Abstract
Self-dual codes over the ring \(\mathbb {Z}_{4}\) are related to combinatorial designs and unimodular lattices. First, we discuss briefly how to construct self-dual cyclic codes over \(\mathbb {Z}_{4}\) of arbitrary even length. Then we focus on solving one key problem of this subject: for any positive integers k and m such that m is even, we give a direct and effective method to construct all distinct Hermitian self-dual cyclic codes of length 2k over the Galois ring GR(4,m). This then allows us to provide explicit expressions to accurately represent all these Hermitian self-dual cyclic codes in terms of binomial coefficients. In particular, several numerical examples are presented to illustrate our applications.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
The class of self-dual codes is closely related to other fields of mathematics, such as lattices, cryptography, invariant theory, block designs, etc. In particular, self-dual codes over \(\mathbb {Z}_{4}\) are related to combinatorial designs and unimodular lattices (cf. [2, 6, 8, 23, 25,26,27,28]). The construction of self-dual codes over \(\mathbb {Z}_{4}\) is an interesting topic in coding theory.
In general, the construction of optimal self-dual codes requires computer searches. To reduce the search field, self-dual codes can be constructed from linear codes with some special algebraic structures, such as cyclic codes, constacyclic codes and quasi-cyclic codes, etc.
The study of cyclic codes over finite rings started to attract much attention in the 1990s, when it was observed that some good nonlinear codes over \(\mathbb {F}_{2}\) can be viewed as binary images of linear cyclic codes over \(\mathbb {Z}_{4}\) under a Gray map [24]. In particular, [24] motivated the study of cyclic and negacyclic codes over Galois rings (see, for example, [1, 3,4,5, 15, 30, 35, 40, 44, 47, 48]). Using the Gray map from \(\mathbb {Z}_{4}\) onto \(\mathbb {F}_{2}^{2}\), defined by: 0↦00, 1↦01, 2↦11, 3↦10, binary formally self-dual codes can be obtained from self-dual codes over \(\mathbb {Z}_{4}\) with good parameters. The construction of self-dual codes over \(\mathbb {Z}_{4}\) and other rings has since become a research topic of much interest (cf. [31, 34, 36, 43]).
Cyclic codes were initially studied where their length is relatively prime to the characteristic of the ring. The structure of this class of cyclic codes over rings was studied in [7, 15, 35, 39, 40] and certain special generating sets for these codes were determined therein. Cyclic codes (resp. negacyclic codes) whose length is not relatively prime to the characteristic of the ring are called repeated-root cyclic codes (resp. negacyclic codes). The first study for this latter class of cyclic codes was done in [1], where the generators for cyclic codes over \(\mathbb {Z}_{4}\) of length 2e were determined. Then the generators for cyclic codes over \(\mathbb {Z}_{4}\) of length 2n were presented in [4], where n is odd. Repeated-root cyclic and negacyclic codes are also interesting as they allow very simple syndrome-forming and decoding circuitry and, in some cases (see [38, 41]), they are maximum distance separable. A partial list of references for the theory of repeated-root cyclic codes includes [14, 16,17,18,19,20,21, 32, 33, 37, 41, 42, 45, 49].
Another important reason for studying cyclic codes over \(\mathbb {Z}_{4}\) of even length is that there are more self-dual codes among them than there are among cyclic codes over \(\mathbb {Z}_{4}\) of odd length. For example: the numbers of self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 23, 22 and 20 are equal to 3, 33 and 63, respectively; the numbers of self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 25, 26 and 28 are equal to 1, 65 and 339, respectively. In fact, some good binary self-dual codes or formally self-dual codes can be obtained from self-dual cyclic codes over \(\mathbb {Z}_{4}\) of even length. Here is a simple example: there is only one binary self-dual cyclic code of length 8 and its basic parameters are [8,4,2]. However, there are 3 self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 4, and two of them give binary self-dual codes having optimal parameters [8,4,4] by the Gray map defined above.
Now, we briefly review some main results on the determination of self-dual cyclic codes of even length over \(\mathbb {Z}_{4}\) in the literature. A concatenated structure and an explicit representation for all distinct self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 2n and 4n were given by [9, 10], for any positive odd integer n. For length 2kn, where k ≥ 3, using the methods in [9, 10] will result in complex representations.
Let k,n ≥ 3 be any integers such that n is odd. Using the standard Discrete Fourier Transform decomposition, which may be viewed as an extension of the approaches in [4] and [21], and by [22, Theorem 3.2 and Corollary 3.3] and [29, Lemma 4.3 and Proposition 4.5], the problem of determining all (Euclidean) self-dual cyclic codes of length 2kn over \(\mathbb {Z}_{4}\) can be translated into solving the following three problems (see Section 2 of this paper for details):
-
(♭) Determining all cyclic codes and their Euclidean dual codes of length 2k over a Galois extension ring of \(\mathbb {Z}_{4}\), say GR(4,m), where m ≥ 1.
-
(♮) Constructing and expressing explicitly all Euclidean self-dual cyclic codes of length 2k over GR(4,m).
-
(♯) Constructing and expressing explicitly all Hermitian self-dual cyclic codes of length 2k over GR(4,m), where m is an even positive integer.
So far, several results on the three problems have been obtained:
-
For Problem (♭): All cyclic codes and their Euclidean dual codes over GR(4,m) of length 2k have been determined by [22, Lemma 2.4 (iv), Proposition 2.5 and Theorem 5.3] and [32, 33].
-
For Problem (♮): The number of Euclidean self-dual cyclic codes over GR(4,m) of length 2k was determined by [33, Corollary 3.5]. Then explicit expressions for all distinct Euclidean self-dual cyclic codes over GR(4,m) of length 2k were given in [11], using binomial coefficients.
-
For Problem (♯): The number \(N_{\mathrm {H}}\left (\text {GR}(4,m),2^{k}\right )\) of all Hermitian self-dual cyclic codes over GR(4,m) of length 2k was determined by [29, Theorem 3.4]: \(N_{\mathrm {H}}\left (\text {GR}(4,m),2^{k}\right ) =\frac {\left (2^{\frac {m}{2}}\right )^{2^{k-1}+1}-1}{2^{\frac {m}{2}}-1}\), where m is even.
However, to the best of our knowledge, there are no general results on the construction and explicit representation of all distinct Hermitian self-dual cyclic codes over GR(4,m) of length 2k.
In order to represent explicitly all Euclidean self-dual cyclic codes of length 2kn over \(\mathbb {Z}_{4}\), we need to solve Problem (♯) completely. This is the main contribution of this current work.
The paper is organised as follows. In Section 2, we discuss briefly how to construct Euclidean self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 2kn, for any positive odd integer n. In Section 3, we introduce necessary notation for the Galois ring GR(4,m) and Hermitian dual codes over GR(4,m). In Section 4, we give a direct and effective approach to construct precisely all distinct Hermitian self-dual cyclic codes over GR(4,m) of length 2k by Theorem 1. This then allows us to provide an explicit expression to accurately represent all these Hermitian self-dual cyclic codes by Theorem 2, using binomial coefficients. In Section 5, we prove Theorem 1 in detail. As an application, we give explicitly all distinct Hermitian self-dual cyclic codes of length 2k over GR(4,m), for the cases of k = 3,4,5, in Section 6. Section 7 concludes the paper.
2 Constructing self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 2k n
In this section, we describe how to construct all distinct Euclidean self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 2kn, where n is an odd positive integer.
As in [22] and [29, Section 4], define the ring \(\mathcal {R}=\frac {\mathbb {Z}_{4}[u]}{\langle u^{2^{k}}-1\rangle }\). Then we have a \(\mathbb {Z}_{4}\)-module isomorphism \({{\Phi }}: \mathcal {R}^{n}\rightarrow \frac {\mathbb {Z}_{4}[x]}{\langle x^{2^{k}n}-1\rangle }\) defined by: for any \(c_{i}(u)={\sum }_{j=0}^{2^{k}-1}c_{i,j}u^{j}\in \mathcal {R}\) with \(c_{i,j}\in \mathbb {Z}_{4}\), i = 0,1,…,n − 1, let
Let M be the the multiplicative order of 2 modulo n and let ζ be a primitive n th root of unity in the Galois ring GR(4,M). We define the Discrete Fourier Transform of
as the vector \((\widehat {c}_{0},\widehat {c}_{1},\ldots , \widehat {c}_{n-1})\in \left (\frac {\text {GR}(4,M)[u]}{\langle u^{2^{k}}-1\rangle } \right )^{n}\) with
where \(nn^{\prime }\equiv 1\) (mod 2k), and define the Mattson-Solomon polynomial of c(Z) to be \(\widehat {c}(Z)={\sum }_{h=0}^{n-1}\widehat {c}_{n-h \ (\text {mod} \ n)}Z^{h}\). By [22, Lemma 3.1] or [29, Lemma 4.1], we have the following
where ⋆ indicates the componentwise multiplication.
For any integer h, 0 ≤ h ≤ n − 1, denote by S2(h) the 2-cyclotomic coset modulo n containing h, i.e., S2(h) = {h2i (mod n)∣i = 0,1,…}. The 2-cyclotomic coset S2(h) is said to be self-inverse if S2(−h) = S2(h). Set J0 = I0 = {0}. Let I1 be the union of all self-inverse 2-cyclotomic cosets modulo n excluding I0 and set I2 = {0,1,…,n − 1}∖ (I0 ∪ I1). The set I2 is the union of pairs of 2-cyclotomic cosets of the form S2(h) ∪ S2(−h), where h∉I0 ∪ I1. Let J1 and J2 be complete sets of representatives of 2-cyclotomic cosets in I1 and I2, respectively. Without loss of generality, we assume that J2 is chosen such that h ∈ J2 if and only if n − h ∈ J2. For any h ∈ J0 ∪ J1 ∪ J2, denote by mh the size of S2(h). Then mh is even for all h ∈ J1.
By [22, Theorem 3.2 and Corollary 3.3] or [29, Lemma 4.3], we know that \(\frac {\mathbb {Z}_{4}[x]}{\langle x^{2^{k}n}-1\rangle }\cong {\prod }_{h\in J_{0}\cup J_{1}\cup J_{2}} \frac {\text {GR}(4,m_{h})[x]}{\langle x^{2^{k}}-1\rangle }\) via the following ring isomorphism
Then every Euclidean self-dual cyclic code \(\mathcal {C}\) over \(\mathbb {Z}_{4}\) of length 2kn can be constructed as follows (cf. [29, Proposition 4.5]):
where
-
◇ C0 is a Euclidean self-dual cyclic code over \(\mathbb {Z}_{4}\) of length 2k,
-
◇ Cj is a Hermitian self-dual cyclic code of length 2k over the Galois ring GR(4,mj) for all j ∈ J1,
-
◇ Ch is a cyclic code of length 2k over the Galois ring GR(4,mh) and \(C_{n-h}=C_{h}^{\bot _{E}}\), where \(C_{h}^{\bot _{E}}\) is the Euclidean dual of Ch, for all h ∈ J2.
In the following sections, we focus on the problem of constructing Hermitian self-dual cyclic code of length 2k over the Galois ring GR(4,m).
3 Preliminaries
In this section, we introduce the notation needed in the following sections.
Let \(\mathbb {Z}_{4}=\{0,1,2,3\}\) in which the arithmetic is done modulo 4, and let \(\mathbb {Z}_{2}=\{0,1\}\) in which the arithmetic is done modulo 2. In this paper, we regard \(\mathbb {Z}_{2}\) as a subset of \(\mathbb {Z}_{4}\), although \(\mathbb {Z}_{2}\) is not a subfield of the ring \(\mathbb {Z}_{4}\). In that sense, we have that \(2\mathbb {Z}_{2}=\{0,2\}\subseteq \mathbb {Z}_{4}\), and each element a in \(\mathbb {Z}_{4}\) has a unique 2-adic expansion: a = b0 + 2b1, where \(b_{0},b_{1}\in \mathbb {Z}_{2}\).
Define \(\overline {a}=b_{0}=a\) (mod 2), and let \(\overline {a}(z)={\sum }_{i=0}^{d}\overline {a}_{i}z^{i}\in \mathbb {Z}_{2}[z]\), for any \(a(z)={\sum }_{i=0}^{d}a_{i}z^{i}\in \mathbb {Z}_{4}[z]\). Then the map − is a surjective homomorphism of rings from \(\mathbb {Z}_{4}[z]\) onto \(\mathbb {Z}_{2}[z]\). A monic polynomial a(z) in \(\mathbb {Z}_{4}[z]\) of positive degree is said to be basic irreducible if \(\overline {a}(z)\) is an irreducible polynomial in \(\mathbb {Z}_{2}[z]\). From now on, we adopt the following notation:
-
Let m be an arbitrary even positive integer, set \(q=2^{\frac {m}{2}}\) and let ς(z) be a fixed monic basic irreducible polynomial in \(\mathbb {Z}_{4}[z]\) of degree m.
-
Let \(R=\frac {\mathbb {Z}_{4}[z]}{\langle {\varsigma }(z)\rangle }=\{{\sum }_{i=0}^{m-1}a_{i}z^{i}\mid a_{0},a_{1},\ldots ,a_{m-1}\in \mathbb {Z}_{4}\}\) in which the arithmetic is done modulo ς(z). Then R is a Galois ring of characteristic 4 and 4m = q4 elements, i.e., R = GR(4,m) (cf. [46, Theorem 14.1]).
-
Let \(\mathbb {F}_{q^{2}}=\mathbb {F}_{2^{m}}=\frac {\mathbb {Z}_{2}[z]}{\langle \overline {{\varsigma }}(z)\rangle } =\left \{{\sum }_{i=0}^{m-1}b_{i}z^{i}\mid b_{0},b_{1},\ldots ,b_{m-1}\in \mathbb {Z}_{2}\right \}\) in which the arithmetic is done modulo \(\overline {{\varsigma }}(z)\). Then \(\mathbb {F}_{q^{2}}\) is a finite field of q2 elements.
-
Let \(\mathbb {F}_{q}=\{\xi \in \mathbb {F}_{q^{2}}\mid \xi ^{q}=\xi \}\subseteq \mathbb {F}_{q^{2}}\). Then \(\mathbb {F}_{q}\) is the unique subfield of \(\mathbb {F}_{q^{2}}\) with q elements (cf. [46, Theorem 6.18]). In particular, \(\mathbb {F}_{2}=\mathbb {Z}_{2}\).
As we have regarded \(\mathbb {Z}_{2}\) as a subset of \(\mathbb {Z}_{4}\), we will regard \(\mathbb {F}_{q^{2}}\) as a subset of R in the natural way, though \(\mathbb {F}_{q^{2}}\) is not a subfield of R. In this sense, we have 2 ⋅ 1 = 2 ∈ R, where \(2\in \mathbb {Z}_{4}\subseteq R\) and \(1\in \mathbb {Z}_{2}\subseteq \mathbb {F}_{q^{2}}\).
Let \(\alpha ={\sum }_{i=0}^{m-1}a_{i}z^{i}\in R\), where \(a_{i}=b_{i0}+2b_{i1}\in \mathbb {Z}_{4}\) with \(b_{i0}, b_{i1}\in \mathbb {Z}_{2}\), for all i = 0,1,…,m − 1. Then α can be uniquely expressed as: α = β0 + 2β1, where \(\beta _{j}={\sum }_{i=0}^{m-1}b_{ij}z^{i}\in \mathbb {F}_{q^{2}} \ \text {for} \ j=0,1\). Define
Then the map − is a surjective homomorphism of rings from R onto \(\mathbb {F}_{q^{2}}\) with the following kernel:
Here, we emphasize that \(\mathbb {F}_{q^{2}}\) is only regarded as a subset of R, but \(\mathbb {F}_{q^{2}}\) is not a subfield of R. Then we have \(\mathbb {F}_{q^{2}}=\overline {R}=\{\overline {\alpha }\mid \alpha \in R\}\).
As 2m = q2 and R is a Galois ring of characteristic 4 and 4m elements, by [46, Theorem 14.8], we can choose a fixed invertible element ζ of R with multiplicative order q2 − 1. From now on, we let
By [46, Theorem 14.8]), each element α of R has a unique 2-adic expansion: α = t0 + 2t1, where \(t_{0},t_{1}\in \mathcal {T}\). This implies \(\overline {R}=\overline {\mathcal {T}}\), \(2R=2\mathcal {T}\) and \(|2R|=|\mathcal {T}|=q^{2}\). Then by (1), we have \(\overline {\mathcal {T}}=\mathbb {F}_{q^{2}} \ \text {and} \ 2\mathcal {T}=2\mathbb {F}_{q^{2}}\). Moreover, we let
As the multiplicative order of ζq+ 1 is \(\frac {q^{2}-1}{q+1}=q-1\), we see that \(|\mathcal {T}_{0}|=q\) and \(\mathcal {T}_{0}=\{\xi \in \mathcal {T}\mid \xi ^{q}=\xi \}\). Hence the subset R0 of R, defined by
is the unique Galois subring of R with \(q^{2}=4^{\frac {m}{2}}\) elements. Therefore, R is a Galois extension ring of R0 with degree 2. Moreover, by \(\overline {\mathcal {T}}=\mathbb {F}_{q^{2}}\) we have
where \(\mathbb {F}_{q}\) is the subfield of \(\mathbb {F}_{q^{2}}\) with q elements. From now on, we define the map \(\phi : R\rightarrow R\) by
By [46, Theorem 14.30], we know that ϕ is the generalized Frobenius automorphism of R over R0 with multiplicative order 2 satisfying
Especially, by \(\mathbb {F}_{q}=\overline {\mathcal {T}}_{0}\subset \overline {\mathcal {T}}\), it follows that cq = c for all \(c\in \mathbb {F}_{q}\).
Now, let \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}\) be the trace function from \(\mathbb {F}_{q^{2}}\) onto \(\mathbb {F}_{q}\) defined by:
and set \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}^{-1}(c)=\left \{\alpha \in \mathbb {F}_{q^{2}}\mid \text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}(\alpha )=c\right \}\) for any \(c\in \mathbb {F}_{q}\). Then for any \(\alpha \in \mathbb {F}_{q^{2}}\), we know that \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}(\alpha )=0\) if and only if \(\alpha \in \mathbb {F}_{q}\). This implies \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}^{-1}(0)=\mathbb {F}_{q}\). Moreover, for any element \(c\in \mathbb {F}_{q}\), we have \(|\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}^{-1}(c)|=q\) (cf. [46, Corollary 7.17]).
The following lemma is one of the key results for this paper.
Lemma 1
Using the notation above, let w be a fixed element of the finite field \(\mathbb {F}_{q^{2}}\) satisfying \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}(w)=1\), i.e., w + wq = 1. Then
-
(i)
\(\mathbb {F}_{q^{2}}=\{a+bw\mid a,b\in \mathbb {F}_{q}\}\).
-
(ii)
For any \(a,b\in \mathbb {F}_{q}\), we have (a + bw)q = (a + b) + bw.
-
(iii)
Let \(c\in \mathbb {F}_{q^{2}}\). We regard 2c as an element of the Galois ring R. Then we have ϕ(2 ⋅ c) = 2 ⋅ cq. In particular, we have that ϕ(2 ⋅ c) = 2 ⋅ c if \(c\in \mathbb {F}_{q}\).
Proof
(i) By \(\text {Tr}_{\mathbb {F}_{q^{2}}/\mathbb {F}_{q}}(w)=1\), we see that \(w\not \in \mathbb {F}_{q}\). Since \(\mathbb {F}_{q^{2}}\) is an \(\mathbb {F}_{q}\)-linear space of dimension 2, the set {1,w} is an \(\mathbb {F}_{q}\)-basis of \(\mathbb {F}_{q^{2}}\). Therefore, \(\mathbb {F}_{q^{2}}=\left \{a+bw\mid a,b\in \mathbb {F}_{q}\right \}\).
(ii) By w + wq = 1, we have wq = 1 + w. As \(q=2^{\frac {m}{2}}\) and \(a,b\in \mathbb {F}_{q}\), we have aq = a and bq = b. Therefore, in the finite field \(\mathbb {F}_{q^{2}}\), we have (a + bw)q = aq + bqwq = a + b(1 + w) = (a + b) + bw.
(iii) As we regard \(\mathbb {F}_{q^{2}}\) as a subset of the Galois ring R, the element \(c\in \mathbb {F}_{q^{2}}\) has a unique 2-adic expansion c = t0 + 2t1, where \(t_{0},t_{1}\in \mathcal {T}\). This implies 2c = 2t0. From this and by the definition of ϕ, we deduce that \(\phi (2c)=2{t_{0}^{q}}\).
On the other hand, by c = t0 + 2t1 and 4 = 0 in \(\mathbb {Z}_{4}\subset R\), we have \(c^{2}={t_{0}^{2}}+4\left (t_{0}t_{1}+{t_{2}^{2}}\right )={t^{2}_{0}}\). From this and by \(q=2^{\frac {m}{2}}\), we obtain \(c^{q}={t_{0}^{q}}\).
As stated above, we conclude that \(\phi (2c)=2{t_{0}^{q}}=2c^{q}\). □
At the end of this section, we emphasize that the element w and the subset \(\mathbb {F}_{q}\) of the Galois ring GR(4,m) play important roles in the construction and representation of Hermitian self-dual cyclic codes of length 2k over GR(4,m). Specifically, the element w and subset \(\mathbb {F}_{q}\) can be constructed as follows:
-
1. Choose a monic basic irreducible polynomial ς(z) in \(\mathbb {Z}_{4}[z]\) of degree m, say \({\varsigma }(z)={\sum }_{i=0}^{m-1}{\varsigma }_{i}z^{i}+z^{m}\), where \({\varsigma }_{i}\in \mathbb {Z}_{4}\) for all i = 0,1,…,m − 1.
Then \(\overline {{\varsigma }}(z)={\sum }_{i=0}^{m-1}\overline {{\varsigma }}_{i}z^{i}+z^{m}\) is an irreducible polynomial in \(\mathbb {Z}_{2}[z]\).
-
2. Set the Galois ring \(R=\text {GR}(4,m)=\{{\sum }_{i=0}^{m-1}a_{i}z^{i}\mid a_{i}\in \mathbb {Z}_{4}, \ i=0,1,\ldots\), m − 1} in which \(z^{m}=-{\sum }_{i=0}^{m-1}{\varsigma }_{i}z^{i}=3{\sum }_{i=0}^{m-1}{\varsigma }_{i}z^{i}\).
Set the finite field \(\mathbb {F}_{q^{2}}=\{{\sum }_{i=0}^{m-1}b_{i}z^{i}\mid b_{i}\in \mathbb {Z}_{2}, \ i=0,1,\ldots ,m-1\}\) in which \(z^{m}={\sum }_{i=0}^{m-1}\overline {{\varsigma }}_{i}z^{i}\).
-
3. Choose a primitive element ζ of the finite field \(\mathbb {F}_{q^{2}}\). Then ζq+ 1 is a primitive element of the subfield \(\mathbb {F}_{q}\subset \mathbb {F}_{q^{2}}\). Hence
$$\mathbb{F}_{q}=\{0\}\cup\left\{\zeta^{l(q+1)}\mid l=0,1,\ldots, q-2\right\} \ (\text{mod} \ \overline{{\varsigma}}(z)).$$ -
4. Select a fixed \(w\in \mathbb {F}_{q^{2}}\) such that w + wq ≡ 1 (mod \(\overline {{\varsigma }}(z)\)) in \(\mathbb {Z}_{2}[z]\). Then
$$\mathbb{F}_{q^{2}}=\{\alpha+w\beta\mid \alpha,\beta\in \mathbb{F}_{q}\} \ (\text{mod} \ \overline{{\varsigma}}(z))$$and \(R=\{\xi +2\eta \mid \xi ,\eta \in \mathbb {F}_{q^{2}}\}\). Here we regard \(\mathbb {F}_{q^{2}}\) as a subset of R.
Finally, we give two examples to describe the above constructions:
Example 1
Let m = 2. Then q = 2. Choose ς(z) = z2 + z + 1. Therefore,
-
\(R=\text {GR}(4,2)=\{a_{0}+a_{1}z\mid a_{0},a_{1}\in \mathbb {Z}_{4}\}\) in which z2 = 3 + 3z.
\(\mathbb {F}_{4}=\{b_{0}+b_{1}z\mid b_{0},b_{1}\in \mathbb {Z}_{2}\}=\{0,1,z,1+z\}\) in which z2 = 1 + z.
-
ζ = z is a primitive element of \(\mathbb {F}_{4}\). Then ζq+ 1 = z3 = 1 is a primitive element of \(\mathbb {F}_{2}\). Hence \(\mathbb {F}_{2}=\{0,1\}\).
-
Let w = z. Then w2 + w ≡ 1 (mod z2 + z + 1) in \(\mathbb {Z}_{2}[z]\). Hence \(\mathbb {F}_{4}=\{\alpha +z\beta \mid \alpha ,\beta \in \mathbb {F}_{2}\}\).
Example 2
Let m = 4. Then q = 4. Choose ς(z) = z4 + z3 + 1. We have the following:
-
\(R=\text {GR}(4,4)=\{a_{0}+a_{1}z+a_{2}z^{2}+a_{3}z^{3}\mid a_{0},a_{1},a_{2},a_{3}\in \mathbb {Z}_{4}\}\) in which z4 = 3 + 3z3.
\(\mathbb {F}_{16}=\{b_{0}+b_{1}z+b_{2}z^{2}+b_{3}z^{3}\mid b_{0},b_{1},b_{2},b_{3}\in \mathbb {Z}_{2}\}\) in which z4 = 1 + z3.
-
ζ = 1 + z + z2 is a primitive element of the finite field \(\mathbb {F}_{16}\).
ζq+ 1 = (1 + z + z2)5 = 1 + z + z3 is a primitive element of \(\mathbb {F}_{4}\). Hence \(\mathbb {F}_{4}=\{0,1,\zeta ^{q+1},(\zeta ^{q+1})^{2}\}\) (mod z4 + z3 + 1), i.e.,
$$\mathbb{F}_{4}=\left\{0,1,1+z+z^{3},z+z^{3}\right\}.$$ -
Let w = z2 + z3. Then w2 + w ≡ 1 (mod z4 + z3 + 1) in \(\mathbb {Z}_{2}[z]\). Hence
$$\mathbb{F}_{16}=\left\{\alpha+\left( z^{2}+z^{3}\right)\beta\mid \alpha,\beta\in \mathbb{F}_{4}\right\}.$$
4 Hermitian self-dual cyclic codes of length 2k over GR(4,m)
In this section, we give an explicit representation for every Hermitian self-dual cyclic code over GR(4,m) of length 2k. To do this, we first review necessary concepts and facts for Euclidean and Hermitian self-dual codes.
Using the notation of Section 2, let k be any fixed positive integer and assume \(\alpha =(\alpha _{0},\alpha _{1},\ldots ,\alpha _{2^{k}-1}), \beta =(\beta _{0},\beta _{1},\ldots ,\beta _{2^{k}-1})\in R^{2^{k}}\). We let \(\phi (\beta )=(\phi (\beta _{0}),\phi (\beta _{1}),\ldots ,\phi (\beta _{2^{k}-1}))\). Recall that the Euclidean inner product [α,β]E and the Hermitian inner product [α,β]H of α and β are defined by
respectively. Then both [−,−]E and [−,−]H are nondegenerate bilinear quadratic forms on \(R^{2^{k}}\).
Let \(\mathcal {C}\) be a linear code over R of length 2k. Then the Euclidean dual code \(\mathcal {C}^{\bot _{E}}\) and the Hermitian dual code \(\mathcal {C}^{\bot _{H}}\) of \(\mathcal {C}\) are defined by
and
respectively. Both \(\mathcal {C}^{\bot _{E}}\) and \(\mathcal {C}^{\bot _{H}}\) are also linear codes over R of length 2k. As R is a Galois ring, we have \(|\mathcal {C}||\mathcal {C}^{\bot _{H}}|=|\mathcal {C}||\mathcal {C}^{\bot _{E}}|=|R|^{2^{k}}\), and so \(|\mathcal {C}^{\bot _{H}}|=|\mathcal {C}^{\bot _{E}}|\).
In particular, \(\mathcal {C}\) is said to be Hermitian self-dual (resp. Euclidean self-dual) if \(\mathcal {C}^{\bot _{H}}=\mathcal {C}\) (resp. \(\mathcal {C}^{\bot _{E}}=\mathcal {C}\)). It is well known that the number of codewords in each Hermitian (Euclidean) self-dual code over the Galois ring R of length 2k is equal to \(\left (|R|^{2^{k}}\right )^{\frac {1}{2}}=\left (\left (4^{m}\right )^{2^{k}}\right )^{\frac {1}{2}}=(2^{m})^{2^{k}}=(q^{2})^{2^{k}}\).
In this paper, we write \(\phi (\mathcal {C})=\{\phi (\alpha )\mid \alpha \in \mathcal {C}\}\subseteq R^{2^{k}}\). As ϕ is a ring automorphism on R of multiplicative order 2, by the definition of inner products [−,−]E and [−,−]H, we conclude that
The latter implies \(\phi (\mathcal {C}^{\bot _{H}})\subseteq \mathcal {C}^{\bot _{E}}\), and hence \(\phi (\mathcal {C}^{\bot _{H}})=\mathcal {C}^{\bot _{E}}\). Therefore,
Next, we review the notation and some known results for Kronecker products of matrices of specific types. These are key to the constructions in this paper. In the following, set \(q=2^{\frac {m}{2}}\) and let the field \(\mathbb {F}_{q}\) be the same as that defined in Section 2.
Let A = (aij) and B be matrices over \(\mathbb {F}_{q}\) of sizes s × t and l × v respectively. We denote by Atr the transpose of A. Recall that the Kronecker product of A and B is defined by A ⊗ B = (aijB), which is a matrix over \(\mathbb {F}_{q}\) of size sl × tv. Then we define
For any integers l and s, where 1 ≤ l ≤ 2λ and 1 ≤ s ≤ 2λ− 1, we denote by Il the identity matrix of order l and adopt the following notation:
-
Let Gl be the submatrix of size l × l in the upper left corner of \(G_{2^{\lambda }}\) and define Ml = Il + Gl, i.e.,
$$\left( \begin{array}{cc} G_{l} & 0 \\ \ast & \ast \end{array}\right)=G_{2^{\lambda}} \ \text{and} \ \left( \begin{array}{cc} M_{l} & 0 \\ \ast & \ast \end{array}\right)=I_{2^{\lambda}}+G_{2^{\lambda}}.$$(4)Then Ml is a matrix over \(\mathbb {F}_{2}\) of size l × l and \(M_{2^{\lambda }}=I_{2^{\lambda }}+G_{2^{\lambda }}\).
-
We label the rows of the matrix Ml from top to bottom as: 0th row, 1st row, …, (l − 1)st row; and label the columns of Ml from left to right as: 1st column, 2nd column, …, l th column.
For l = 2s − 1, we denote by \({{\varUpsilon }}_{j}^{[0,2s-1)}\) the j th column vector of the matrix M2s− 1, for all j = 1,2,…,2s − 1. Then \({{\varUpsilon }}_{j}^{[0,2s-1)}\in \mathbb {F}_{2}^{2s-1}\) and
$$M_{2s-1}=\left( {{\varUpsilon}}_{1}^{[0,2s-1)},{{\varUpsilon}}_{2}^{[0,2s-1)},\ldots,{{\varUpsilon}}_{2s-1}^{[0,2s-1)}\right).$$ -
For any vector \(\alpha ^{[0,2s-1)}\in \mathbb {F}_{q}^{2s-1}\), define its truncated vector α[s− 1,2s− 1) by
(5) -
Let \(\varepsilon _{2s-1}^{(2s-1)}=(0,\ldots ,0,1)^{\text {tr}}\in \mathbb {F}_{2}^{2s-1}\).
The solution space of the homogeneous linear equations with coefficient matrix Ml is determined by the following lemma, when l is odd.
Lemma 2
(cf. [11, Theorem 1]) For any positive integer s, let \(\mathcal {S}_{2s-1}\) be the solution space for the homogeneous linear equations over \(\mathbb {F}_{q}\):
Then we have the following conclusions:
-
(i)
\(\dim _{\mathbb {F}_{q}}(\mathcal {S}_{2s-1})=s\) and the following s column vectors:
$${{\varUpsilon}}_{1}^{[0,2s-1)}, {{\varUpsilon}}_{3}^{[0,2s-1)},\ldots,{{\varUpsilon}}_{2s-3}^{[0,2s-1)}, \varepsilon_{2s-1}^{(2s-1)}$$form a basis of the \(\mathbb {F}_{q}\)-linear space \(\mathcal {S}_{2s-1}\).
-
(ii)
\(\mathcal {S}_{2s-1}=\Big \{{\sum }_{i=1}^{s-1}a_{2i-1}{{\varUpsilon }}_{2i-1}^{[0,2s-1)}+a_{2s-2}\varepsilon _{2s-1}^{(2s-1)} \mid a_{2i-1}\in \mathbb {F}_{q} \ \text {for} \ \text {all} \ 1\leq i\leq s-1, \ \text {and} \ a_{2s-2}\in \mathbb {F}_{q}\Big \}\).
To determine Hermitian self-dual cyclic codes over GR(4,m) of length 2k, the following conclusion plays an essential role.
Lemma 3
Using the notation above, we set
Then \(\mathcal {S}_{2s-1}^{[s-1]}\) is an \(\mathbb {F}_{q}\)-subspace of \(\mathbb {F}_{q}^{s}\). Moreover, we have that \(\dim _{\mathbb {F}_{q}}\left (\mathcal {S}_{2s-1}^{[s-1]}\right )=\lceil \frac {s+1}{2}\rceil\) and an \(\mathbb {F}_{q}\)-basis of \(\mathcal {S}_{2s-1}^{[s-1]}\) is given by:
Therefore, we have \(|\mathcal {S}_{2s-1}^{[s-1]}|=q^{\lceil \frac {s+1}{2}\rceil }\) and
Proof
By (3) and (4), we see that M2s− 1 is a strictly lower triangle matrix and its column vectors \({{\varUpsilon }}_{1}^{[0,2s-1)},{{\varUpsilon }}_{2}^{[0,2s-1)},\ldots ,{{\varUpsilon }}_{2s-1}^{[0,2s-1)}\) satisfy the following properties: \({{\varUpsilon }}_{2s-1}^{[0,2s-1)}=\textbf {0}_{2s-1}\), and
where 0t is the zero column vector of length t for any integer t ≥ 0. Hence
From this and by Lemma 2 (i), we deduce that \(\dim _{\mathbb {F}_{q}}\left (\mathcal {S}_{2s-1}^{[s-1]}\right )=\left \lceil \frac {s+1}{2}\right \rceil\), and \(\mathcal {S}_{2s-1}^{[s-1]}= \left \{{\sum }_{\lfloor \frac {s+1}{2}\rfloor \leq i\leq s-1}a_{2i-1}{{\varUpsilon }}_{2i-1}^{[s-1,2s-1)}+a_{2s-2}\varepsilon _{s}^{(s)} \mid a_{2i-1},a_{2s-2}\in \mathbb {F}_{q}, \lfloor \frac {s+1}{2}\rfloor \leq i\leq s-1\right \}\) by Lemma 2 (ii).
□
In this paper, cyclic codes of length 2k over the Galois ring R = GR(4,m) are identified with ideals of the following ring
-
\(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }=R[x]/\langle x^{2^{k}}-1\rangle =\left \{{\sum }_{i=0}^{2^{k}-1}a_{i}x^{i}\mid a_{0},a_{1},\ldots ,a_{2^{k}-1}\in R\right \}\) in which the arithmetic is done modulo the polynomial \(x^{2^{k}}-1\),
under the identification map \(\theta : R^{2^{k}} \rightarrow \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) defined by
for all ai ∈ R and i = 0,1,…,2k − 1. Further, we set
-
\(\frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }=\mathbb {F}_{q^{2}}[x]/\langle x^{2^{k}}-1\rangle =\left \{{\sum }_{i=0}^{2^{k}-1}b_{i}x^{i}\mid b_{0},b_{1},\ldots ,b_{2^{k}-1}\in \mathbb {F}_{q^{2}}\right \}\) in which the arithmetic is done modulo \(x^{2^{k}}-1\).
As we have regarded \(\mathbb {F}_{q^{2}}\) as a subset of R, we will regard \(\frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\) as a subset of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) in the natural way, though \(\frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\) is not a subring of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). In that sense, each element ξ of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) has a unique 2-adic expansion:
This implies \(2\cdot \frac {R[x]}{\langle x^{2^{k}}-1\rangle }=2\cdot \frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle } =\left \{2\xi _{0}\mid \xi _{0}\in \frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\right \}\subset \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). Here we only regard \(\frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\) as a subset of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\).
For any polynomial \(b(x)={\sum }_{i=0}^{2^{k}-1}b_{i}x^{i}\in \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\), where bi ∈ R for all i, we define \(\phi (b(x))={\sum }_{i=0}^{2^{k}-1}\phi (b_{i})x^{i}\). Then ϕ is an automorphism of multiplicative order 2 on the ring \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). By Lemma 1 (iii), we have that
Let \(\mathcal {C}\) be an ideal of the ring \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). We set \(\phi (\mathcal {C})=\{\phi (b(x))\mid b(x)\in \mathcal {C}\},\) which is an ideal of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) as well. Hence ϕ introduces a bijection \(\mathcal {C} \mapsto \phi (\mathcal {C})\) on the set of ideals in \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\).
Let \(f(x), g(x)\in \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). In this paper, we denote by 〈f(x),g(x)〉 the ideal of the ring \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) generated by f(x) and g(x), i.e.,
Now, using the notation above and in Section 2, we determine all distinct Hermitian self-dual cyclic codes of length 2k over the Galois ring R = GR(4,m) by the following theorem. Its detailed proof is given in Section 4.
Theorem 1
For any positive integer k, we have the following:
⋄ If k = 1, there are \(1+2^{\frac {m}{2}}\) Hermitian self-dual cyclic codes of length 2 over R:
⋄ If k = 2, there are \(1+2^{\frac {m}{2}}+2^{m}\) Hermitian self-dual cyclic codes of length 22 over R:
〈2〉;
〈(x − 1)3 + 2b0,2(x − 1)〉, where \(b_{0}\in \mathbb {F}_{q}\);
〈(x − 1)2 + 2(b0 + w + b1(x − 1))〉, where \(b_{0}, b_{1}\in \mathbb {F}_{q}\).
⋄ Let k ≥ 3 and set \(q=2^{\frac {m}{2}}\). Then all distinct Hermitian self-dual cyclic codes of length 2k over R are given by the following four cases:
-
I. 1 code: 〈2〉.
-
II. q codes: \(\langle (x-1)^{2^{k}-1}+2b_{0},2(x-1)\rangle\) where \(b_{0}\in \mathbb {F}_{q}\).
-
III. For every integer s, 2 ≤ s ≤ 2k− 1 − 1, there are qs codes:
in which \(b_{s}(x)={\sum }_{j=0}^{s-1}(b_{j,0}+wb_{j,1})(x-1)^{j}\) is determined by:
where
and \(c_{2t-1},a_{2i-1},a_{2s-2}\in \mathbb {F}_{q}\), for all integers t and i: \(\lceil \frac {s+1}{2}\rceil \leq t\leq s-1\) and \(\lfloor \frac {s+1}{2}\rfloor \leq i\leq s-1\).
-
IV. \(q^{2^{k-1}}\) codes:
where
and \(a_{i},c_{i},c_{0},c_{2^{k-2}}\in \mathbb {F}_{q}\), for all i = 1,2,…,2k− 2 − 1.
Hence the number NH(GR(4,m),2k) of all Hermitian self-dual cyclic codes of length 2k over R is \(N_{\mathrm {H}}(\text {GR}(4,m),2^{k})={\sum }_{s=0}^{2^{k-1}}(2^{\frac {m}{2}})^{s} =\frac {(2^{\frac {m}{2}})^{2^{k-1}+1}-1}{2^{\frac {m}{2}}-1}\).
Remark 1
-
(i)
The formula for the number of Hermitian self-dual cyclic codes of length 2k over GR(4,m) has been given in [29, Theorem 3.4].
-
(ii)
In Theorem 1, b(x) is expressed as a polynomial in x in Case IV, while bs(x) is expressed as a polynomial in x − 1 in case III.
Finally, we give an explicit expression for every Hermitian self-dual cyclic code of length 2k over GR(4,m). For any integers K and t satisfying 1 ≤ t ≤ K, let \(\left (\begin {array}{c} K \\ t \end {array}\right )\) be the binomial coefficient defined by
Then by [13, Proposition 2], we have
where
From this, by (4) and (5), we have the following conclusion:
Theorem 2
For any integer k ≥ 3, all distinct Hermitian self-dual cyclic codes of length 2k over GR(4,m) are given by the following four cases:
-
I. 1 code: 〈2〉.
-
II. q codes: \(\langle (x-1)^{2^{k}-1}+2b_{0},2(x-1)\rangle\), where \(b_{0}\in \mathbb {F}_{q}\).
-
III. For each integer s: 2 ≤ s ≤ 2k− 1 − 1, there are qs codes:
where
and \(c_{2t-1},a_{2i-1},a_{2s-2}\in \mathbb {F}_{q}\), for all integers t and i satisfying
-
IV. \(q^{2^{k-1}}\) codes: \(\langle (x-1)^{2^{k-1}}+2b(x)\rangle ,\) where
and \(a_{i},c_{i},c_{0},c_{2^{k-2}}\in \mathbb {F}_{q}\), for all i = 1,2,…,2k− 2 − 1.
Proof
Obviously, we only need to prove the conclusion in Case III. For any integer s, 2 ≤ s ≤ 2k− 1 − 1, let j ∈{i,t}, where \(\lfloor \frac {s+1}{2}\rfloor \leq i\leq s-1\) and \(\lceil \frac {s+1}{2}\rceil \leq t\leq s-1\). By the definition for the truncated vector \({{\varUpsilon }}_{2j-1}^{[s-1,2s-1)}\) (see (5)) and (7), we have \({{\varUpsilon }}_{2j-1}^{[s-1,2s-1)}=\left (\begin {array}{c}g_{s-1,2j-1} \\ g_{s,2j-1}\\ {\vdots } \\ g_{2s-2,2j-1} \end {array}\right )\) in which gs− 1+γ,2j− 1 satisfies the following conditions:
-
(◇) If 0 ≤ γ < 2j − 1 − s, \(g_{s-1+\gamma ,2j-1}=\left (\begin {array}{c}2^{k}-(2j-1)\\ s+\gamma -(2j-1) \end {array}\right )=0\).
-
(◇) If γ = 2j − 1 − s, \(g_{s-1+\gamma ,2j-1}=\left (\begin {array}{c}2^{k}-(2j-1)\\ s+\gamma -(2j-1) \end {array}\right )+1=0\).
-
(◇) If γ = 2j − 1 − s + ν, where 1 ≤ ν ≤ 2(s − j),
$$g_{s-1+\gamma,2j-1}=\left( \begin{array}{c}2^{k}-(2j-1)\\ \nu \end{array}\right) =\left( \begin{array}{c}2^{k}-2j+1\\ \nu \end{array}\right).$$
From these and by Theorem 1, we deduce the conclusions in Case III directly. Here, we omit the trivial verification process.
□
5 Proof of Theorem 1
In this section, we prove Theorem 1. To save space, we will refer directly to some of the results in the literature later in this paper.
Lemma 4
(cf. [11, Lemma 1]) We have \((x-1)^{2^{k}}=2(x-1)^{2^{k-1}}\) in \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\).
Lemma 5
(cf. [11, Lemma 2]) Let s be an integer: 1 ≤ s ≤ 2k− 1. For any vector \(\underline {b}=(b_{0},b_{1},\ldots ,b_{s-1})^{\text {tr}}\in \mathbb {F}_{q^{2}}^{s}\), we set \(b(x)={\sum }_{j=0}^{s-1}b_{j}(x-1)^{j}\), and let \(\mathcal {C}_{\underline {b}}\) be the ideal of \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) generated by \((x-1)^{2^{k}-s}+2b(x)\) and 2(x − 1)s, i.e.,
Then we have the following:
-
(i)
The ideal \(\mathcal {C}_{\underline {b}}\) is a cyclic code of length 2k over R containing \((|R|^{2^{k}})^{\frac {1}{2}}\) codewords.
-
(ii)
We have \(\mathcal {C}_{\underline {b}}\neq \mathcal {C}_{\underline {c}}\), for any \(\underline {b}, \underline {c}\in \mathbb {F}_{q^{2}}^{s}\) satisfying \(\underline {b}\neq \underline {c}\).
As \(x^{2^{k}}=1\) in the rings \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) and \(\frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\), we have \(x^{-l}=x^{2^{k}-l}\) for all integers l, 1 ≤ l ≤ 2k − 1. Here is the key conclusion for proving Theorem 1:
Lemma 6
Using the notation of Lemma 5, if b(x) satisfies the following congruence relation in the ring \(\mathbb {F}_{q^{2}}[x]\):
where \(\phi (b(x))={\sum }_{j=0}^{s-1}{b_{j}^{q}}(x-1)^{j}\in \mathbb {F}_{q^{2}}[x]\), the code \(\mathcal {C}_{\underline {b}}\) defined by (8) is a Hermitian self-dual cyclic code of length 2k over R.
Proof
Let b(x) satisfy (9). By Lemma 5 (i), we know that \(\mathcal {C}_{\underline {b}}\) is a cyclic code of length 2k over R containing \((2^{m})^{2^{k}}=(|R|^{2^{k}})^{\frac {1}{2}}\) codewords. Moreover, by Lemma 1 (iii) and ϕ(a) = a for all a ∈ R0, it follows that
For any ideal \(\mathcal {D}\) of the ring \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\), recall that the annihilator \(\text {Ann}(\mathcal {D})\) of \(\mathcal {D}\) is defined by
Let \(\chi : \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\rightarrow \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) be the conjugate map defined by
Then it is well known that (cf. [32, Theorem 4.1])
Since b(x) satisfies (9), there exists \(g(x)\in \frac {\mathbb {F}_{q^{2}}[x]}{\langle x^{2^{k}}-1\rangle }\) such that
This implies \(2x^{-s}b(x^{-1})=2(\phi (b(x))+(x-1)^{2^{k-1}-s})+g(x)\cdot 2(x-1)^{s}\) in \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\). As x is invertible in \(\frac {R[x]}{\langle x^{2^{k}}-1\rangle }\), by − 2 = 2 in \(\mathbb {Z}_{4}\subset R\), it follows that
Moreover, by Lemma 4, we have that
Similarly, as 2k − 2s ≥ 2k − 2 ⋅ 2k− 1 = 0, we obtain
and hence
From these, we deduce that \(\phi (\mathcal {C}_{\underline {b}})\cdot \chi (\mathcal {C}_{\underline {b}})=\{0\}\). Since R is a Galois ring and \(|\chi (\mathcal {C}_{\underline {b}})|=|\phi (\mathcal {C}_{\underline {b}})|=|\mathcal {C}_{\underline {b}}|=(|R|^{2^{k}})^{\frac {1}{2}}\), we conclude that \(\text {Ann}(\phi (\mathcal {C}_{\underline {b}}))=\chi (\mathcal {C}_{\underline {b}})\).
As χ− 1 = χ, we have \((\phi (\mathcal {C}_{\underline {b}}))^{\bot _{E}}=\chi (\text {Ann}(\phi (\mathcal {C}_{\underline {b}}))) =\chi ^{2}(\mathcal {C}_{\underline {b}})=\mathcal {C}_{\underline {b}}\). This implies \(\phi (\mathcal {C}_{\underline {b}})=\mathcal {C}_{\underline {b}}^{\bot _{E}}\). From this and by the condition in (2) of Section 2, we conclude that \(\mathcal {C}_{\underline {b}}\) is a Hermitian self-dual code.
Finally, by the definition of ϕ and Lemma 1 (iii), it follows that
This implies \(\phi (b(x))={\sum }_{j=0}^{s-1}{b_{j}^{q}}(x-1)^{j}\) (mod 2).
□
By − 1 = 1 in the finite field \(\mathbb {F}_{2}\subset \mathbb {F}_{q^{2}}\), we have the following conclusion.
Lemma 7
([12, Theorem 1 (ii) and its proof]) Let l be an integer satisfying 1 ≤ l ≤ 2k − 1, and let Gl be the matrix defined by (4). Let Bl = (b0,b1,…,bl− 1)tr \(\in \mathbb {F}_{q^{2}}^{l}\) and set \(\beta (x)={\sum }_{j=0}^{l-1}b_{j}(x-1)^{j}\). Then we have
where \(x^{-1}=x^{2^{k}-1} (\text {mod} \ (x-1)^{l})\).
Lemma 8
Let k ≥ 3 be any fixed integer. For any integer s, 2 ≤ s ≤ 2k− 1 − 1, we set
Then ρs(x) satisfies (9) in Lemma 6, i.e.,
Proof
As ϕ(a) = a for all \(a\in \mathbb {Z}_{4}\subseteq R_{0}\), we have ϕ(ρs(x)) = ρs(x). Then by [11, Lemma 5]:
we conclude that \(\phi (\rho _{s}(x))+x^{-s}\rho _{s}(x^{-1})\equiv (x-1)^{2^{k-1}-s}\) (mod (x − 1)s).
□
We are now ready to prove Theorem 1 in Section 3. It is obvious that \(\langle 2\rangle =2\cdot \frac {R[x]}{\langle x^{2^{k}}-1\rangle }\) is a trivial Hermitian self-dual cyclic code of length 2k over R for any positive integer k. Then we only need to determine the nontrivial codes.
Case 1: k = 1.
In this case, by [29, Theorem 3.4], there are \(2^{\frac {m}{2}}\) nontrivial Hermitian self-dual cyclic codes of length 2 over R. As 1 ≤ s ≤ 2k− 1 = 1, we have s = 1.
Let b(x) = b0 + w, where \(b_{0}\in \mathbb {F}_{q}\) and \(q=2^{\frac {m}{2}}\). Then we have ϕ(b(x)) = b0 + 1 + w, \((x-1)^{2^{k-1}-s}=(x-1)^{0}=1\) and x− 1b(x− 1) = x− 1(b0 + w) ≡ b0 + w (mod x − 1). From these, we deduce that
Then by Lemma 6, we conclude that 〈(x − 1) + 2b(x),2(x − 1)〉 is a nontrivial Hermitian self-dual cyclic code of length 2 over R, for any \(b_{0}\in \mathbb {F}_{q}\).
Moreover, by Lemma 5, all these cyclic codes are distinct from each other. Further, by 2(x − 1) = 2((x − 1) + 2b(x)) ∈〈(x − 1) + 2b(x)〉, it follows that 〈(x − 1) + 2b(x),2(x − 1)〉 = 〈(x − 1) + 2b(x)〉.
Therefore, all Hermitian self-dual cyclic codes of length 2 over R have been given in Theorem 1.
Case 2: k ≥ 2.
As 1 ≤ s ≤ 2k− 1, we have two cases for nontrivial Hermitian self-dual cyclic codes of length 2k over R: when s = 1 and when 2 ≤ s ≤ 2k− 1.
(i) Let s = 1. For any \(b_{0}\in \mathbb {F}_{q}\), set b(x) = b0. Then ϕ(b(x)) = b0. As x ≡ 1 (mod x − 1), we get x− 1 ≡ 1 (mod x − 1). By k ≥ 2, we have 2k− 1 − 1 ≥ 1. This implies \((x-1)^{2^{k-1}-1}\equiv 0\) (mod x − 1). From these, we deduce that
Then by Lemma 6, \(\langle (x-1)^{2^{k}-1}+2b(x),2(x-1)^{s}\rangle\) is a nontrivial Hermitian self-dual cyclic code of length 2k over R. Therefore, by Lemma 5, we obtain q distinct nontrivial Hermitian self-dual cyclic codes of length 2k over R: \(\langle (x-1)^{2^{k}-1}+2b_{0},2(x-1)\rangle\), where \(b_{0}\in \mathbb {F}_{q}\).
(ii) Let 2 ≤ s ≤ 2k− 1. We further split this case into two subcases: when k = 2 and when k ≥ 3.
(ii-1) Let k = 2. Then we have s = 2, which is the only case. For any \(b_{0}, b_{1}\in \mathbb {F}_{q}\), set b(x) = b0 + w + b1(x − 1). Then we have that ϕ(b(x)) = b0 + 1 + w + b1(x − 1) = 1 + b(x) and \((x-1)^{2^{2-1}-2}=1\). Further, by \(q=2^{\frac {m}{2}}\), we have (x − 1)2 = x2 − 1. This implies x2 ≡ 1 and x− 1 ≡ x (mod (x − 1)2), and hence x− 2b(x− 1) ≡ b(x) (mod (x − 1)2). From these, we deduce that
Therefore, by Lemmas 6 and 5, we obtain q2 distinct nontrivial Hermitian self-dual cyclic codes of length 2k over R:
where \(b_{0}, b_{1}\in \mathbb {F}_{q}\), since 2(x − 1)2 = 2((x − 1)2 + 2(b0 + w + b1(x − 1))).
Now, by Lemma 5, we have obtained q + q2 distinct nontrivial Hermitian self-dual cyclic codes of length 22 over R given by Case (i) and Case (ii-1).
Moreover, by [29, Theorem 3.4], q + q2 is the number of all nontrivial Hermitian self-dual cyclic codes of length 22 over R. Hence all Hermitian self-dual cyclic codes of length 22 over R have been given in Theorem 1.
(ii-2) Let k ≥ 3 and 2 ≤ s ≤ 2k− 1. For any integer l ≥ 2, we set
Let \(\underline {b}=(b_{0},b_{1},\ldots ,b_{s-1})^{\text {tr}}\), where \(b_{j}=b_{j,0}+wb_{j,1}\in \mathbb {F}_{q^{2}}\) with \(b_{j,0},b_{j,1}\in \mathbb {F}_{q}\) for all j = 0,1,…,s − 1. Let \(B_{s;(0)}=\left (\begin {array}{c} b_{0,0}\\ b_{1,0}\\ \vdots \\ b_{s-1,0} \end {array}\right ) \ \text {and} \ B_{s;(1)}=\left (\begin {array}{c} b_{0,1}\\ b_{1,1}\\ \vdots \\ b_{s-1,1} \end {array}\right ),\) which are vectors in \(\mathbb {F}_{q}^{s}\). Then we have \(\underline {b}=B_{s;(0)}+wB_{s;(1)}\). Set
By Lemma 1 (ii), we have \({b_{j}^{q}}=b_{j,0}+(1+w)b_{j,1}=(b_{j,0}+b_{j,1})+wb_{j,1}\) for all j. Then it follows that
By 2 ≤ s ≤ 2k− 1, where k ≥ 3, we have the following two cases: (‡) 2 ≤ s ≤ 2k− 1 − 1, and (†) s = 2k− 1.
(‡) Let 2 ≤ s ≤ 2k− 1 − 1. We adopt the following notation:
-
◇ Set \(\widehat {b}(x)=\rho _{s}(x)+b(x)\), where \(\rho _{s}(x)=(x-1)^{(2^{k-1}-1)-s}\) (see Lemma 8).
-
◇ Let \(\mathcal {C}_{\widehat {b}(x)}=\langle (x-1)^{2^{k}-s}+2\widehat {b}(x),2(x-1)^{s}\rangle\), which is a cyclic code of length 2k over R by Lemma 5. Then
$$\mathcal{C}_{\widehat{b}(x)}=\langle (x-1)^{2^{k}-s}+2(x-1)^{(2^{k-1}-1)-s}+2b(x),2(x-1)^{s}\rangle.$$
Obviously, we have that \(b(x)=\rho _{s}(x)+\widehat {b}(x)\), \(\phi (\widehat {b}(x))=\phi (b(x))+\phi (\rho _{s}(x))\) and \(x^{-s}\widehat {b}(x^{-1})=x^{-s}b(x^{-1})+x^{-s}\rho _{s}(x^{-1})\). Furthermore, by Lemma 8, we have that \(\phi (\rho _{s}(x))+x^{-s}\rho _{s}(x^{-1})\equiv (x-1)^{2^{k-1}-s}\) (mod (x − 1)s). These imply
From this and by Lemma 6, we deduce that
From now on, we let
where 0t is the zero column vector of length t, for any integer t ≥ 0. Then we have \(\phi (\beta (x)) =X_{2s-1}\left (\left (\begin {array}{c} \textbf {0}_{s-1}\\ B_{s;(0)} \end {array}\right ) +\left (\begin {array}{c} \textbf {0}_{s-1}\\ B_{s;(1)} \end {array}\right ) +w\left (\begin {array}{c} \textbf {0}_{s-1}\\ B_{s;(1)} \end {array}\right )\right )\) and
by Lemma 7. On the other hand, we have
and x− 1β(x− 1) = x− 1(x− 1 − 1)s− 1b(x− 1) = (x − 1)s− 1 ⋅ x−sb(x− 1) (mod 2). From these, we deduce that
where M2s− 1 = I2s− 1 + G2s− 1, by (4).
Using (4), (5) and (6) in Section 3, we can write
where \(\widetilde {M}_{s}=\left ({{\varUpsilon }}_{s}^{[s-1,2s-1)}, {{\varUpsilon }}_{s+1}^{[s-1,2s-1)},\ldots , {{\varUpsilon }}_{2s-1}^{[s-1,2s-1)}\right )\). Therefore, the matrices Bs;(1) and Bs;(0) satisfy the following matrix equations:
if and only if \(\widetilde {M}_{s}B_{s;(1)}=\textbf {0}_{s} \ \text {and} \ \widetilde {M}_{s}B_{s;(0)}=B_{s;(1)}\).
Now, by Lemmas 2 and 3 in Section 3, it follows that
Therefore, \(\mathcal {S}_{2s-1}^{[s-1]}\) is the solution space of the system of homogeneous linear equations \(\widetilde {M}_{s}Y=\textbf {0}_{s}\), where Y = (y0,y1,…,ys− 1)tr. By \(\widetilde {M}_{s}B_{s;(0)}=B_{s;(1)}\), we see that Bs;(0) is a solution vector of the following system of linear equations:
As 2 ≤ s ≤ 2k− 1 − 1, we split this case into two subcases: when s is even and when s is odd.
(‡-1) Let s be even, where 2 ≤ s ≤ 2k− 1 − 2. Assume \(B_{s;(1)}\in \mathcal {S}_{2s-1}^{[s-1]}\). Then by Lemma 3, the vector Bs;(1) is uniquely expressed as
where \(c_{2s-2}, c_{2t-1}\in \mathbb {F}_{q}\) for all integers t: \(\frac {s}{2}\leq t\leq s-1\). Applying column elementary transformation to the augmented matrix of (11), from \(\widetilde {M}_{s}=\left ({{\varUpsilon }}_{s}^{[s-1,2s-1)}, {{\varUpsilon }}_{s+1}^{[s-1,2s-1)},\ldots , {{\varUpsilon }}_{2s-1}^{[s-1,2s-1)}\right )\), we obtain
From this, by Lemma 3 and (6) in Section 3, we deduce that there are solutions to the linear equation system (11) if and only if cs− 1 = c2s− 2 = 0. Now, let this condition be satisfied. By (12), we have
and the components of \(\widehat {\underline {c}}\) are defined by:
⊳ \(\widehat {c}_{j}=c_{2t-1}\), if j = 2t − 1 and \(\frac {s}{2}+1\leq t\leq s-1\);
⊳ \(\widehat {c}_{j}=0\), otherwise.
Then the vector \(\widehat {\underline {c}}\) is a solution of the linear equation system (11). Furthermore, since \(\mathcal {S}_{2s-1}^{[s-1]}\) is the solution space of \(\widetilde {M}_{s}Y=\textbf {0}_{s}\), \(\widehat {\underline {c}}+\mathcal {S}_{2s-1}^{[s-1]}\) must be the set of all solutions of 11), for any vector Bs,(1) given by (13).
As stated above, we obtain the following \(q^{\frac {s}{2}-1}q^{\frac {s}{2}+1}=q^{s}\) nontrivial Hermitian self-dual cyclic codes of length 2k over R:
where b(x) = Xs(Bs;(0) + wBs;(1)) is determined by
and \(c_{2t-1}\in \mathbb {F}_{q}\) for all integers t: \(\frac {s}{2}+1\leq t\leq s-1\).
(‡-2) Let s be odd, where 3 ≤ s ≤ 2k− 1 − 1. Assume \(B_{s;(1)}\in \mathcal {S}_{2s-1}^{[s-1]}\). Then by Lemma 3, the vector Bs;(1) is uniquely expressed as
where \(c_{2s-2}, c_{2t-1}\in \mathbb {F}_{q}\) for all integers t: \(\frac {s+1}{2}\leq t\leq s-1\). Applying column elementary transformation to the augmented matrix of (11), by \(\widetilde {M}_{s}= ({{\varUpsilon }}_{s}^{[s-1,2s-1)}, {{\varUpsilon }}_{s+1}^{[s-1,2s-1)},\ldots , {{\varUpsilon }}_{2s-1}^{[s-1,2s-1)})\), we obtain
From this, by Lemma 3 and (6) in Section 3, we deduce that there are solutions to the linear equation system (11) if and only if c2s− 2 = 0. Now, let this condition be satisfied. By (14), we have
and the components of \(\widehat {\underline {c}}\) are defined by: \(\widehat {c}_{j}=c_{2t-1}\), if j = 2t − 1 and \(\frac {s+1}{2}\leq t\leq s-1\); and \(\widehat {c}_{j}=0\), otherwise. Then the vector \(\widehat {\underline {c}}\) is a solution of the linear equation system (11). Furthermore, since \(\mathcal {S}_{2s-1}^{[s-1]}\) is the solution space of \(\widetilde {M}_{s}Y=\textbf {0}_{s}\), we conclude that \(\widehat {\underline {c}}+\mathcal {S}_{2s-1}^{[s-1]}\) is the set of all solutions of (11), for any vector Bs,(1) given by (15).
As stated above, we obtain the following \(q^{\frac {s-1}{2}}q^{\frac {s+1}{2}}=q^{s}\) nontrivial Hermitian self-dual cyclic codes of length 2k over R:
where \(b(x)=X_{s}(\widehat {B}_{s;(0)}+wB_{s;(1)})\) is determined by
and \(c_{2t-1}\in \mathbb {F}_{q}\) for all integers t: \(\frac {s+1}{2}\leq t\leq s-1\).
(†) Let s = 2k− 1 and set
where \(a_{i},c_{i},c_{0},c_{2^{k-2}}\in \mathbb {F}_{q}\), for all i = 1,2,…,2k− 2 − 1. As \(q=2^{\frac {m}{2}}\), we have \((x-1)^{2^{k-1}}=x^{2^{k-1}}-1\). This implies \(x^{2^{k-1}}\equiv 1\) and \(x^{-j}\equiv x^{2^{k-1}-j}\) (mod \((x-1)^{2^{k-1}}\)), and hence
From this, we deduce that \(\phi (b(x))+x^{-2^{k-1}}b(x^{-1})\equiv 1\) (mod \((x-1)^{2^{k-1}}\)). Then by Lemmas 6 and 5, we obtain \(q^{2\cdot (2^{k-2}-1)+2}=q^{2^{k-1}}\) distinct nontrivial Hermitian self-dual cyclic codes of length 2k over R:
where \(a_{i},c_{i},c_{0},c_{2^{k-2}}\in \mathbb {F}_{q}\), for all i = 1,…,2k− 2 − 1.
Summarizing the results above, we have constructed
distinct Hermitian self-dual cyclic codes of length 2k over R.
As the number of all Hermitian self-dual cyclic codes of length 2k over R is \(N_{\mathrm {H}}(\text {GR}(4,m),2^{k})=\frac {q^{2^{k-1}+1}-1}{q-1}\) (cf. [29, Theorem 3.4]), where \(q=2^{\frac {m}{2}}\), the codes listed by Theorem 1 are exactly all the distinct Hermitian self-dual cyclic codes of length 2k over R. Therefore, we have proved Theorem 1.
6 Applications
In this section, we list all distinct Hermitian self-dual cyclic codes of length 2k over the Galois ring R = GR(4,m), where m is even, using Theorem 1 or Theorem 2. To save space, we only consider the cases k = 3,4,5.
Example 3
All 1 + q + q2 + q3 + q4 Hermitian self-dual cyclic codes of length 8 over R are given by the following four cases:
-
(i)
1 code: 〈2〉.
-
(ii)
q codes: 〈(x − 1)7 + 2b0,2(x − 1)〉, where \(b_{0}\in \mathbb {F}_{q}\).
-
(iii)
q2 + q3 codes: 〈(x − 1)8−s + 2(x − 1)3−s + 2bs(x),2(x − 1)s〉, where s = 2,3 and
$$\begin{array}{@{}rcl@{}}b_{2}(x)&=&a_{1}+(a_{1}+a_{2})(x-1), a_{1},a_{2}\in\mathbb{F}_{q};\\ b_{3}(x)&=&c_{3}+(a_{3}+wc_{3})(x-1)+a_{4}(x-1)^{2}, c_{3}, a_{3},a_{4}\in\mathbb{F}_{q}. \end{array}$$ -
(iv)
q4 codes: 〈(x − 1)4 + 2b(x)〉, where
$$b(x)=(c_{0}+w)+(a_{1}+c_{1}+wa_{1})x+c_{2}x^{2}+(c_{1}+a_{1}w)x^{3} \text{and} a_{1}, c_{0}, c_{1}, c_{2}\in\mathbb{F}_{q}.$$
Example 4
All \(1+{\sum }_{s=1}^{8}q^{s}\) Hermitian self-dual cyclic codes of length 16 over R are given by the following four cases:
-
(i)
1 code: 〈2〉.
-
(ii)
q codes: 〈(x − 1)15 + 2b0,2(x − 1)〉, where \(b_{0}\in \mathbb {F}_{q}\).
-
(iii)
\({\sum }_{s=2}^{7}q^{s}\) codes: 〈(x − 1)16−s + 2(x − 1)7−s + 2bs(x),2(x − 1)s〉,
where s = 2,3,4,5,6,7 and
b2(x),b3(x) are the same as those in Example 3;
-
(iv)
q8 codes: 〈(x − 1)8 + 2b(x)〉, where
and \(a_{i},c_{i},c_{0},c_{4}\in \mathbb {F}_{q}\), for all i = 1,2,3.
Example 5
All \(1+{\sum }_{s=1}^{15}q^{s}\) Hermitian self-dual cyclic codes of length 32 over R are given by the following four cases:
-
(i)
1 code: 〈2〉.
-
(ii)
q codes: 〈(x − 1)31 + 2b0,2(x − 1)〉, where \(b_{0}\in \mathbb {F}_{q}\).
-
(iii)
\({\sum }_{s=2}^{15}q^{s}\) codes: 〈(x − 1)32−s + 2(x − 1)15−s + 2bs(x),2(x − 1)s〉, where s = 2,3,4,5,6,7,8,9,10,11,12,13,14,15 and
b2(x),b3(x),b4(x),b5(x),b6(x),b7(x) are the same as those in Example 4;
and \(a_{i},c_{2t-1}\in \mathbb {F}_{q}\) for all integers i,t: 1 ≤ i ≤ 28 and 2 ≤ t ≤ 14.
-
(iv)
q16 codes: 〈(x − 1)16 + 2b(x)〉, where
and \(a_{i},c_{i},c_{0},c_{8}\in \mathbb {F}_{q}\), for all i = 1,2,3,4,5,6,7.
Remark 2
Let the Galois rings GR(4,2) and GR(4,4) be constructed by Examples 1 and 2, respectively. Then by Examples 4 and 5, we obtain:
◇ 511 distinct Hermitian self-dual cyclic codes of length 16 over GR(4,2);
◇ 87381 distinct Hermitian self-dual cyclic codes of length 16 over GR(4,4);
◇ 131071 distinct Hermitian self-dual cyclic codes of length 32 over GR(4,2);
◇ 5726623061 distinct Hermitian self-dual cyclic codes of length 32 over GR(4,4).
Finally, we give an example of the construction of self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 6. First, we have that z3 − 1 = (z − 1)(z2 + z + 1), where z − 1 and z2 + z + 1 correspond to the 2-cyclotomic cosets modulo 3 S2(0) = {0} and S2(1) = {1,2} respectively. Obviously, S2(1) is self-inverse.
Let \(\text {GR}(4,2)=\frac {\mathbb {Z}_{4}[z]}{\langle z^{2}+z+1\rangle }\). As \(\mathbb {Z}_{4}=\frac {\mathbb {Z}_{4}[z]}{\langle z-1\rangle }\), we have the following isomorphism of rings from \(\mathbb {Z}_{4}\times \text {GR}(4,2)\) onto \(\frac {\mathbb {Z}_{4}[z]}{\langle z^{3}-1\rangle }\) defined by:
for all \(b,a_{0},a_{1}\in \mathbb {Z}_{4}\). Hence \(\mathcal {C}\) is a self-dual cyclic code over \(\mathbb {Z}_{4}\) of length 6 if and only if \(\mathcal {C}\cong C_{0}\times C_{1}\), where C0 (as an ideal of \(\frac {\mathbb {Z}_{4}[y]}{\langle y^{2}-1\rangle }\)) is an Euclidean self-dual cyclic code over \(\mathbb {Z}_{4}\) of length 2 and C1 (as an ideal of \(\frac {\text {GR}(4,2)[y]}{\langle y^{2}-1\rangle }\)) is a Hermitian self-dual cyclic code over GR(4,2) of length 2.
By [11, Theorem 2] and Theorem 1 in Section 4, we give all 3 self-dual cyclic codes \(\mathcal {C}\) over \(\mathbb {Z}_{4}\) of length 6 by the following table:
C 0 | C 1 | \(\mathcal {C}\) (as ideals of the ring \(\frac {\mathbb {Z}_{4}[x]}{\langle x^{6}-1\rangle }\)) |
---|---|---|
〈2〉 | 〈2〉 | 〈2〉 |
〈2〉 | 〈(y − 1) + 2z〉 | 〈2 + x + 3x2 + 2x3 + x4 + x5〉 |
〈2〉 | 〈(y − 1) + 2(1 + z)〉 | 〈2 + x + x2 + 2x3 + 3x4 + x5〉 |
7 Conclusions and further work
For any positive integers m and k, where m is even, we have given a direct and effective approach to construct all distinct Hermitian self-dual cyclic codes of length 2k over the Galois ring GR(4,m) precisely. In particular, using binomial coefficients, we have provided an explicit expression to accurately represent this class of Hermitian self-dual cyclic codes over GR(4,m).
Theoretically, using the results in [11, 22] and this paper, any self-dual cyclic code over \(\mathbb {Z}_{4}\) of arbitrary even length can be constructed by the Discrete Fourier Transform decomposition given in [29]. This approach is, however, not easy to be implemented in practice.
A natural extension of this work will be to give directly an explicit representation for all self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 2kn, for any positive odd integer n. Moreover, it would be interesting to investigate the parameters of these codes and obtain good self-dual cyclic \(\mathbb {Z}_{4}\)-codes and formally self-dual binary codes, using the representations obtained.
References
Abualrub, T., Oehmke, R.: On the generators of \(\mathbb {Z}_{4}\) cyclic codes of length 2e. IEEE Trans. Inform. Theory 49, 2126–2133 (2003)
Bachoc, C.: Application of coding theory to the construction of modular lattices. J. Combin. Theory Ser. A 78, 92–119 (1997)
Blackford, T., Ray-Chaudhuri, D.K.: A transform approach to permutation groups of cyclic codes over Galois rings. IEEE Trans. Inform. Theory 46 (7), 2350–2358 (2000)
Blackford, T.: Cyclic codes over \(\mathbb {Z}_{4}\) of oddly even length. Discret. Appl. Math. 128, 27–46 (2003)
Blackford, T.: Negacyclic codes over \(\mathbb {Z}_{4}\) of even length. IEEE Trans. Inform. Theory 49(6), 1417–1424 (2003)
Bonnecaze, A., Rains, E., Solé, P.: 3-colored 5-designs and Z4,-codes. J. Statist. Plann. Inference 86(2), 349–368 (2000). Special issue in honor of Professor Ralph Stanton. MR 1768278 (2001g:05021)
Calderbank, A.R., Sloane, N.J.A.: Modular and p-adic cyclic codes. Des. Codes Cryptogr. 6(1), 21–35 (1995)
Calderbank, A.R., Sloane, N.J.A.: Double circulant codes over \(\mathbb {Z}_{4}\) and even unimodular lattices. J. Alg. Combin. 6, 119–131 (1997)
Cao, Y., Cao, Y., Dougherty, S.T., Ling, S.: Construction and enumeration for self-dual cyclic codes over \(\mathbb {Z}_{4}\) of oddly even length. Des. Codes Cryptogr. 87, 2419–2446 (2019)
Cao, Y., Cao, Y., Fu, F.-W., Wang, G.: Self-dual cyclic codes over \(\mathbb {Z}_{4}\) of length 4n. Appl. Algebra in Engrg. Comm. Comput. 33, 21–51 (2022)
Cao, Y., Cao, Y., Ling, S., Wang, G.: An explicit expression for Euclidean self-dual cyclic codes of length 2k over Galois ring GR(4,m). Finite Fields Appl. 72, 101817 (2021)
Cao, Y., Cao, Y., Dinh, H.Q., Jitman, S.: An explicit representation and enumeration for self-dual cyclic codes over \(\mathbb {F}_{2^{m}}+u\mathbb {F}_{2^{m}}\) of length 2s. Discrete Math. 342, 2077–2091 (2019)
Cao, Y., Cao, Y., Dinh, H.Q., Jitman, S.: An efficient method to construct self-dual cyclic codes of length ps over \(\mathbb {F}_{p^{m}}+u\mathbb {F}_{p^{m}}\). Discret. Math. 343, 111868 (2020)
Castagnoli, G., Massey, J.L., Schoeller, P.A., von Seemann, N.: On repeated-root cyclic codes. IEEE Trans. Inform. Theory 37(2), 337–342 (1991)
Dinh, H.Q., López-Permouth, S.R.: Cyclic and negacyclic codes over finite chain rings. IEEE Trans. Inform. Theory 50(8), 1728–1744 (2004)
Dinh, H.Q.: Negacyclic codes of length 2s over Galois rings. IEEE Trans. Inform. Theory 51(12), 4252–4262 (2005)
Dinh, H.Q.: Complete distances of all negacyclic codes of length 2s over \(\mathbb {Z}_{2^{a}}\). IEEE Trans. Inform. Theory 53(1), 147–161 (2007)
Dinh, H.Q.: On the linear ordering of some classes of negacyclic and cyclic codes and their distance distributions. Finite Fields Appl. 14(1), 22–40 (2008)
Dinh, H.Q.: Constacyclic codes of length 2s over Galois extension rings of \(\mathbb {F}_{2} + u\mathbb {F}_{2}\). IEEE Trans. Inform. Theory 55(4), 1730–1740 (2009)
Dinh, H.Q.: Constacyclic codes of length ps over \(\mathbb {F}_{p^{m}} + u\mathbb {F}_{p^{m}}\). J. Algebra 324(5), 940–950 (2010)
Dougherty, S.T., Park, Y.H.: On modular cyclic codes. Finite Fields Appl. 13, 31–57 (2007)
Dougherty, S.T., Ling, S.: Cyclic codes over \(\mathbb {Z}_{4}\) of even length. Des. Codes Cryptogr. 39, 127–153 (2006)
Gaborit, P., Natividad, A.M., Solé, P.: Eisenstein lattices, Galois rings and quaternary codes. Int. J. Number Theory 2, 289–303 (2006)
Hammons, Jr.A. R., Kumar, P.V., Calderbank, A.R., Sloane, N.J.A., Solé, P.: The Z4-linearity of Kerdock, Preparata, Goethals, and related codes. IEEE Trans. Inform. Theory 40(2), 301–319 (1994)
Harada, M.: Self-dual \(\mathbb {Z}_{4}\)-codes and Hadamard matrices. Discret. Math. 245, 273–278 (2002)
Harada, M., Kitazume, M., Munemasa, A., Venkov, B.: On some self-dual codes and unimodular lattices in dimension 48. Eur. J. Combin. 26, 543–557 (2005)
Harada, M., Miezaki, T.: An optimal odd unimodular lattice in dimension 72. Arch. Math. 97(6), 529–533 (2011)
Harada, M., Solé, P., Gaborit, P.: Self-dual codes over \(\mathbb {Z}_{4}\) and unimodular lattices: a survey. In: Algebras and Combinatorics, Hong Kong, 1997, pp 255–275. Springer, Singapore (1999)
Jitman, S., Ling, S., Sangwisut, E.: On self-dual cyclic codes of length pa over GR(p2,s). Adv. Math. Commun. 10, 255–273 (2016)
Kai, X., Zhu, S.: On the distance of cyclic codes of length 2e over \(\mathbb {Z}_{4}\). Discret. Math. 310(1), 12–20 (2010)
Kaya, A., Yildiz, B.: Various constructions for self-dual codes over rings and new binary self-dual codes. Discret. Math. 339, 460–469 (2016)
Kiah, H.M., Leung, K.H., Ling, S.: Cyclic codes over GR(p2,m) of length pk. Finite Fields Appl. 14, 834–846 (2008)
Kiah, H.M., Leung, K.H., Ling, S.: A note on cyclic codes over GR(p2,m) of length pk. Des. Codes Cryptogr. 63, 105–112 (2012)
Kiermaier, M.: There is no self-dual Z4,-linear codes whose Gray image has the parameter (72, 236, 16). IEEE Trans. Inform. Theory 59(6), 3384–3386 (2013)
Kanwar, P., López-Permouth, S.R.: Cyclic codes over the integers modulo pm. Finite Fields Appl. 3(4), 334–352 (1997)
Liu, H., Youcef, M.: Some repeated-root constacyclic codes over Galois rings. IEEE Trans. Inform. Theory 63(10), 6247–6255 (2017)
López-Permouth, S.R., Szabo, S.: On the Hamming weight of repeated root cyclic and negacyclic codes over Galois rings. Adv. Math. Commun. 3(4), 409–420 (2009)
Massey, J.L., Costello, D.J., Justesen, J.: Polynomial weights and code constructions. IEEE Trans. Inform. Theory 19, 101 (1973)
Norton, G.H., Ana Sălăgean, A.: On the structure of linear and cyclic codes over a finite chain ring. Appl. Algebra Engrg. Comm. Comput. 10(6), 489–506 (2000)
Pless, V.S., Qian, Z.: Cyclic codes and quadratic residue codes over \(\mathbb {Z}_{4}\). IEEE Trans. Inform. Theory 42(5), 1594–1600 (1996)
Roth, R.M., Seroussi, G.: On cyclic MDS codes of length q over GF(q). IEEE Trans. Inform. Theory 32(2), 284–285 (1986)
Sălăgean, A.: Repeated-root cyclic and negacyclic codes over a finite chain ring. Discret. Appl. Math. 154(2), 413–419 (2006)
Shi, M., Qian, L., Sok, L., Aydin, N., Solé, P.: On constacyclic codes over \(\mathbb {Z}_{4}[u]/\langle u^{2}-1\rangle\) and their Gray images. Finite Fields Appl. 45, 86–95 (2017)
Vega, G., Wolfmann, J.: Some families of \(\mathbb {Z}_{4}\)-cyclic codes. Finite Fields Appl. 10(4), 530–539 (2004)
van Lint, J.H.: Repeated-root cyclic codes. IEEE Trans. Inform. Theory 37(2), 343–345 (1991)
Wan, Z.-X.: Lectures on finite fields and Galois rings. World Scientific Pub Co Inc (2003)
Wolfmann, J.: Negacyclic and cyclic codes over \(\mathbb {Z}_{4}\). IEEE Trans. Inform. Theory 45(7), 2527–2532 (1999)
Wolfmann, J.: Binary images of cyclic codes over \(\mathbb {Z}_{4}\). IEEE Trans. Inform. Theory 47(5), 1773–1779 (2001)
Zimmermann, K.-H.: On generalizations of repeated-root cyclic codes. IEEE Trans. Inform. Theory 42(2), 641–649 (1996)
Acknowledgements
This research is supported in part by the National Natural Science Foundation of China (Grant Nos. 12071264, 11801324, 11671235), the Shandong Provincial Natural Science Foundation, China (Grant No. ZR2018BA007), Nanyang Technological University, Singapore (Grant No. 04INS000047C230GRT01), the IC Program of Shandong Institutions of Higher Learning For Youth Innovative Talents and the Scientific Research Fund of Hubei Provincial Key Laboratory of Applied Mathematics (Hubei University)(Grant Nos. HBAM201906).
Funding
Not available.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Cao, Y., Cao, Y., Ling, S. et al. On the construction of self-dual cyclic codes over \(\mathbb {Z}_{4}\) with arbitrary even length. Cryptogr. Commun. 14, 1117–1143 (2022). https://doi.org/10.1007/s12095-022-00579-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12095-022-00579-2