian group has a largest subgroup , then it is directly indivisible. If a prime number, then has the largest subgroup . So it cannot be decomposed directly.
≠
A.
{\ displaystyle \ neq A}
A.
{\ displaystyle A}
p
{\ displaystyle p}
A.
=
Z
/
p
n
Z
{\ displaystyle A = \ mathbb {Z} / p ^ {n} \ mathbb {Z}}
p
Z
/
p
n
Z
≠
A.
{\ displaystyle p \ mathbb {Z} / p ^ {n} \ mathbb {Z} \ neq A}
A.
{\ displaystyle A}
Are coprime integers, then is .
a
,
b
{\ displaystyle a, b}
Z
/
(
a
⋅
b
)
Z
=
a
Z
/
(
a
⋅
b
)
Z
⊕
b
Z
/
(
a
⋅
b
)
Z
{\ displaystyle \ mathbb {Z} / (a \ cdot b) \ mathbb {Z} = a \ mathbb {Z} / (a \ cdot b) \ mathbb {Z} \ oplus b \ mathbb {Z} / (a \ cdot b) \ mathbb {Z}}
The last example has a strong generalization: be a group and be with . In addition, be coprime with . Then is .
A.
{\ displaystyle A}
n
∈
N
{\ displaystyle n \ in \ mathbb {N}}
A.
⋅
n
=
0
{\ displaystyle A \ cdot n = 0}
n
=
r
⋅
s
{\ displaystyle n = r \ cdot s}
r
,
s
{\ displaystyle r, s}
A.
=
A.
⋅
r
⊕
A.
⋅
s
{\ displaystyle A = A \ cdot r \ oplus A \ cdot s}
Is , so is , where is. The complement of is by no means clearly determined. For example, it is also for everyone .
Z
2
=
{
(
z
1
,
z
2
)
|
z
1
,
z
2
∈
Z
}
{\ displaystyle \ mathbb {Z} ^ {2} = \ {(z_ {1}, z_ {2}) | z_ {1}, z_ {2} \ in \ mathbb {Z} \}}
Z
2
=
e
1
→
Z
⊕
e
2
→
Z
{\ displaystyle \ mathbb {Z} ^ {2} = {\ vec {e_ {1}}} \ mathbb {Z} \ oplus {\ vec {e_ {2}}} \ mathbb {Z}}
e
1
→
=
(
1
,
0
)
,
e
2
→
=
(
0
,
1
)
{\ displaystyle {\ vec {e_ {1}}} = (1.0), {\ vec {e_ {2}}} = (0.1)}
e
1
→
Z
{\ displaystyle {\ vec {e_ {1}}} \ mathbb {Z}}
Z
2
=
e
1
→
Z
⊕
(
z
,
1
)
Z
{\ displaystyle \ mathbb {Z} ^ {2} = {\ vec {e_ {1}}} \ mathbb {Z} \ oplus (z, 1) \ mathbb {Z}}
z
∈
Z
{\ displaystyle z \ in \ mathbb {Z}}
The last example is more general. Be a natural number. the set of - tuples with components . Next is the tuple that has one here and in other places . Then is .
n
≥
1
{\ displaystyle n \ geq 1}
Z
n
: =
{
(
z
1
,
...
,
z
n
)
|
z
i
∈
Z
}
{\ displaystyle \ mathbb {Z} ^ {n}: = \ {(z_ {1}, \ dots, z_ {n}) | z_ {i} \ in \ mathbb {Z} \}}
n
{\ displaystyle n}
Z
{\ displaystyle \ mathbb {Z}}
e
i
→
{\ displaystyle {\ vec {e_ {i}}}}
i
{\ displaystyle i}
1
{\ displaystyle 1}
0
{\ displaystyle 0}
Z
n
=
⨁
i
=
1
n
e
→
i
Z
{\ displaystyle \ mathbb {Z} ^ {n} = \ bigoplus \ limits _ {i = 1} ^ {n} {\ vec {e}} _ {i} \ mathbb {Z}}
To determine whether a subgroup is a direct summand, there is a simple criterion:
Z
2
{\ displaystyle \ mathbb {Z} ^ {2}}
Be . Then the following statements are equivalent.
a
→
=
(
a
1
,
a
2
)
∈
Z
2
{\ displaystyle {\ vec {a}} = (a_ {1}, a_ {2}) \ in \ mathbb {Z} ^ {2}}
a
→
Z
{\ displaystyle {\ vec {a}} \ mathbb {Z}}
is a direct summand in .
Z
2
{\ displaystyle \ mathbb {Z} ^ {2}}
There is with .
b
1
,
b
2
∈
Z
{\ displaystyle b_ {1}, b_ {2} \ in \ mathbb {Z}}
a
1
⋅
b
1
+
a
2
⋅
b
2
=
1
{\ displaystyle a_ {1} \ cdot b_ {1} + a_ {2} \ cdot b_ {2} = 1}
Some grids and determinants as area
A grid with the generating vectors
(
1
,
0
)
Z
,
(
0
,
1
)
Z
{\ displaystyle (1,0) \ mathbb {Z}, (0,1) \ mathbb {Z}}
A grid with the generating vectors is shown
(
1
,
1
)
Z
,
(
-
2
,
-
1
)
Z
{\ displaystyle (1,1) \ mathbb {Z}, (- 2, -1) \ mathbb {Z}}
Property 2. of the last sentence has a geometric meaning: The subgroup is a direct summand in if and only if there is a vector such that a parallelogram of area 1 spans.
a
→
Z
{\ displaystyle {\ vec {a}} \ mathbb {Z}}
Z
2
{\ displaystyle \ mathbb {Z} ^ {2}}
b
→
=
(
b
1
,
b
2
)
{\ displaystyle {\ vec {b}} = (b_ {1}, b_ {2})}
a
→
,
b
→
{\ displaystyle {\ vec {a}}, {\ vec {b}}}
The last statement can be generalized. If so the following applies: is a direct summand in if and only if the numbers have the greatest common factor .
a
→
=
(
a
1
,
...
,
a
n
)
∈
Z
n
{\ displaystyle {\ vec {a}} = (a_ {1}, \ dots, a_ {n}) \ in \ mathbb {Z} ^ {n}}
a
→
Z
{\ displaystyle {\ vec {a}} \ mathbb {Z}}
Z
n
{\ displaystyle \ mathbb {Z} ^ {n}}
a
1
,
...
,
a
n
{\ displaystyle a_ {1}, \ dots, a_ {n}}
1
{\ displaystyle 1}
Primary groups
The following theorem makes a statement about the decomposition of torsion groups. For this purpose it is defined: Let be a prime number. The group is called -primarily if and only if everyone has a with . The sum of all -primary subgroups of a group is -primary. It is the largest primary subgroup of . It is
denoted by and is called the primary component of . The following applies:
p
{\ displaystyle p}
A.
{\ displaystyle A}
p
{\ displaystyle p}
a
∈
A.
{\ displaystyle a \ in A}
n
∈
N
{\ displaystyle n \ in \ mathbb {N}}
a
⋅
p
n
=
0
{\ displaystyle a \ cdot p ^ {n} = 0}
p
{\ displaystyle p}
A.
{\ displaystyle A}
p
{\ displaystyle p}
p
{\ displaystyle p}
A.
{\ displaystyle A}
A.
p
{\ displaystyle A_ {p}}
p
{\ displaystyle p}
A.
{\ displaystyle A}
If a torsion group is, then is . It is the direct sum of its primary components.
A.
{\ displaystyle A}
A.
=
⨁
p
prim
A.
p
{\ displaystyle A = \ bigoplus \ limits _ {p {\ text {prim}}} A_ {p}}
A.
{\ displaystyle A}
Universal property
Be for two subgroups and the canonical inclusions. They are equivalent:
A.
=
A.
1
+
A.
2
{\ displaystyle A = A_ {1} + A_ {2}}
A.
1
,
A.
2
{\ displaystyle A_ {1}, A_ {2}}
q
i
:
A.
i
↪
A.
{\ displaystyle q_ {i} \ colon A_ {i} \ hookrightarrow A}
A.
=
A.
1
⊕
A.
2
{\ displaystyle A = A_ {1} \ oplus A_ {2}}
.
For every two homomorphisms there is exactly one homomorphism with for .
f
i
:
A.
i
→
B.
,
i
∈
{
1
,
2
}
{\ displaystyle f_ {i} \ colon A_ {i} \ to B, i \ in \ {1,2 \}}
f
:
A.
→
B.
{\ displaystyle f \ colon A \ to B}
f
∘
q
i
=
f
i
{\ displaystyle f \ circ q_ {i} = f_ {i}}
i
∈
{
1
,
2
}
{\ displaystyle i \ in \ {1,2 \}}
The second statement of the theorem is the so-called universal property of the direct sum. It applies to any index set.
Be a family of subgroups with . And be the inclusions. Then are equivalent:
(
A.
i
|
i
∈
I.
)
{\ displaystyle (A_ {i} | i \ in I)}
∑
i
∈
I.
A.
i
=
A.
{\ displaystyle \ sum \ limits _ {i \ in I} A_ {i} = A}
q
i
:
A.
i
↪
A.
{\ displaystyle q_ {i} \ colon A_ {i} \ hookrightarrow A}
It is .
⨁
i
∈
I.
A.
i
=
A.
{\ displaystyle \ bigoplus \ limits _ {i \ in I} A_ {i} = A}
For each family of homomorphisms there is exactly one with . That means, the following diagram is commutative for all .
f
i
:
A.
i
→
B.
{\ displaystyle f_ {i} \ colon A_ {i} \ to B}
f
:
A.
→
B.
{\ displaystyle f \ colon A \ to B}
f
∘
q
i
=
f
i
{\ displaystyle f \ circ q_ {i} = f_ {i}}
i
∈
I.
{\ displaystyle i \ in I}
Let and be two abelian groups with and . If there is exactly one with and exactly one with for every family , then and are isomorphic.
(
A.
,
q
i
)
{\ displaystyle (A, q_ {i})}
(
S.
,
s
i
)
{\ displaystyle (S, s_ {i})}
q
i
:
A.
i
→
A.
{\ displaystyle q_ {i} \ colon A_ {i} \ to A}
s
i
:
A.
i
→
S.
{\ displaystyle s_ {i} \ colon A_ {i} \ to S}
f
i
:
A.
i
→
B.
{\ displaystyle f_ {i} \ colon A_ {i} \ to B}
f
:
A.
→
B.
{\ displaystyle f \ colon A \ to B}
f
i
=
f
∘
q
i
{\ displaystyle f_ {i} = f \ circ q_ {i}}
G
:
S.
→
B.
{\ displaystyle g \ colon S \ to B}
f
i
=
G
∘
s
i
{\ displaystyle f_ {i} = g \ circ s_ {i}}
A.
{\ displaystyle A}
S.
{\ displaystyle S}
Some structure sentences
Theorem: If there is a homomorphism, then with and .
f
:
A.
→
Z
{\ displaystyle f \ colon A \ to \ mathbb {Z}}
A.
=
core
(
f
)
⊕
a
Z
{\ displaystyle A = \ operatorname {core} (f) \ oplus a \ mathbb {Z}}
a
∈
A.
{\ displaystyle a \ in A}
a
⋅
Z
≅
Z
{\ displaystyle a \ cdot \ mathbb {Z} \ cong \ mathbb {Z}}
Theorem: Each subgroup of is the direct sum of at most cyclic subgroups.
Z
n
{\ displaystyle \ mathbb {Z} ^ {n}}
n
{\ displaystyle n}
Theorem: If it is torsion-free and generated by elements, then there is a monomorphism .
F.
{\ displaystyle F}
n
{\ displaystyle n}
F.
→
Z
n
{\ displaystyle F \ to \ mathbb {Z} ^ {n}}
Conclusion: If there is a torsion-free group generated by elements, then there is a such that is isomorphic to .
F.
{\ displaystyle F}
n
{\ displaystyle n}
k
≤
n
{\ displaystyle k \ leq n}
F.
{\ displaystyle F}
Z
k
{\ displaystyle \ mathbb {Z} ^ {k}}
If finitely generated, the torsion subgroup is a direct summand of .
A.
{\ displaystyle A}
A.
{\ displaystyle A}
Individual evidence
^ László Fuchs: Abelian Groups. Springer, 2015, ISBN 978-3-319-19421-9 , p. 43.
^ Frank W. Anderson, Kent R. Fuller: Rings and Categories of Modules. Springer, 1992, ISBN 0-387-97845-3 , p. 66.
literature
Web links
Since it is quite tedious to look for the evidence to the facts in the given literature together, evidence is compiled here .
<img src="https://de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">