This paper explores the connection between $S$-subgroups in $ extbf{Z}_2^n$ and binary codes, revealing conditions under which these subgroups are free and their invariance properties under permutation actions.
Contribution
It establishes the relationship between $S$-subgroups and specific classes of binary codes, detailing conditions for freeness and invariance under permutation groups.
Findings
01
$S$-subgroups are free when codes are both $P(T)$-codes and $G$-codes.
02
Constructed codes are cyclic, decimated, or symmetric, leading to specific invariance properties.
03
No codes generate the entire $ extbf{Z}_2^n$ in any $ extbf{G}_n(a)$-complete $S$-set.
Abstract
In this paper the relationship between S-subgroups in Z2n and binary codes is shown. If the codes used are both P(T)-codes and G-codes, then the S-subgroup is free. The codes constructed are cyclic, decimated or symmetric and the S-subgroups obtained are free under the action the cyclic permutation subgroup, invariants under the action the decimated permutation subgroup and symmetric under the action of symmetric permutation subgroup, respectively. Also it is shows that there is no codes generating whole Z2n in any \Gn(a)-complete S-set of the S-ring S(Z2n,Sn).
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCoding theory and cryptography · graph theory and CDMA systems · Finite Group Theory Research
Full text
Schur ring and Codes for S-subgroups over Z2n
Ronald Orozco López
Abstract
In this paper the relationship between S-subgroups in Z2n and binary codes is shown.
If the codes used are both P(T)-codes and G-codes, then the S-subgroup is free. The codes
constructed are cyclic, decimated or symmetric and the S-subgroups obtained are free under the
action the cyclic permutation subgroup, invariants under the action the decimated permutation
subgroup and symmetric under the action of symmetric permutation subgroup, respectively. Also
it is shows that there is no codes generating whole Z2n in any Gn(a)-complete
S-set of the S-ring S(Z2n,Sn).
Let G be a finite group with identity element e and C[G] the group algebra of all formal sums
∑g∈Gagg, ag∈C, g∈G. For T⊂G, the element ∑g∈Tg will
be denoted by T. Such an element is also called a simple quantity.
The transpose of T=∑g∈Gagg is defined as T⊤=∑g∈Gag(g−1). Let {T0,T1,...,Tr} be a partition of G and let S be the subspace
of C[G] spanned by T1,T2,...,Tr. We say that S is
a Schur ring (S-ring, for short) over G if:
T0={e},
2. 2.
for each i, there is a j such that Ti⊤=Tj,
3. 3.
for each i and j, we have Ti⋅Tj=∑k=1rλi,j,kTk, for constants λi,j,k∈C.
The numbers λi,j,k are the structure constants of S with respect to the linear base
{T0,T1,...,Tr}. The sets Ti are called the
basic sets of the S-ring S. Any union of them is called an S-sets. Thus,
X⊆G is an S-set if and only if X∈S. The set of all S-set is closed
with respect to taking inverse and product. Any subgroup of G that is an S-set, is called an
S-subgroup of G or S-group. A partition
{T0,...,Tr} of G is called Schur partition or S-partition
if T0={e} and if for each i there is some j such that
Ti−1={g−1:g∈Ti}=Tj. It is known that there is a 1-1 correspondence between
S-rings over G and S-partitions of G. By using this correspondence, in this paper we will
refer to an S-ring by mean of its S-partition.
The concept of S-ring was iniciated by I. Schur in their classical paper [1] which
was published in 1933. Later, the theory of S-ring was developed for Wielandt [2].
But the main objective of theory was purely group theoretical concept, especially in
problem concerning the permutations groups. In the 80s and 90s, the theory received a notable
impulse by the study of S-ring over cyclic groups and their applications to the graph theory
[3],[4],[5],[6].
With the papers [8],[9] and [11] was initiated the research of S-ring over the group Z2n
and was shown the relationship between this and Hadamard matrices, perfect binary sequences and
periodic compatible binary sequences. In this paper we will show the relationship between
S-ring over Z2n and binary codes. In particular we will use codes for to construct
S-subgroups over Z2n. This point of view will be shown as an alternative to the cocyclic
matrices and to the difference sets used for to research hadamard matrices, since with the
schur rings and its generator codes will be possible to understand the structure of the special
binary sequences above.
In this paper a code Xn generating whole Z2n is found. Then the others codes for
S-subgroups are constructed by using Xn as a base. A code Xn′ will be called
G-codes if there exists a permutation subgroup G in Aut(Z2n) such that
GXn′=Xn′. Other types of codes studied are the P(T)-codes, closely
related to the partition P(T) of the subset T of N={0,1,...,n−1}. In fact, the
S-subgroups constructed are both P(T)-codes and G-codes. This codes generating free
S-subgroups in Z2n.
This paper is organized as follows. In section 2 basic concepts of theory of codes are shown.
In section 3 properties of P(T)-codes are stablished. Also is shown that a P(T)-code generates
a free subgroup in Z2n and that a G-code generates an S-subgroup in the same group.
In section 4 is shown the connection between G-codes and Gn(a)-complete S-set.
S-subgroups and its generator codes in S-rings induced by the permutation subgroups
Cn, Δn, HnCn, ΔnCn and HnΔnCn of
Aut(Z2n) are studied in the sections 5 to 10.
2 Some terminology of the Theory of Codes
In [11] the following terminology related to the theory of codes can be found
An arbitrary set A will be called an alphabet and its elements are called
letters. A finite sequence of letters written in the form s1s2⋯sn,
n≥0, with every si in A, is called a word. Any subsequence of
consecutive letters of a word is a subword. When n=0 the word is
the empty word and denoted by 1A. Given a word w=s1s2⋯sn, the
number n is called the lenght of w and is denoted l(w). Then the empty word
1A has lenght [math], i.e., l(1A)=0.
Let A∗ denote all finite words defined on A and let A+
denote all finite nonempty words on A. A∗ is equipped with an
associative binary operation obtained by concatenating two sequences:
[TABLE]
The empty word is the unit element with respect to this operation and consequently the sets
A∗ and A+ are a monoid and a semigroup, respectivelly.
A factorization of a word s∈A∗ is a sequence {s1,s2,...,sn} of
n≥0 words in Σ∗ such that s=s1s2⋯sn. For a subset X of
A∗, we denote by X∗ the submonoid generated by X,
[TABLE]
Similarly, we denote by X+ the subsemigroup generated by X,
[TABLE]
By definition, each word s in X∗ admits a least one factorization x1x2⋯xn
with all xi in X. Such a factorization is called an X-factorization.
A monoid M is called free if it has a subset B such that:
M=B∗, and
2. 2.
For all n,m≥1 and x1,⋯xn, y1,⋯ym∈B we have
[TABLE]
From condition 1., B is a generating set of M and condition 2. say us that each element in M
has an unique representation as a product of elements of B. The set B satisfying 1. y 2.
is called a base of M.
Let M be a monoid and B its generating set. We say that B is minimal generating
set if no proper subset of B is a generating set. A element x of M is called
indecomposable or atomic if it cannot be expressed in the form x=yz with
y,z=1.
Now, by reformulating the condition (2.1) we obtain the following definition. A subset
X⊆A∗ is a code if it satisfies the following condition:
For all n,m≥1 and x1,...,xn, y1,...,ym∈X
[TABLE]
Note that the empty word 1A is never in a code. The following theorem shows the
equivalence between codes and free generating set
Theorem 1**.**
Let X⊆A∗. Then the following conditions are equivalent:
X* is a code,*
2. 2.
X* is a free generating set, or a base, of the monoid X∗,*
3. 3.
X∗* is free and X is its minimal generating set.*
On the other hand, let G be a group with A⊆G. Elements of A∗
represent elements of G closed under concatenation and inversion. The empty word represents
1G, the unity of G. Then A∗ is a subgroup of G. A word
w=s1s2⋯sn in A∗ is called reduced if w contains
no subwords xx−1 or x−1x for x∈A. A group G is called a free group if
there exists a generating set X of G such that every non-empty reduced group word in X
defines a non-trivial element of G. Let G be a free group on X. Then the cardinality of X
is called the rank of G.
3 Schur rings and codes for S-subgroups over Z2n
In this paper denote by Z2 the cyclic group of order 2 with elements +
and −(where + and − mean 1 and −1 respectively). Let
Z2n=Z2×⋯×Z2n. Then all X∈Z2n are sequences of + and − and will be called Z2-sequences or binary sequences.
All binary sequence in Z2n is of the form (x0,x1,...,xn−1).
Let 1 denote the sequence (1,1,...,1). As X2=1 for all X in
Z2n, then all reduced word in Z2n contains no the subword XX. Now we will find
a code generating whole Z2n.
We define the following subset of Z2n
[TABLE]
where each − is in the i-th position. In the following theorem we shall show that
Xn is a base for all Z2n
Theorem 2**.**
Xn* is a code for Z2n.*
Proof.
As ∣Xn∣=n, then all word on Xn has the form
[TABLE]
with ϵi=0,1. Thus, the number of words on Xn of lenght k is
(kn), with k ranging in [1,n−1]. As the empty word corresponds to 1
and as X0X1⋯Xn−1=−1, the number total of words constructed
with the codewords Xi is 2n. Hence there exist a 1-1 correspondence between all words on
Xn and all binary sequences in Z2n. Consequently
Xn∗=Z2n as we announce.
∎
Let Aut(Z2n) denote the automorphism group of Z2n and take G by a subgroup
of Aut(Z2n). We shall denote with S(Z2n,G) an S-partition of
Z2n under the action of G. As Xn generates whole Z2n, we wish
to find codes on Xn, this is, with codewords factorizable on
Xn, for S-subgroups of S(Z2n,G). We start with
the following definition
Definition 1**.**
Let T={i1,i2,...,ir} be a subset of N={0,1,...,n−1} and let
P(T)={T1,...,Ts} denote a partition on T. A code Xn′ on Xn is a
P(T)-code if Xn′={YT1,...,YTs}, where
YTj=Xj1⋯Xjk and Tj={j1,...,jk}.
The map Tj↦YTj establishes an 1-1 correspondence between the blocks Tj of
T and the codewords YTj of Xn′. Then it is easily inferred that
Theorem 3**.**
Let Xn′ be a P(T)-code. Then
[TABLE]
We will call to Xn′∗ a P(T)-free group.
Proof.
Let P(T)={T1,...,Ts} be a partition of some subset T of N. Then all word on
Xn′ is irreducible. Hence the number of words on Xn′
of lenght k is (k∣X′∣) and
[TABLE]
∎
Corollary 1**.**
Xn* in (3.6) is the only P(T)-code generating whole Z2n.*
Proof.
Xn is a P(T)-code with P(T)={{0},{1},...,{n−1}}. The statement is followed from
here.
∎
Now, we will find the number of P(T)-free subgroups in Z2n
Theorem 4**.**
The number of P(T)-free subgroup in Z2n is B∣T∣+1
where B∣T∣ are the Bell numbers.
Proof.
By the correspondence is clear that the number of P(T)-free subgroups for any subset T is
B∣T∣. As (∣T∣n) indicates the number of ∣T∣-element
subsets of an n-element set, then (∣T∣n)B∣T∣ indicates the number
of P(T)-free subgroups with fixed size. Assuming that B0 is the number of empty words we
obtain ∑∣T∣=0n(∣T∣n)B∣T∣=B∣T∣+1.
∎
For example, the following are all P(T)-free subgroup of Z23
[TABLE]
The reason for deal with P(T)-code will be showed now. Let
[TABLE]
a code for Z27 and define the set
[TABLE]
on X7. It is easy to show that X7′ is not a code. For example,
X0X2X5X6 has at least two factorization in X′, namely
X0X2X3⋅X3X5X6 and X2X4X5⋅X4X6X0.
Also, ∣X7′∗∣=16 and not 128 as desirable. Hence
X7′∗ is not free group. However, from theorem 3
a P(T)-code always generates a free group.
The following theorem say us as to obtain new P(T)-free subgroup from old.
Theorem 5**.**
Let Xi be P(Ti)-codes, 1≤i≤r, in Z2n such that
Ti∩Tj=∅, i=j, and Ti⊂N. Then
[TABLE]
and
[TABLE]
Proof.
Follows by induction on number of Ti-codes Xi.
∎
Now we will obtain a necessary condition for the existence of an S-subgroup
Theorem 6**.**
Let G be a permutation automorphic subgroup of Aut(Z2n) acting on some
set X in Z2n. Then X∗ is an S-subgroup in
S(Z2n,G).
Proof.
We take a word Yi1Yi2⋯Yir in X′∗ and a g in G. Then
[TABLE]
As g is arbritary, then g(Yi1Yi2⋯Yir) is in X′∗
for all g in G. Hence G defines a partition on X′∗ and X′∗ is
an S-subgroup of S(Z2n,G).
∎
Not all S-subgroup is free. X7′∗ in (3.9) is an S-subgroup of
S(Z27,C7), where C7=⟨C⟩ is the cyclic
permutation automorphic subgroup of Aut(Z27) of order 7 with C the cyclic permutation
acting on all component of some Y in Z27. But X7′∗ is not a free subgroup of
Z27.
Again let G be a permutation automorphic subgroup of Aut(Z2n) and let YG denote the
orbit of some Y in Z2n under the action of G. From previous theorem YG∗ is an
S-subgroup. We will called to YG∗ a basicS-subgroup of
S(Z2n,G). If for a code X it is true that X=YG for some Y in
Z2n, then we will say that X is a G-code. In the following sections we construct
S-subgroups by using G-codes.
4 Schur ring S(Z2n,Sn)
Let ω(X) denote the Hamming weight of X∈Z2n. Thus, ω(X) is the number of
+ in any Z2−sequences X of Z2n. Now let Gn(k) be the subset of
Z2n such that ω(X)=k for all X∈Gn(k), where 0≤k≤n.
We let Ti=Gn(n−i). It is straightforward to prove that the partition
S(Z2n,Sn)={Gn(0),...,Gn(n)} induces an S-partition over
Z2n, where Sn≤Aut(Z2n) is the permutation group on n objects. From [7]
it is know that the constant structure λi,j,k is equal to
From (3.6) we know that Xn=Gn(n−1). Then Xn is an Sn-code for
S(Z2n,Sn). In the following corollary we found another Sn-code for
S(Z2n,Sn)
Corollary 2**.**
Gn(1)* is an Sn-code for S(Z2n,Sn).*
Proof.
It is enough to take into account that −Gn(n−1)=Gn(1).
∎
We prefer to use the Sn-code Gn(n−1) and not Gn(1) because in Gn(n−1)
the positions of the negative components are easily obtained. Indeed, +++−⋅+−++=+−+− in
G4(3) but in G4(1) we have −−−+⋅−+−−=+−+−.
Next we will see that a G-code is contained in any S-set of
S(Z2n,Sn)
Proposition 1**.**
A G-code X is contained in Gn(a) for some a ranging in [1,n−1].
Proof.
Take X in X. It is easy to note that ω(gX)=ω(X) for all g∈G.
Hence X⊆Gn(a) for some a in [1,n−1].
∎
On the other hand, it is follows directly from (4.1) that λi,j,2k+1=0 if
i+j is even and λi,j,2k=0 if i+j is odd. The union of all basic sets Gn(2a)
in S(Z2n,Sn) will be denoted by En and the union of all basic sets
Gn(2a+1) in S(Z2n,Sn) will be denoted On. The sets E2n
and O2n+1 are subgroups of order 22n−1 and 22n, respectively. Then
[TABLE]
and
[TABLE]
are S-subgroups of S(Z22n,S2n) and S(Z22n+1,S2n+1),
respectively.
From (4.2), G2n(n)2=⋃i=0nG2n(2i)=E2n and
G2n+1(n)2=⋃i=0nG2n+1(2i+1)=O2n+1. Therefore, neither G4n(2n)
nor G4n+3(2n+1) contains some code X generating whole Z24n and Z22n+1,
respectively. This remark is generalized below.
From [7] is obtained the following definition
Definition 2**.**
Take Gn(a) in S(Z2n,Sn). Let S′⊂S(Z2n,Sn) be a set of basic sets. We will call S′ a
Gn(a)-completeS-set if it holds
Gn(i)Gn(j)⊃Gn(a)* for all Gn(i),Gn(j)∈S′,*
2. 2.
There is no Gn(b)∈S(Z2n,Sn) such that
Gn(b)2⊃Gn(a) and Gn(b)Gn(k)⊃Gn(a) for all
Gn(k)∈S′.
A important result obtained is that there is no Gn(a)-complete for all n and all a
Theorem 7**.**
There is no G2n(2a+1)-complete S-sets in S(Z22n,Sn).
2. 2.
There is no G2n+1(2a)-complete S-sets in S(Z22n+1,Sn).
In the following theorem is shown the relationship between codes generating whole
Z2n and non Gn(a)-complete S-sets in S(Z2n,Sn)
Theorem 8**.**
There is no a code X generating whole Z2n in a Gn(a)-complete S-set.
Proof.
Let S′ denote a G2n(2a)-complete S-set. From (4.2)
[TABLE]
for all G2n(2b) in S′. Then all powers of G2n(2b) will
contain basic sets G2n(2k) only. Therefore the basic sets in a
G2n(2a)-complete can generate the S-subgroup E2n at the most. With a similar
argument is shown for basic sets in G2n+1(2a+1)-complete S-sets.
∎
We finish this section showing some basic sets of S(Z2n,Sn) that can
to contain G-codes generating all Z2n.
then G4n(2n−1)3=O4n.
From G4n(2n−1)G4n(2n+1)⊃G4n(0) and from (4) is followed
that G4n(2n−1)4=E4n.
∎
As Sn induces a S-partition on Z2n is straightforward to prove that G induces a
S-partition on Z2n for all G≤Sn≤Aut(Z2n). In the following
sections we will construct S-subgroups by using G-codes in S-ring S(Z2n,G).
5 Schur ring S(Z2n,Cn)
Let C denote the cyclic permutation on the components + and − of X in
Z2n such that
[TABLE]
that is, C(xi)=x(i+1)modn. The permutation C is a generator of cyclic group
Cn=⟨C⟩ of order n. Let
XC=OrbCnX={Ci(X):Ci∈Cn}. Therefore, Cn defines a partition in
equivalent class on Z2n which is an S-partition and this we shall denote by
Z2Cn=S(Z2n,Cn). It is worth mentioning that this Schur ring
corresponds to the orbit Schur ring induced by the cyclic permutation automorphic subgroup
Cn≤Sn≤Aut(Z2n).
On the other hand, let X={xi} and Y={yi} be two complex-valued sequences of period
n. The periodic correlation of X and Y at shift k is the product defined by:
[TABLE]
where a denotes the complex conjugation of a and i+k is calculated modulo n.
If Y=X, the correlation PX,Y(k) is denoted by PX(k) and is the
autocorrelation of X. Obviously,
[TABLE]
for all 0≤i≤n−1 and for all X in Z2n.
If X is a Z2-sequence of length n, PX(k)=2ω{Yk}−n,
where Yk=XCkX. Also by (4.2), if X∈Gn(a), then
[TABLE]
for some 0≤ik≤a and n−PX(k) is divisible by 4 for all k.
We know from theorem 2 that Gn(n−1) is a code for
S(Z2n,Sn) for all n. As Gn(n−1)={X,CX,C2X,...,Cn−1}=XC,
then XC is a Cn-code for S(Z2n,Cn). Then for to obtain
information from each basic set YC in S(Z2n,Cn) we must do it through
of its XC-factorization with the Cn-code XC. The advantage of using this code lies
in its simplicity, since each CiX has exactly a − as its component and thereby it is
possible to know exactly the Hamming weight of each word writing with this basis.
All word Y in Gn(a) has the form Ci1XCi2X⋯CirX with length
∣Y∣=r and with a=n−r. Then, every basic set in GnC(a) has form
[TABLE]
and in this way if Z=Cj1XCj2X⋯CjsX, we have
[TABLE]
Each word YCkZ=Ci1X⋯CirXCj1+kX⋯Cjs+kX can be reduced if
exist two equal letters. Thereupon YCkZ decreases its length an even number. Therefore YCkZ
belong to Gn(b) with b=n−(r+s)+2w where 2w is the number of canceled letters. If both
Y and Z belongs to Gn(a), then b=n−2r+2w and PY,Z(k)=n−4r+4wk.
Next we will obtain the algebraic version of (5.3), (5.4) and
(5.5)
Proposition 3**.**
Let Y denote the binary sequence Ci1XCi2X⋯CirX. If YCkY∈Gn(a),
then
YCn−kY* and (CjX)Ck(CjX) are in Gn(a) too.*
2. 2.
(−Y)Ck(−Y)∈Gn(a).
Proof.
It is clear that
[TABLE]
with a=n−2r+2w, where 2w are the number of canceled letters. We wish to show that the
cancellation numbers of YCkY and YCn−kY coincide. Suppose that ij=i1+k for some
j and some k. Then this implies that n−k+ij=i1 reduced module n. Therefore
[TABLE]
has the same number of cancellations as YCkY. Equally is proved for (CjX)Ck(CjX).
2. As Ck(−Y)=−CkY, then (−Y)Ck(−Y)=YCkY∈Gn(a).
∎
Now we will show other advantage of to use the Cn-code XC
Proposition 4**.**
For all n≤2 we have
[TABLE]
Proof.
All word Y in Gn(n−2) has the form CiXCjX with i<j. Then
YCkY=CiXCjXCi+kXCj+kX and there exist a k such that either i+k=j or j+k=i
and for all the remaining values of k we have that CiXCjXCi+kXCj+kX is a reduced
word. As YCkY and YCn−kY are in Gn(a) for some a, then YC2 contains
2 words in Gn(n−2), n−3 words in Gn(n−4) and the trivial word in Gn(n).
∎
On the other hand, let
[TABLE]
Clearly d divides to n and the X∈Fd(Z2n) have the form X=(Y,Y,...,Y), with
Y∈Z2d. Then Fd(Z2Cn)=⋃∣XC∣=dXC is an S-set of
Z2Cn, for each d∣n. When d=n, we will to say that Cn acts freely on
XC and we denote Fn(Z2Cn) as F(Z2Cn). When d<n, we will to say that
Cn don’t act freely on XC and let F(Z2Cn) denote the set of the
XC which are not frees under the action of Cn, namely
[TABLE]
Therefore,
[TABLE]
The set Fd(Z2n) is constructed with codewords in
[TABLE]
where Ai,dX=CiXCi+dX⋯Ci+kdX for i=0,1,2,...,d−1 with k=dn−1,
X∈Gn(n−1) and
[TABLE]
is an arithmetic progression. Then XF,d is a P(T)-code with
[TABLE]
and any word in XF,d∗ has the form
[TABLE]
ϵi=0,1. It is clear that CiAj,dX=Ai+j,dX. Hence XF,d is a
Cn-code.
If d=1, then k=n−1, i=0 and
[TABLE]
is a code with a codeword. As XCXC2X⋯Cn−1X=−1, then
XF,1∗={1,−1} for all n.
If d=n, then k=0, i=0,1,⋯,n−1 and
[TABLE]
is an code with exactly n codewords and XF,d=XC. Let ∣XF,d∣ be the rank
of XF,d. We then note that 1<∣XF,d∣<n for 1<d<n.
Now we will see the relationship between free subgroup XF,d∗ and the sets
Fd(Z2n)
Theorem 9**.**
XF,d∗* is a subgroup of Z2n of order 2d with
XF,d∗=⋃r∣dFd(Z2n). We will donote this subgroup with
Gd(n).*
Proof.
As XF,d is a P(T)-code, then Gd(n) is a subgroup of Z2n of order
2d for all divisor d of n. Then we only will show that Gd(n) has the
desired structure. If d is a prime divisor of n, then all words of Gd(n) are
in Fd(Z2n), except for 1 and −1. Hence
[TABLE]
with
[TABLE]
Suppose that d is no prime. Then
[TABLE]
is contained in Fr(Z2n), r∣d and i=0,1,...,r−1. Therefore
Gd(n)=⋃r∣dFr(Z2n).
∎
Corollary 3**.**
GdC(n)* is an S-subgroup of S(Z2n,Cn).*
Proof.
The group ⟨Cn⟩ defines a partition on each Fr(Z2n) in
Gd(n) and hence we obtain the desired statement.
∎
Example 1**.**
The subgroup G3(9) of Z29 is given by
[TABLE]
And the S-subgroup G3C(9) of S(Z29,C9) is given by
[TABLE]
Theorem 10**.**
Gd(n)⊆⋃a=0dGn(dna)* with equality only for
d=1,n.*
Proof.
Follows from Fd(Z2n)⊆Gn(dn)∪Gn(n−dn).
∎
Now we show the lattice of S-subgroups GdC(60) of
S(Z260,C60) ordered by inclusion.
We finish this section we provide other proof to the Theorems 6 and 7 in [7]. We start with
the lemma
Lemma 1**.**
[TABLE]
for all X in XC=G2n(2n−1)
Proof.
Clearly XCnX is in Fn(Z22n) when X=−+++⋯+++. As
[TABLE]
for any other codeword CiX in the code XC, then
Ci(XCnX)∈Fn(Z22n) for all 1≤i≤2n−1, since
⟨Cn⟩ defines a partition on Fn(Z22n).
∎
Theorem 11**.**
If YC∈F(Z2C2n), then YC2∖{1}∈F(Z2C2n).
Proof.
Take Y=Ci1X⋯CirX in F(Z22n). Then
[TABLE]
From the above lemma XCnX is in Fn(Z22n) for all X in XC. Then
YCnY∈Fn(Z22n) for all Y in F(Z2n) and
[TABLE]
as we promised to show.
∎
Lemma 2**.**
[TABLE]
for no X in XC=G2n+1(2n) and for no k ranging in [1,2n], d<2n+1.
Proof.
Take X=−++⋯++ in XC. Then XCkX=−+⋯+2n−k−+⋯+k−1. If we want k−1=2n−k, then k=22n+1, which is
not possible. Hence XCkX is no contained in Fd(Z22n+1), d<2n+1. As
Ci(XCkX)=(CiX)Ck(CiX), it is followed the statement.
∎
Theorem 12**.**
If XC∈F(Z2C2n+1), then XC2∖{1}∈F(Z2C2n+1).
Proof.
Let Y=Ci1XCi2X⋯CirX such that all the ij are not in arithmetic
progression. If YCkY=Ci1(XCkX)Ci2(XCkX)⋯Cir(XCkX) is
contained in some Fd(Z22n+1), d∣(2n+1), d<2n+1, then XCkX must be
contained in Fd(Z22n+1), but is not possible by the previous lemma.
∎
6 Schur ring S(Z2n,Δn)
Let δa∈Sn act on X∈Z2n by decimation, that is,
δa(xi)=xai(modn) for all xi in X, (a,n)=1 and let Δn denote
the set of this δa. The set Δn is a group of order ϕ(n) isomorphic to
Zn∗, the group the units of Zn, where ϕ is called the Euler totient function.
Clearly S(Z2n,Δn) is an S-partition of Z2n. In this section,
we will construct Δn-codes for S-subgroups of S(Z2n,Δn).
We will use the commutation relation Ciδa=δaCia for to prove all of results
in this section. We begin for show that Gn(n−1) is partitioned in three equivalence class
Proposition 5**.**
[TABLE]
where X=−++⋯++ and
[TABLE]
r=n−ϕ(n)−1.
Proof.
It is very easy to see that X=−++⋯++ is fixed under the action of Δn. Also, as
δaCiX=Ca−1iδaX=Ca−1iX, then ΔnXZn∗=XZn∗ and
ΔnXZn∖Zn∗=XZn∖Zn∗.
∎
As each CiX in XZn∗ or in XZn∖Zn∗
is atomic, then XZn∗ and XZn∖Zn∗ are
Δn-code and hence XZn∗∗ and
XZn∖Zn∗∗ are S-subgroups in
S(Z2n,Δn) with ∣XZn∗∗∣=2ϕ(n) and
∣XZn∖Zn∗∗∣=2n−ϕ(n)−1.
On the other hand, let
[TABLE]
denote the autocorrelation vector of Y in Z2n and let A(Z2n)
denote the set of all this. Let X1+X2+⋯+Xn=a denote the plane in Zn in the
indeterminates Xi, i=1,2,...,n and let
θ:Z2n→A(Z2n) be the map defined by
θ(Y)=(PY(0),PY(1),…,PY(n−1)).
The decimation group Δn do not alter the set of values which PX(k) takes
on, but merely the order in which they appear, i.e., if Y=δaX then
PY(k)=PX(ka). Therefore, we have the commutative diagram
[TABLE]
and θ∘δr=δr∘θ.
Let Y∈Z2n such that θ(Y)=(n,d,d,...,d). Such a binary sequence is known as
binary sequence with 2-levels autocorrelation value and are important by its applications on
telecommunication. We want to construct a Δn-code for some S-subgroup H of
S(Z2n,Δn) containing such Y. From (6.4) is followed that
θ(Y)=δaθ(Y)=θ(δaY), for all δa∈Δn. Hence Y
and δaY have the same autocorrelation vector. For Y fullfilling δaY=Y for some
δa in Δn we have the following definition
Definition 3**.**
Let a be a unit in Zn∗. A word Y in Z2n is δa-invariant
if δaY=Y. Denote by In(a) the set of these Y.
If Y is in In(a), then δrY is in In(a), too. Also
δa(YZ)=δaYδaZ=YZ for all Y,Z in In(a). Then
In(a) is an S-subgroup of S(Z2n,Δn). Now, we shall see
that all factorization of words in In(a) is relationated with cyclotomic coset
of a module n. First, we have the following definition
Definition 4**.**
Let a relative prime to n. The cyclotomic coset of a module n is defined by
[TABLE]
where sat≡smodn. A subset {s1,s2,…,sr} of
Zn is called complete set of representatives of cyclotomic coset of a modulo n if
Ci1,Ci2,…, Cir are distinct and are a
partition of Zn.
Take Y=Ci1XCi2X⋯CirX in In(a) with X=−++⋯++.
We want δaY=Y. Then
[TABLE]
since δaX=X. As must be δaY=Y, then ik=a−1ij or ij=aik for
1≤k,j≤r. Let CsX denote the word CsXCsaX⋯Csats−1X.
Then all Y in In(a) has the form
Y=Cs1ϵ1XCs2ϵrX⋯CsrϵX, with ϵi=0,1. As In(a) is an S-subgroup in
S(Z2n,Δn), δrCsiX=CsjX and
[TABLE]
is a Δn-code for In(a). Also XI(a) is a P(T)-code
with
[TABLE]
and {s1,s2,...,sr} a complete set of representatives. Hence XI(a)∗
has order 2r+1, where r is the number of cyclotomic cosets of a module n
In the table 1, binary sequences with 2-level autocorrelation values with their
respective δa-invariants S-subgroups are shown
On the other hand, we have the following theorem
Theorem 13**.**
If ⟨b⟩ is a subgroup of ⟨a⟩, then
In(a)≤In(b).
Proof.
Let C1a and C1b denote the classes {1,a,a2,...,at−1}
and {1,b,b2,...,bs−1}. By hypothesis
⟨b⟩≤⟨a⟩, then C1b⊆C1a. Hence there exists yi in ⟨a⟩ such that
[TABLE]
and k=[⟨a⟩:⟨b⟩]. Then is follows that
[TABLE]
Therefore ∣XIn(a)∣≤∣XIn(b)∣ and
In(a)≤In(b).
∎
We finish this section constructing some δa-invariants S-subgroups
We proof 1. The proof of 2 it is analogous. We note that
[TABLE]
then C1={1,2n}. As 2n+1∤2n−1 and q<2n+1, then 2nq≡qmod(2n+1)
and the Cq={q,2nq} are cyclotomic cosets of 2n module 2n+1. Finally, it is easy
to note that 2nq is congruent to 2n+1−q module 2n+1,
[TABLE]
∎
Proposition 7**.**
Let 2p+1 be an prime number with p an odd prime number. The S-subgroups invariants in
Z22p+1 are I2p+1(x), I2p+1(y) and I2p+1(2p),
where x is a primitive root module 2p+1 and y is not neither primitive root module 2p+1
nor 2p.
Proof.
Let P={x1,x2,...,xt} denote the set of primitive roots module 2p+1. Then
[TABLE]
As ∣⟨xi⟩∣=2p for any xi∈P, then
⟨xi⟩ has exactly a subgroup of order 2 and a subgroup of order p.
Therefore there exist S-subgroups invariants I2p+1(y) and I2p+1(2p)
where y∈Pc∖{2p}, with Pc the complement of P in Z2p+1∗.
∎
7 Schur ring S(Z2n,Hn)
We note by RY the reversed sequence RY=(yn−1,...,y1,y0) and let Hn denote the
permutation automorphic subgroup Hn={1,R}≤Sn≤Aut(Z2n). Hence Hn defines
a partition on Z2n and S(Z2n,Hn) is a schur ring.
Definition 5**.**
Let Y∈Z2n. We shall call Y symmetric if RY=Y and otherwise we say it is
non symmetric. We make Sym(Z2n) the set of all Y symmetric and Sym(Z2n)
the set of all Y nonsymmetric.
Take Y∈Z2n such that Y is of the form Ci1XCi2X⋯CirX. We want
to understand the structure of the words in Sym(Z2n). As it must be fulfilled that RY=Y,
then taking X=+⋯+−+⋯+ in Gn(n−1) with n an odd number we have
[TABLE]
where we have used that RX=X. Hence if Y is symmetric, then must be n−ij=ik for j=k
ranging in [1,r]. Thereby Y has the form
[TABLE]
for ϵi=0,1 and Y0=X and Yi=CiXCn−iX. As CiXCn−iX is in
Sym(Z2n) for all i, then
[TABLE]
is a Hn-code for Sym(Z2n). For the case n an even number it is easily followed that
[TABLE]
is a Hn-code for Sym(Z2n), where X=+++⋯++−. Also it is clear that
XSymE and XSymO are P(T)-codes, therefore
Sym(Z2n) is an free S-subgroup in S(Z2n,Hn) and
Sym(Z22n+1) and Sym(Z22n) have order 2n+1 and 2n, respectively.
Finally, the relationship between the symmetric subgroup Sym(Z22n+1) and the
δ2n-invariant S-subgroup I2n+1(2n) is shown
Theorem 14**.**
Sym(Z22n+1)=I2n+1(2n).
Proof.
By proposition 6 the codewords in XI2n+1(2n) are X
and CqXC2n+1−qX, 1≤q≤n, with X=+⋯+−+⋯+. We want to show that all of
codewords in XI2n+1(2n) is symmetric. For this we note that RX=X and
[TABLE]
∎
8 Schur ring S(Z2n,HnCn)
In this section we will use the commutation relation
[TABLE]
to show that the S-subgroups Gd(n) and Sym(Z2n) are S-subgroups in
S(Z2n,HnCn).
Theorem 15**.**
GdC(n)* is an S-subgroup in S(Z2n,HnCn)*
Proof.
By having in mind the commutation relation (8.1), we can to show that
XF,d is an Hn-code. In this way we have
[TABLE]
As X=−++⋯++, then RX=CX. Hence
[TABLE]
Finally, reordering and rewriting
[TABLE]
∎
Definition 6**.**
A S-set YC in Z2Cn is symmetric if R⋅YC=YC, where R⋅YC means
the action of R on the elements of YC. The set of all symmetric S-sets will be denoted by
Sym(Z2Cn) and the set of all non-symmetric S-sets will be denoted by
Sym(Z2Cn).
It is easy to note that R⋅YC=(RY)C. Therefore YC is a symmetric S-set if and
only if contains some CiY symmetric.
Theorem 16**.**
Sym(Z2Cn)* is an S-subgroup of S(Z2n,HnCn).*
Proof.
Take YC,ZC in Sym(Z2Cn) and suppose that Y,Z∈Sym(Z2n). As
[TABLE]
then R(YCZC)=R(YC)R(ZC)=YCZC.
∎
9 Schur ring S(Z2n,ΔnCn)
In this section we will use the commutation relation
[TABLE]
to show that the S-subgroups Gd(n) and In(a) are S-subgroups in
S(Z2n,ΔnCn).
Theorem 17**.**
GdC(n)* is an S-subgroup in S(Z2n,ΔnCn).*
Proof.
We want to show that XF,d is a Δn-code. Take Ai,dX in XF,d. Then if
k=dn−1 we have
[TABLE]
Define the map ϑ:XF,d→{ni/d:i=1,2,...,d−1} by
[TABLE]
Then ϑ is a biyection. As ϑ(δaAi,dX)≡da−1nimodn
and ϑ(ClδaAi,dX)≡(a−1+l)dnimodn, is followed that
δaAi,dX∈XF,d and therefore XF,d is a Δn-code.
∎
Definition 7**.**
Let a be a unit in Zn∗. A basic set YC in Z2Cn is
δa-invariant if δa⋅YC=YC. Denote by InC(a)
the set of these Y.
By having in mind the basic set YC, we can note that δa⋅YC=(δaY)C.
Hence YC is δa-invariant in Z2Cn if and only if contains some CiYδa-invariant in Z2n.
Theorem 18**.**
InC(a)* is an S-subgroup in S(Z2n,ΔnCn).*
Proof.
Take YC,ZC in InC(a) and suppose that Y,Z∈In(a). As
δa(YCkZ)C=(δaYδaCkZ)C=(YCka−1Z)C,
then δa(YCZC)=δa(YC)δa(ZC)=YCZC.
∎
10 Schur ring S(Z2n,HnΔnCn)
Finally we show that the S-subgroups Gd(n), Sym(Z2n) and
In(a) are S-subgroups in S(Z2n,HnΔnCn).
Let YC be any basic set in Z2Cn. It is a very easy to notice that
YC=(CkY)C for all k. We will use this fact for to prove the following lemma
Lemma 3**.**
**
δaYiC=Ya−1iC* for Yi=CiXC2n+1−iX∈XSymO.*
2. 2.
δaYiC=Y(a−1i+2a−1−1)C* for
Yi=CiXC2n−1−iX∈XSymE.*
Proof.
1. Take δa in the group Δn. It is clear that δaX=CkaX for
some ka depending on a, where X=+⋯+−+⋯+ is the word in XC used to
construct all codewords in XSymO. Then
[TABLE]
2. Equally, δaX=CkaX for some ka depending on a, where X=++⋯++−
in XC is used to construct all codewords in XSymE. Then
[TABLE]
∎
We will use this lemma for to show that the symmetric binary sequences form an S-subgroup in
S(Z2n,HnΔnCn)
Theorem 19**.**
Sym(Z2Cn)* is an S-subgroup in S(Z2n,HnΔnCn).*
Proof.
Clearly Sym(Z2Cn) is an S-subgroup of S(Z2n,HnCn). From
the previous lemma is followed that Δn defines a partition on XSymE and
XSymO. Hence they are Δn-codes and Sym(Z2Cn) is an S-subgroup of
S(Z2n,HnΔnCn).
∎
From theorem 19 the case a=n−1 is
excluded. In the previous section already was proved that InC(a) is an S-subgroup in
S(Z2n,ΔnCn). Now, we wish to show that Hn defines a partition
on InC(a) by using (8.1). Take the codeword CsX in
XIn(a). We have then
[TABLE]
∎
Bibliography11
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] I. Schur. Zur Theorie der einfach transitiven Permutationsgruppen , Sitzungsber. Preuss. Akad. Wiss., Phys.-Math. Kl., 598–623, 1993.
2[2] H. Wielandt. Finite Permutation Groups , Academic Press, New York-London, 1964.
3[3] M. Klin, R. Poschel. The konig problem, the isomorphism problem for cyclic graphs and the method of schur rings , Algebraic Methods in Graph Theory, 1, 2, 1978.
4[4] S. L. Ma. On association schemes, schur rings, strongly regular graphs and partial difference sets , Ars Combin., 21:211-220, 1989.
5[5] A. Heinze. Applications of Schur rings in algebraic combinatorics: graphs, partial difference sets and cyclotomic schemes , Ph D thesis, Universitat Oldenburg, 2001.
6[6] M. Muzychuk, M. Klin, R. Poschel. The isomorphism problem for circulant graphs via Schur ring theory , Dis. Math. The. Com. Sci. 56 , 241-264, 2001.
7[7] M. E. Muzichuk, The Subschemes of the Hamming Scheme , Investigations in Algebraic Theory of Combinatorial Objects Volume 84 of the series Mathematics and Its Applications pp 187-208, 1992.
8[8] R. Orozco, Schur Ring over Group ℤ 2 n superscript subscript ℤ 2 𝑛 \mathbb{Z}_{2}^{n} , Circulant S − limit-from 𝑆 S- Sets Invariant by Decimation and Hadamard Matrices , 2018, ar Xiv:1802.05788.