Abstract
For any positive integer m > 2 and an odd prime p, let \(\mathbb {F}_{p^{m}}\) be the finite field with pm elements and let \( \text {Tr}^{m}_{e}\) be the trace function from \(\mathbb {F}_{p^{m}}\) onto \(\mathbb {F}_{p^{e}}\) for a divisor e of m. In this paper, for the defining set \(D=\{x\in \mathbb {F}_{p^{m}}:\text {Tr}^{m}_{e}(x)=1\text { and } \text {Tr}^{m}_{e}(x^{2})=0\}=\{d_{1}, d_{2}, \ldots , d_{n}\}\)(say), we define a pe-ary linear code \(\mathcal {C}_{D}\) by
Then we determine the complete weight enumerator and weight distribution of the linear code \(\mathcal {C}_{D}\). The presented code is optimal with respect to the Griesmer bound provided that \(\frac {m}{e}=3\). In fact, it is MDS when \(\frac {m}{e}=3\). This paper gives the results of S. Yang, X. Kong and C. Tang (Finite Fields Appl. 48 (2017)) if we take e = 1. In addition to the generalization of the results of Yang et al., we study the dual code \(\mathcal {C}_{D}^{\perp }\) of the code \(\mathcal {C}_{D}\) as well as find some optimal constant composition codes.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Throughout this paper, let p be an odd prime, and let m = es, where m, e and s (> 2) are positive integers. \(\mathbb {F}_{p^{m}}\) denotes a finite field with pm elements. The trace function from \(\mathbb {F}_{p^{m}}\) onto \(\mathbb {F}_{p^{e}}\) is denoted by \(\text {Tr}^{m}_{e}\). Moreover, the absolute trace functions of \(\mathbb {F}_{p^{m}}\) and \(\mathbb {F}_{p^{e}}\) are denoted by \(\text {Tr}^{m}_{1}\) and \(\text {Tr}^{e}_{1}\) respectively. An (n, M) code over \(\mathbb {F}_{p^{e}}\) is a subset of \(\mathbb {F}^{n}_{p^{e}}\) of size M. A linear code \(\mathcal {C}\) of length n over \(\mathbb {F}_{p^{e}}\) is a subspace of \(\mathbb {F}^{n}_{p^{e}}\). An [n, k, d] linear code \(\mathcal {C}\) over \(\mathbb {F}_{p^{e}}\) is a k-dimensional subspace of \(\mathbb {F}^{n}_{p^{e}}\) with minimum distance d. The members of the code \(\mathcal {C}\) are known as codewords. The number of nonzero coordinates in \( \textbf {c}\in \mathcal {C} \) is called the Hamming-weight wt(c) of the codeword c. Let Ai denote the number of codewords with Hamming weight i in a linear code \(\mathcal {C}\) of length n. The weight enumerator of the code \(\mathcal {C}\) is a polynomial defined by
where (1,A1,…,An) is called the weight distribution of the code \(\mathcal {C}\). There is much literature on the weight distribution of some special linear codes [2, 5, 6, 11, 13]. The complete weight enumerator of a linear code \(\mathcal {C}\) gives the frequency of each symbol contained in each codeword (see [18]). Assume \(\mathbb {F}_{p^{e}}=\{w_{0},w_{1},\ldots , w_{p^{e}-1}\}\), where w0 = 0. Composition of a vector v=\((v_{0}, v_{1},\ldots , v_{n-1})\in \mathbb {F}_{p^{e}}^{n}\), denoted by comp(v), is defined as
where kj is the number of components vi(0 ≤ i ≤ n − 1) of v that equal wj. It is obvious that \({\sum }_{j=0}^{p^{e}-1}k_{j}=n\). Let \(A(k_{0},k_{1},\ldots ,k_{p^{e}-1})\) be the number of codewords \(\textbf {c}\in \mathcal {C}\) with comp(c)=\((k_{0},k_{1},\ldots ,k_{p^{e}-1})\). Then the complete weight enumerator of the code \(\mathcal {C}\) is the polynomial
where \(\mathbb {B}_{n}=\{(k_{0},k_{1},\ldots ,k_{p^{e}-1}):0\leq k_{j}\leq n,{\sum }_{j=0}^{p^{e}-1}k_{j}=n\}\). The complete weight enumerators of linear codes not only give the weight enumerators but also demonstrate the frequency of each symbol appearing in each codeword. Consequently, the complete weight enumerators of linear codes have been of fundamental importance to theories and practices. Recently, linear codes with their complete weight enumerators have been studied extensively. Ding et al. in [7, 8] showed that complete weight enumerators can be applied to the calculation of the deception probabilities of certain authentication codes. Constructions of some families of optimal constant composition codes and the complete weight enumerators of some constant composition codes were given in [3, 4, 9].
In [1, 12, 16, 21,22,23,24], authors constructed linear codes with their complete weight enumerators over \(\mathbb {F}_{p}\) by employing absolute trace function. Construction of linear codes over \(\mathbb {F}_{p^{e}}\) by considering \(\text {Tr}_{e}^{m}\) in place of \(\text {Tr}_{1}^{m}\) result in improved relative minimum distance of the codes compared with [13, 22] (see Remarks 3.13 and 3.21, Tables 9 and 10).
In the present work, we find linear codes over \(\mathbb {F}_{p^{e}}\) by considering new defining set obtained by replacing Tr by \(\text {Tr}_{e}^{m}\) in the defining set D given in [22]. Now, we define the trace function from \(\mathbb {F}_{p^{m}}\) onto \(\mathbb {F}_{p^{e}}\) as follows:
Set \(D=\{x\in \mathbb {F}_{p^{m}}:\text {Tr}^{m}_{e}(x)=1~\text {and}~\text {Tr}^{m}_{e}(x^{2})=0\}=\{d_{1},d_{2},\ldots ,d_{n}\}\). We define a linear code \(\mathcal {C}_{D}\) associated with D by
We determine the complete weight enumerator and weight distribution of the linear code \(\mathcal {C}_{D}\) of (1). We show that the several constructed linear codes are optimal with respect to the Griesmer bound. In fact, the constructed optimal codes are MDS. In [22], it is shown that there exists a unique MDS code when m = 3 while we have shown that there are infinitely many MDS codes when s = 3. Moreover, we have constructed some optimal dual codes of the codes defined by (1), and finally we have shown an application to constant composition codes.
Rest of the paper is organized as follows. In Section 2, we give some definitions and results on cyclotomic numbers and Gauss sums over finite fields. In Section 3.1, we present the complete weight enumerator and weight distribution of the proposed linear code \(\mathcal {C}_{D}\). Some examples to illustrate our main results also discussed in Section 3.1. In Section 3.2, we give some optimal dual codes. In Section 4, we have shown an application of complete weight enumerator to constant composition codes. Section 5 concludes the paper with some concluding remarks.
2 Preliminaries
We begin with some preliminaries by introducing the concept of cyclotomic numbers. Let a be a primitive element of \(\mathbb {F}_{p^{m}}\), and let pm = Nh + 1 for two positive integers N > 1 and h > 1. The cyclotomic classes of order N in \(\mathbb {F}_{p^{m}}\) are the cosets \(\mathcal {C}_{i}^{(N,p^{m})}=a^{i}\langle a^{N}\rangle \) for i = 0,1,…,N − 1, where 〈aN〉 denotes the subgroup of \(\mathbb {F}_{p^{m}}^{*}\) generated by aN. It is obvious that \(\#\mathcal {C}_{i}^{(N, p^{m})}=h\), where #X, for any set X, denotes the cardinality of the set X. For fixed i and j, we define the cyclotomic number \((i,j)^{(N,p^{m})}\) to be the number of solutions of the equation
where 1 = a0 is the multiplicative identity of \(\mathbb {F}_{p^{m}}\). That is, \((i,j)^{(N,p^{m})}\) is the number of ordered pairs (s, t) such that
Now, we present, from [14], some notions and results about group characters and Gauss sums for later use. An additive character χ of \(\mathbb {F}_{p^{m}}\) is a mapping from \(\mathbb {F}_{p^{m}}\) into the multiplicative group of complex numbers of absolute value 1 with χ (g1 + g2) = χ (g1)χ(g2) for all \( g_{1}, g_{2}\in \mathbb {F}_{p^{m}}\). By ([14], Theorem 5.7), for any \( b\in \mathbb {F}_{p^{m}} \),
defines an additive character of \( \mathbb {F}_{p^{m}} \), where \(\zeta _{p}=e^{\frac {2\pi \sqrt {-1}}{p}}\), and every additive character can be obtained in this way. An additive character defined by χ0(x) = 1 for all \( x\in \mathbb {F}_{p^{m}}\) is called the trivial character while all other characters are called nontrivial characters. The character χ1 in (2) is called the canonical additive character of \( \mathbb {F}_{p^{m}}\). The orthogonal property of additive character χ of \( \mathbb {F}_{p^{m}} \) can be found in ([14], Theorem 5.4) and is given as
Characters of the multiplicative group \( \mathbb {F}^{*}_{p^{m}}\) of \(\mathbb {F}_{p^{m}} \) are called multiplicative characters of \( \mathbb {F}_{p^{m}} \). By ([14], Theorem 5.8), for each j = 0,1,…,pm − 2, the function ψj with
defines a multiplicative character of \( \mathbb {F}_{p^{m}} \), where g is a generator of \( \mathbb {F}^{*}_{p^{m}} \). For \( j=\frac {p^{m}-1}{2} \), we have the quadratic character \( \eta =\psi _{\frac {p^{m}-1}{2}} \) defined by
Moreover, we extend this quadratic character by letting η(0) = 0. The quadratic Gauss sum G = G(η, χ1) over \( \mathbb {F}_{p^{m}} \) is defined by
Now, let \( \overline {\eta } \) and \( \overline {\chi }_{1} \) denote the quadratic and canonical character of \( \mathbb {F}_{p^{e}} \) respectively. Then we define quadratic Gauss sum \( \overline {G}=G(\overline {\eta }, \overline {\chi }_{1}) \) over \( \mathbb {F}_{p^{e}} \) by
The explicit values of quadratic Gauss sums are given by the following lemma.
Lemma 2.1
[14, Theorem 5.15] Let the symbols have the same meanings as before. Then
Lemma 2.2
[20] Let the notations have the same significations as before. Then, for N = 2, the cyclotomic numbers are given by:
-
1.
h even: \((0,0)^{(2,p^{m})}=\frac {h-2}{2}\), \((0,1)^{(2,p^{m})}=(1,0)^{(2,p^{m})}=(1,1)^{(2,p^{m})}=\frac {h}{2}\);
-
2.
h odd: \((0,0)^{(2,p^{m})}=(1,0)^{(2,p^{m})}=(1,1)^{(2,p^{m})}=\frac {h-1}{2}\), \((0,1)^{(2,p^{m})}=\frac {h+1}{2}\).
Lemma 2.3
[16, Lemma 2] Let η and \(\overline {\eta }\) be the quadratic characters of \(\mathbb {F}_{p^{m}}^{*}\) and \(\mathbb {F}_{p^{e}}^{*}\) respectively. Then the following assertions hold:
-
1.
if s ≥ 2 is even, then η(y) = 1 for each \(y\in \mathbb {F}^{*}_{p^{e}} \);
-
2.
if s is odd, then \( \eta (y)=\overline {\eta }(y) \) for each \(y\in \mathbb {F}^{*}_{p^{e}} \).
Lemma 2.4
[14, Theorem 5.33] Let χ be a non-trivial additive character of \( \mathbb {F}_{p^{m}} \) and let \(f(x)=a_{2}x^{2} + a_{1}x + a_{0}\in \mathbb {F}_{p^{m}}[x]\) with a2≠ 0. Then
Lemma 2.5
[14, Theorem 2.26] Let \( \text {Tr}^{m }_{1}\) and \( \text {Tr}^{e}_{1}\) be the absolute trace functions of \( \mathbb {F}_{p^{m}} \) and \( \mathbb {F}_{p^{e}} \) respectively, and let \( \text {Tr}^{m}_{e}\) be the trace function from \( \mathbb {F}_{p^{m}} \) onto \( \mathbb {F}_{p^{e}}\). Then
for all \( x\in \mathbb {F}_{p^{m}}\).
Lemma 2.6
[10, Griesmer bound] Let \(\mathcal {C}\) be an [n, k, d] linear code over \( \mathbb {F}_{p^{e}} \), where k ≥ 1. Then
where the symbol \(\left \lceil {x}\right \rceil \) denotes the smallest integer not less than x.
3 Main results
We divide this section into two subsections, namely 3.1 and 3.2.
3.1 Determination of the complete weight enumerator of \(\mathcal {C}_{D}\)
In this subsection, after proving some lemmas, we determine the complete weight enumerator and weight distribution of \(\mathcal {C}_{D}\) defined by (1). It is clear that the length n of the linear code \(\mathcal {C}_{D}\) is equal to # D which can be found in the following lemma.
Lemma 3.1
For \( \lambda ,\mu \in \mathbb {F}_{p^{e}}\), define
Then
-
1.
if 2∣s and p∣s, we have
$$N(\lambda, \mu)= \left\{\begin{array}{ll} p^{m-2e}+p^{-e}(p^{e}-1)G, &\text{if}~\lambda=0~\text{and}~\mu=0,\\ p^{m-2e} , & \text{if}~\lambda=0~\text{and}~\mu\neq0,\\ p^{m-2e}-p^{-e}G, &\text{if}~\lambda\neq0~\text{and}~\mu=0,\\ p^{m-2e}, &\text{if}~\lambda\neq0~\text{and}~\mu\neq0. \end{array}\right.~~~~~~~~~~~$$ -
2.
if 2∣s and \(p\nmid s \), we have
$$N(\lambda,\mu)= \left\{\begin{array}{ll} p^{m-2e},& \text{if}~\lambda=0~\text{and}~\mu=0,\\ p^{m-2e}+p^{-e}G, & \text{if}~\lambda=0~\text{and}~\mu\neq0,\\ p^{m-2e}, & \text{if}~\lambda\neq0~\text{and}~\mu^{2}-s\lambda=0,\\ p^{m-2e}+\overline{\eta}(\mu^{2}-s\lambda)p^{-e}G, & \text{if}~\lambda\neq0~\text{and}~\mu^{2}-s\lambda\neq0. \end{array}\right.$$ -
3.
if \(2\nmid s\) and p∣s, we have
$$ N(\lambda, \mu)= \left\{\begin{array}{ll} p^{m-2e} , & \text{if}~\lambda=0,\\ p^{m-2e}+\overline{\eta}(-\lambda)p^{-e}G\overline{G}, &\text{if}~\lambda\neq0~\text{and}~\mu=0,\\ p^{m-2e}, &\text{if}~\lambda\neq0~\text{and}~\mu\neq0. \end{array}\right.~~~~~~~~~~$$ -
4.
if \(2\nmid s\) and \(p\nmid s \), we have
$$ N(\lambda,\mu)= \left\{\begin{array}{ll} p^{m-2e}+\overline{\eta}(-s)p^{-2e}(p^{e}-1)G\overline{G},& \text{if}~\mu^{2}-s\lambda=0,\\ p^{m-2e}-\overline{\eta}(-s)p^{-2e}G\overline{G},& \text{if}~\mu^{2}-s\lambda\neq0. \end{array}\right.~~~~~$$
Proof
By the properties of additive character and Lemma 2.5, we have
where
By using Lemma 2.4, it is easy to prove that
By Lemma 2.4, we have
and there are the following four cases to consider: Case 1: Suppose 2∣s and p∣s. Then
Case 2: Suppose 2∣s and \( p\nmid s\). Then, by Lemma 2.4, we have
Case 3: Next, let \( 2\nmid s\) and p∣s. Then
Case 4: Finally, let \( 2\nmid s\) and \( p\nmid s\). Then, by Lemma 2.4, we have
Combining (4) and the values of S1, S2 and S3, the proof of the lemma is completed. □
Lemma 3.2
For \( a\in \mathbb {F}_{p^{m}}^{*} \) and \( c\in \mathbb {F}_{p^{e}}^{*} \), let
Then
Proof
We have
One may easily verify that the equation aw + y = 0 has solutions if and only if \(a\in \mathbb {F}_{p^{e}}^{*} \) for \( y,w\in \mathbb {F}_{p^{e}}^{*} \). Then ℵ2 = 0 if \(a\notin \mathbb {F}_{p^{e}}^{*} \). Hence, if \(a\in \mathbb {F}_{p^{e}}^{*} \), we have
By using orthogonal property of additive character, we get the lemma.□
Let \(a\in \mathbb {F}^{*}_{p^{m} }\) and \(c\in \mathbb {F}^{*}_{p^{e}}\), and let \(s=\text {Tr}^{m}_{e}(1)\). For the sake of simplicity, we denote \(\nabla =(\text {Tr}^{m}_{e}(a))^{2}-s\text {Tr}^{m}_{e}(a^{2})\) and \(\phi (c)=-sc^{2}+2c\text {Tr}^{m}_{e}(a) -\text {Tr}^{m}_{e}(a^{2})\) in the rest of the paper.
Lemma 3.3
For \( a\in \mathbb {F}^{*}_{p^{m}} \) and \( c\in \mathbb {F}^{*}_{p^{e}} \), let
Then
Proof
By Lemmas 2.4 and 2.5, we have
as required.□
Lemma 3.4
For \( a\in \mathbb {F}_{p^{m}}^{*} \) and \( c\in \mathbb {F}_{p^{e}}^{*} \), let
Then we have the following cases:
-
1.
If 2∣s and p∣s, then
$$\aleph_{4}=\left\{\begin{array}{ll} (p^{e}-1)G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})=0~\text{and}~\text{Tr}^{m}_{e}(a)=0,\\ -G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})=0~\text{and}~\text{Tr}^{m}_{e}(a)\neq0,\\ -G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0~\text{and}~\text{Tr}^{m}_{e}(a)=0,\\ (p^{2e}-p^{e}-1)G,&\text{if}~\text{Tr}^{m}_{e}(a^{2})\text{Tr}^{m}_{e}(a)\neq0~\text{and}~\text{Tr}^{m}_{e}(a^{2})=2c\text{Tr}^{m}_{e}(a),\\ -(p^{e}+1)G,&\text{if}~\text{Tr}^{m}_{e}(a^{2})\text{Tr}^{m}_{e}(a)\neq0~\text{and}~\text{Tr}^{m}_{e}(a^{2})\neq2c\text{Tr}^{m}_{e}(a). \end{array}\right. $$ -
2.
If 2∣s and \(p\nmid s\), then
$$\aleph_{4}=\left\{\begin{array}{ll} (p^{2e}-p^{e}-1)G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla=0~\text{and}~c\text{Tr}^{m}_{e}(a)=\text{Tr}^{m}_{e}(a^{2}),\\ (p^{2e}-p^{e}-1)G,&\text{if }\text{Tr}^{m}_{e}(a^{2})=0,~\text{Tr}^{m}_{e}(a)\neq0\text{ and }cs= 2\text{Tr}^{m}_{e}(a),\\ -(p^{e}+1)G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla=0~\text{and}~c\text{Tr}^{m}_{e}(a)\neq \text{Tr}^{m}_{e}(a^{2}),\\ -(p^{e}+1)G,&\text{if }\text{Tr}^{m}_{e}(a^{2})=0,~\text{Tr}^{m}_{e}(a)\neq0\text{ and }cs\neq 2\text{Tr}^{m}_{e}(a),\\ (p^{2e}-2p^{e}-1)G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\overline{\eta}(\nabla)=1~and~\phi(c)=0,\\ -(2p^{e}+1)G,& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\overline{\eta}(\nabla)=1~\text{and}~\phi(c)\neq0,\\ -G,& \text{if}~\text{Tr}_{e}^{m}(a^{2})\neq0\text{ and }\overline{\eta}(\nabla) = -1\text{ or }\text{Tr}^{m}_{e}(a^{2}) = 0\text{ and }\text{Tr}^{m}_{e}(a) = 0. \end{array}\right.$$ -
3.
If \( 2\nmid s \) and p∣s, then
$$ \aleph_{4}=\left\{\begin{array}{ll} 0,&\text{if}~\text{Tr}^{m}_{e}(a^{2})=0~\text{and}~\text{Tr}^{m}_{e}(a)=0,\\ \overline{\eta}(\frac{c\text{Tr}^{m}_{e}(a)}{2})p^{e}G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})=0~\text{and}~\text{Tr}^{m}_{e}(a)\neq0,\\ \overline{\eta}(-\text{Tr}^{m}_{e}(a^{2}))G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0~\text{and}~\text{Tr}^{m}_{e}(a)=0,\\ \overline{\eta}(-\text{Tr}^{m}_{e}(a^{2}))G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\text{Tr}^{m}_{e}(a)\neq0~\text{and}~\text{Tr}^{m}_{e}(a^{2})=2c\text{Tr}^{m}_{e}(a),\\ \overline{\eta}(2c\text{Tr}^{m}_{e}(a)-\text{Tr}^{m}_{e}(a^{2}))p^{e}G\overline{G}\\~~~~~+\overline{\eta}(-\text{Tr}^{m}_{e}(a^{2}))G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\text{Tr}^{m}_{e}(a)\neq0~\text{and}~\text{Tr}^{m}_{e}(a^{2})\neq2c\text{Tr}^{m}_{e}(a). \end{array}\right.~~~~~~~~~~~$$ -
4.
If \( 2\nmid s \) and \(p\nmid s\), then
$$\aleph_{4} = \left\{\!\!\begin{array}{ll} \overline{\eta}(-s)G\overline{G},&\text{if}~\text{Tr}^{m}_{e}(a^{2})=0~and~\text{Tr}^{m}_{e}(a)=0,\\ \overline{\eta}(-s)G\overline{G},&\text{if}~\text{Tr}^{m}_{e}(a^{2})=0,~\text{Tr}^{m}_{e}(a)\neq0\text{ and }cs=2\text{Tr}^{m}_{e}(a),\\ \left( \overline{\eta}(2c\text{Tr}^{m}_{e}(a)-sc^{2})p^{e}+\overline{\eta}(-s)\right)G\overline{G},&\text{if}~\text{Tr}^{m}_{e}(a^{2})=0, ~\text{Tr}^{m}_{e}(a)\neq0\text{ and }cs\neq2\text{Tr}^{m}_{e}(a),\\ -(p^{e}-2)\overline{\eta}(-s)G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla=0~\text{and}~c\text{Tr}^{m}_{e}(a) = \text{Tr}^{m}_{e}(a^{2}),\\ 2\overline{\eta}(-s)G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla = 0~\text{and}~c\text{Tr}^{m}_{e}(a)\neq \text{Tr}^{m}_{e}(a^{2}),\\ \left( \overline{\eta}(-\text{Tr}^{m}_{e}(a^{2}))+\overline{\eta}(-s)\right)G\overline{G},& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla\neq0~\text{and}~\phi(c)=0,\\ \left( \overline{\eta}(\phi(c))p^{e} + \overline{\eta}(-\text{Tr}^{m}_{e}(a^{2})) + \overline{\eta}(\!-s)\right)G\overline{G},\!\!\!& \text{if}~\text{Tr}^{m}_{e}(a^{2})\neq0,~\nabla\neq0~\text{and}~\phi(c)\neq0. \end{array}\right.$$
Proof
By Lemmas 2.4 and 2.5, we have
There are the following cases to consider: Case 1: s > 2 is even and p∣s; Case 2: s > 2 is even and \(p\nmid s \); Case 3: s ≥ 3 is odd and p∣s; Case 4: s ≥ 3 is odd and \(p\nmid s \).
Case 1: Suppose that s > 2 is even and p∣s. Then, by (5), we obtain
If \( \text {Tr}^{m}_{e}(a^{2})=0 \), then
If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then, by Lemma 2.4, we have
If \(\text {Tr}^{m}_{e}(a^{2})\neq 0\) and \(\text {Tr}^{m}_{e}(a)\neq 0\), then, again from Lemma 2.4, we have
Case 2: Now, assume that s > 2 is even and \( p\nmid s \). By (5), we obtain
If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then, by Lemma 2.4, we have
If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then, by Lemma 2.4, we have
Recall that \( \nabla =(\text {Tr}^{m}_{e}(a))^{2}-s\text {Tr}^{m}_{e}(a^{2})\). If \(\text {Tr}^{m}_{e}(a^{2})\neq 0\), then, by Lemma 2.4, we have
Suppose that \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇ = 0. Then
Next, we consider that \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇≠ 0. Then, by Lemma 2.4, we have
where \( \phi (c)=-sc^{2}+2c\text {Tr}^{m}_{e}(a)-\text {Tr}^{m}_{e}(a^{2}) \). From [14, Exercise 5.24], the equation ϕ(c) = 0 over \(\mathbb {F}_{p^{e}}\) has two distinct solutions if and only if \(\overline {\eta }(\nabla )=1\). Thus, when \(\overline {\eta }(\nabla )=1\), we have
Similarly, when \(\overline {\eta }(\nabla )=-1\), we must have ϕ(c)≠ 0 for all \(c\in \mathbb {F}_{p^{e}}^{*}\) and so
Case 3: Suppose that s ≥ 3 is odd and p∣s. Now, by (5), we obtain
If \(\text {Tr}^{m}_{e}(a^{2})=0\), then
If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then, from Lemma 2.4, we have
If \( \text {Tr}^{m}_{e}(a^{2})\neq 0\) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then, again from Lemma 2.4, we have
Case 4: Suppose that s ≥ 3 is odd and \( p\nmid s \). Now, by (5), we obtain
If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then
If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then
If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \), then we deduce that
If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇ = 0, then \( \overline {\eta }(-\text {Tr}^{m}_{e}(a^{2}))=\overline {\eta }(-s) \) and
Suppose that \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇≠ 0. Recall that \( \phi (c)=-sc^{2}+2c\text {Tr}^{m}_{e}(a) -\text {Tr}^{m}_{e}(a^{2}) \). Then
This completes the proof of the lemma. □
Lemma 3.5
Suppose that \( \lambda \in \mathbb {F}_{p^{e}}^{*} \) and \( \mu \in \mathbb {F}_{p^{e}}\). For i ∈{1,− 1}, let Ki denote the number of pairs (λ, μ) such that \( \overline {\eta }(\mu ^{2}-s\lambda )=i\). Then
Proof
First, we take μ = 0. Then μ2 − sλ = −sλ, and the number of pairs (λ,0) satisfying \( \overline {\eta }(\mu ^{2}-s\lambda )=i \) is \( \frac {(p^{e}-1)}{2} \). Further, we consider that μ≠ 0. Then, for each pair \((\lambda ,\mu )\in \mathbb {F}_{p^{e}}^{*}\times \mathbb {F}_{p^{e}}^{*}\) and fixed \(s\in \mathbb {F}_{p^{e}}^{*}\), we define a mapping \({\mathscr{L}}\) from \(\mathbb {F}_{p^{e}}^{*}\times \mathbb {F}_{p^{e}}^{*}\) into \(\mathbb {F}_{p^{e}}\) by \({\mathscr{L}}(\lambda ,\mu )=\mu ^{2}-s\lambda \). For each \(c_{0}\in \mathbb {F}_{p^{e}}^{*}\), let
Set pe = 2h + 1. Now, for a fixed c0 such that \(\overline {\eta }(c_{0})=1\), the number of pairs (λ, μ2) satisfying μ2 − sλ = c0 is equal to \((0,0)^{(2,p^{e})}+(1,0)^{(2,p^{e})}=h-1~(\text {by Lemma 2.2)}\). Similarly, for a fixed c0 such that \(\overline {\eta }(c_{0})=-1\), the number of pairs (λ, μ2) satisfying μ2 − sλ = c0 is equal to \((0,1)^{(2,p^{e})}+(1,1)^{(2,p^{e})}=h~(\text {from Lemma 2.2)}\). Consequently, we have
We conclude that \(K_{1}=\frac {(p^{e}-1)}{2}+(p^{e}-1)(h-1)\) and \(K_{-1}=\frac {(p^{e}-1)}{2}+(p^{e}-1)h\). Thus, the result is established.□
Lemma 3.6
Suppose that \(\lambda \in \mathbb {F}_{p^{e}}^{*}\), \(\mu \in \mathbb {F}_{p^{e}}\) and μ2 − sλ≠ 0. For i ∈{1,− 1}, let ψi denote the number of the pairs (λ, μ) such that \(\overline {\eta }(\lambda )=i\). Then
and
Proof
Following the arguments similar to the arguments used in the proof of Lemma 3.5, one may easily get the proof of the lemma. So, the proof of the lemma is omitted.□
Let \(c\in \mathbb {F}_{p^{e}}^{*}\) and \(a\in \mathbb {F}_{p^{m}}^{*}\). For a codeword ca of \(\mathcal {C}_{D}\), we denote Nc = Nc(a) to be the number of components \(\text {Tr}^{m}_{e}(ax)\) of ca that are equal to c and n to be the length of ca. So, we have
where
and ℵ2, ℵ3 and ℵ4 are defined in Lemmas 3.2, 3.3 and 3.4 respectively. In the upcoming theorems, we have determined Nc for few different cases.
Theorem 3.7
Assume that \( c,c_{0}\in \mathbb {F}_{p^{e}}^{*} \). If 2∣s and p∣s, then the linear code \( \mathcal {C}_{D} \) defined by (1) has parameters [pm− 2e, s], and its complete weight enumerator is given in Table 1.
Proof
From the definition, this code has length n = #D which follows from Lemma 3.1 and dimension s. From (6), we have \( N_{c}=\frac {n}{p^{e}}+p^{-3e}(\aleph _{2}+\aleph _{3}+\aleph _{4}) \), where \( c\in \mathbb {F}_{p^{e}}^{*} \). In this case, the length of the code \( \mathcal {C}_{D} \) is n = pm− 2e.
If \( a\in \mathbb {F}_{p^{e}}^{*} \), then \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)=0 \). Consequently, we have
Each value occurs only once.
Now, suppose that \(a\in \mathbb {F}_{p^{m}}^{*}\setminus \mathbb {F}_{p^{e}}^{*}\). Then ℵ2 = 0. We give the remaining proof in the following cases: Case 1: If \(\text {Tr}^{m}_{e}(a^{2})=\text {Tr}^{m}_{e}(a)=0\), then
By Lemma 3.1, the frequency is pm− 2e + p−e(pe − 1)G − pe as \( a\notin \mathbb {F}_{p^{e}} \). Case 2: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then
By Lemma 3.1, the frequency is (pe − 1)(pm− 2e − p−eG). Hence, we conclude from last two cases that Nc = pm− 3e occurs pm−e − pe times. Case 3: If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then
It follows from Lemma 3.1 that this value occurs (pe − 1)pm− 2e times. Case 4: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0\) and \(\text {Tr}^{m}_{e}(a)\neq 0 \), then
where \( c_{0}=\frac {\text {Tr}^{m}_{e}(a^{2})}{2\text {Tr}^{m}_{e}(a)}\in \mathbb {F}_{p^{e}}^{*} \). By Lemma 3.1, the frequency is (pe − 1)pm− 2e. This completes the proof of the theorem.□
Corollary 3.8
If 2∣s and p∣s, then the weight distribution of the linear code \( \mathcal {C}_{D} \) defined by (1) is given in Table 2.
Example 3.9
Let (p, m, s, e) = (3,12,6,2). Then, by Theorem 3.7, the code \(\mathcal {C}_{D}\) is a [6561,6,5823] linear code. Its complete weight enumerator and weight enumerator are
and 1 + 419904x5823 + 59040x5832 + 52488x5904 + 8x6561 respectively.
Theorem 3.10
Assume that \( c\in \mathbb {F}_{p^{e}}^{*} \). If 2∣s and \( p\nmid s \), then the linear code \( \mathcal {C}_{D} \) defined by (1) has parameters [n, s], where
Its complete weight enumerator is given in Table 3.
Proof
This code has length n = #D which follows from Lemma 3.1 and dimension s. For \( c\in \mathbb {F}_{p^{e}}^{*} \), recall from (6) that \( N_{c}=\frac {n}{p^{e}}+p^{-3e}(\aleph _{2}+\aleph _{3}+\aleph _{4})\). We now consider the case that 2∣s and \( p\nmid s \). In this case the length n of the code \( \mathcal {C}_{D} \) is pm− 2e + p−eG.
If \( a\in \mathbb {F}_{p^{e}}^{*} \), then \( \text {Tr}^{m}_{e}(a^{2})=a^{2}s\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=as\neq 0 \). Consequently, \( \nabla =(\text {Tr}^{m}_{e}(a))^{2}-s\text {Tr}^{m}_{e}(a^{2})=0 \) and
Moreover, each value occurs only once.
Suppose that \( a\in \mathbb {F}_{p^{m}}^{*}\setminus \mathbb {F}_{p^{e}}^{*} \). Under this assumption, we have ℵ2 = 0. We give the remaining proof in the following cases: Case 1: If \( \text {Tr}^{m}_{e}(a^{2})=\text {Tr}^{m}_{e}(a)=0 \), then
By Lemma 3.1, the frequency is pm− 2e − 1 as a≠ 0. Case 2: If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then
where \( c_{0}=\frac {2\text {Tr}^{m}_{e}(a)}{s}\in \mathbb {F}_{p^{e}}^{*} \). By Lemma 3.1, the frequency is pm− 2e + p−eG. Case 3: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇ = 0, then
where \(c_{o}=\frac {\text {Tr}^{m}_{e}(a^{2})}{\text {Tr}^{m}_{e}(a)}=\frac {\text {Tr}^{m}_{e}(a)}{s}\in \mathbb {F}_{p^{e}}^{*} \) since ∇ = 0. By Lemma 3.1, the frequency is pm− 2e − 1. Case 4: If \(\text {Tr}^{m}_{e}(a^{2})\neq 0\) and \(\overline {\eta }(\nabla )=1\), then
where c0, c1 are two distinct roots of the equation \(\phi (c)=-sc^{2}+2c\text {Tr}^{m}_{e}(a)-\text {Tr}^{m}_{e}(a^{2})=0\), since \(\overline {\eta }(\nabla )=1\). By Lemmas 3.1 and 3.5, the frequency is pm− 2e + p−eG. Case 5: If \(\text {Tr}^{m}_{e}(a^{2})\neq 0\) and \(\overline {\eta }(\nabla )=-1\), then
By Lemmas 3.1 and 3.5, the frequency is \(\frac {1}{2}(p^{e}-1)(p^{m-e}-G)\). Thus, the result is established.□
Corollary 3.11
If 2∣s and \( p\nmid s \), then the weight distribution of the linear code \( \mathcal {C}_{D} \) defined by (1) is given in Table 4.
Example 3.12
Let (p, m, s, e) = (3,8,4,2). Then, by Theorem 3.10, the code \(\mathcal {C}_{D}\) is a [72,4,62] linear code. Its complete weight enumerator and weight enumerator are
and 1 + 2016x62 + 640x63 + 3240x64 + 576x71 + 88x72 respectively.
Remark 3.13
Consider the linear codes [81,6,48] and [71,5,42] obtained in [22] and [13] respectively. Then one can see that our code illustrated in the previous example has improved relative minimum distance.
Theorem 3.14
Assume that \( c\in \mathbb {F}_{p^{e}}^{*} \). If \( 2\nmid s \) and p∣s, then the linear code \( \mathcal {C}_{D} \) defined by (1) has parameters [pm− 2e, s]. Its complete weight enumerator is given in Table 5.
Proof
Now consider that \( 2\nmid s \) and p∣s. In this case the length of the code \( \mathcal {C}_{D} \) is n = pm− 2e.
If \( a\in \mathbb {F}_{p^{e}}^{*} \), then \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)=0 \). Consequently, by (6), we have
Each value occurs only once.
Suppose that \( a\in \mathbb {F}_{p^{m}}^{*}\setminus \mathbb {F}_{p^{e}}^{*} \). Under this assumption, we have ℵ2 = 0. We give the remaining proof in the following cases: Case 1: If \( \text {Tr}^{m}_{e}(a^{2})=\text {Tr}^{m}_{e}(a)=0 \), then
By Lemma 3.1, the frequency is pm− 2e − pe as \( a\notin \mathbb {F}_{p^{e}} \). Case 2: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=0 \), then
By Lemma 3.1, the frequency is (pe − 1)pm− 2e. Hence, we conclude from the last two cases that Nc = pm− 3e occurs pm−e − pe times. Case 3: If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then
from which it follows that \( N_{c}=p^{m-3e}+p^{-2e}\overline {\eta }(c)G\overline {G} \) or \( N_{c}=p^{m-3e}-p^{-2e}\overline {\eta }(c)G\overline {G} \). According to Lemma 3.1, the frequency of each value is \( \frac {1}{2}(p^{e}-1)p^{m-2e} \). Case 4: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \), \(\text {Tr}^{m}_{e}(a)\neq 0 \) and \(\text {Tr}^{m}_{e}(a^{2})=2c\text {Tr}^{m}_{e}(a)\), then
Case 5: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \), \(\text {Tr}^{m}_{e}(a)\neq 0 \) and \(\text {Tr}^{m}_{e}(a^{2})\neq 2c\text {Tr}^{m}_{e}(a)\), then
So, one can easily combine the last two cases as
where \( c_{0}=\frac {\text {Tr}^{m}_{e}(a^{2})}{2\text {Tr}^{m}_{e}(a)}\in \mathbb {F}_{p^{e}}^{*} \). This concludes that
or
By Lemma 3.1, the frequency is \( \frac {1}{2}(p^{e}-1)p^{m-2e} \). Thus, the result is established.□
Corollary 3.15
If \( 2\nmid s \) and p∣s, then the weight distribution of the linear code \( \mathcal {C}_{D} \) defined by (1) is given in the following Table 6.
Proof
The result can be extracted from the complete weight enumerator as shown in the last Theorem 3.14, by observing that
where \(c_{0}\in \mathbb {F}_{p^{e}}^{*}\). □
Example 3.16
Let (p, m, s, e) = (3,6,3,2). Then, by Theorem 3.14, the code \(\mathcal {C}_{D}\) is a [9,3,7] linear code, which is a MDS code. Let \(\mathbb {F}_{9}=\{w_{0},w_{1}\ldots ,w_{8}\}\). If w2, w4, w6 and w8 are square elements in \(\mathbb {F}_{9}^{*}\), then its complete weight enumerator is
while its weight enumerator is 1 + 288x7 + 144x8 + 296x9.
Theorem 3.17
Assume that \( c\in \mathbb {F}_{p^{e}}^{*} \). If \( 2\nmid s \) and \( p\nmid s \), then the linear code \( \mathcal {C}_{D} \) defined by (1) has parameters [pm− 2e − K, s], where
Its complete weight enumerator is given in Table 7, where ∇ runs through \(\mathbb {F}_{p^{e}}^{*}\) such that \(\overline {\eta }(-\nabla )=-1\).
Proof
Consider the case that \( 2\nmid s \) and \( p\nmid s \). In this case, the length of the code \( \mathcal {C}_{D} \) is \( n=p^{m-2e}-\overline {\eta }(-s)p^{-2e}G\overline {G} \).
If \( a\in \mathbb {F}_{p^{e}}^{*} \), then \( \text {Tr}^{m}_{e}(a^{2})=a^{2}s\neq 0 \) and \( \text {Tr}^{m}_{e}(a)=as\neq 0 \). Consequently, \( \nabla =(\text {Tr}^{m}_{e}(a))^{2}-s\text {Tr}^{m}_{e}(a^{2}) =0\) and \( \overline {\eta }(-\text {Tr}^{m}_{e}(a^{2}))=\overline {\eta }(-s)\). Thus
Each value occurs only once.
Suppose that \( a\in \mathbb {F}_{p^{m}}^{*}\setminus \mathbb {F}_{p^{e}}^{*} \). Under this assumption ℵ2 = 0. The remaining proof is given in the following cases: Case 1: If \( \text {Tr}^{m}_{e}(a^{2})=\text {Tr}^{m}_{e}(a)=0 \), then
By Lemma 3.1, the frequency is \( p^{m-2e}+p^{-2e}\overline {\eta }(-s)(p^{e}-1)G\overline {G}-1 \). Case 2: If \( \text {Tr}^{m}_{e}(a^{2})=0 \) and \( \text {Tr}^{m}_{e}(a)\neq 0 \), then
where \( c_{0}=\frac {2\text {Tr}^{m}_{e}(a)}{s}\in \mathbb {F}_{p^{e}}^{*} \). By Lemma 3.1, the frequency is n. Case 3: If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and ∇ = 0, then \( \overline {\eta }(-\text {Tr}^{m}_{e}(a^{2}))=\overline {\eta }(-s) \). Consequently, we have
where \(c_{o}=\frac {\text {Tr}^{m}_{e}(a)}{s}\in \mathbb {F}_{p^{e}}^{*} \) since ∇ = 0. One may easily find, by Lemma 3.1, that the frequency is \( p^{m-2e}+\overline {\eta }(-s)(p^{e}-1)p^{-2e}G\overline {G}-1\). Case 4: If \(\text {Tr}^{m}_{e}(a^{2})\neq 0\) and ∇≠ 0, then
where \( \phi (c)=-sc^{2}+2c\text {Tr}^{m}_{e}(a)-\text {Tr}^{m}_{e}(a^{2})\). This case is divided into the following two subcases. Case 4(a): If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \( \overline {\eta }(\nabla )=1 \), then the equation ϕ(c) = 0 must have two distinct roots. Let c0 and c1 be the distinct roots of ϕ(c). Then ϕ(c) can be represented as ϕ(c) = −s(c − c0)(c − c1). Therefore, we have
By Lemma 3.1, the frequency is n. Moreover, by Lemma 3.5, there are \( \frac {1}{2}(p^{e}-1)(p^{e}-2) \) such values. Case 4(b): If \( \text {Tr}^{m}_{e}(a^{2})\neq 0 \) and \(\overline {\eta }(\nabla )=-1\), then
More precisely, by writing \( \phi (c)=-s\left (c-\frac {\text {Tr}^{m}_{e}(a)}{s}\right )^{2}+\frac {\nabla }{s} \), we deduce that
where \( c_{o}=\frac {\text {Tr}^{m}_{e}(a)}{s} \) and \( \text {Tr}^{m}_{e}(a)\neq 0\). The number of such values is \( \frac {1}{2}(p^{e}-1)^{2} \). On the other hand if \( \text {Tr}^{m}_{e}(a)=0 \), then
The number of such values is \( \frac {1}{2}(p^{e}-1) \). Again, from Lemma 3.1, each value occurs n times. Thus, we have the desired result. □
Corollary 3.18
If \( 2\nmid s \) and \( p\nmid s \), then the weight distribution of the linear code \( \mathcal {C}_{D} \) defined by (1) is given in Table 8, where
Proof
For \(a\in \mathbb {F}_{p^{m}}^{*}\), define \(N_{0}=\#\{x\in \mathbb {F}_{p^{m}}:\text {Tr}^{m}_{e}(x)=1,\text {Tr}^{m}_{e}(x^{2})=0 \text { and }\text {Tr}^{m}_{e}(ax)=0\}\). For given n the length of \(\mathcal {C}_{D}\), the Hamming weight of a codeword ca is given by
In the same manner, as in (6), we can find that
where
Let \(2\nmid s\) and \(p\nmid s\). Then, by following the arguments similar to the arguments used in the proofs of Lemmas 3.2, 3.3 and 3.4, one may easily get that
The result is directly follows from (7), (8) and Lemmas 3.1 and 3.6.□
Remark 3.19
If \(b\in \mathbb {F}_{p^{e}}^{*}\) is fixed and \(D_{b}=\{x\in \mathbb {F}_{p^{m}}: \text {Tr}^{m}_{e}(x)=b \text { and } \text {Tr}^{m}_{e}(x^{2})=0\},\) then we get the code \(\mathcal {C}_{D_{b}}\) of the form (1). Now, define a mapping fb : D1 → Db by
This implies that the code \(\mathcal {C}_{D_{b}}\) is equal to \(\mathcal {C}_{D_{1}}\). So, Theorems 3.7, 3.10, 3.14 and 3.17 actually demonstrate the complete weight enumerators of \(\mathcal {C}_{D_{b}}\) for all \(b\in \mathbb {F}_{p^{e}}^{*}\).
Corollary 3.20
Let \(\mathcal {C}_{D}\) be the linear code defined by (1). Then \(\mathcal {C}_{D}\) is optimal with respect to the Griesmer bound only if s = 3. If s = p = 3, then it is MDS and it has parameters [3e,3,3e − 2]. Moreover, for s = 3 and p > 3, it has parameters [pe + 1,3,pe − 1] if \(\overline {\eta }(-3)=-1\) and [pe − 1,3,pe − 3] if \(\overline {\eta }(-3)=1\).
Proof
If \(2\nmid s\) and p∣s, then it directly follows from Corollary 3.15 that \(\mathcal {C}_{D}\) has parameters \([p^{m-2e}, s, (p^{e}-1)p^{m-3e}-p^{\frac {m-3e}{2}}]\). Suppose that \(s=2s^{\prime }+1\), where \(s=\frac {m}{e}\) and \(s^{\prime }\geq 1\). Then we have
By the equation
we have \(s^{\prime }=1\) whence s = 3. Since p∣s, we must have p = 3. Thus, for p = s = 3, the code \(\mathcal {C}_{D}\) is MDS with parameters [3e,3,3e − 2].
Now, we suppose that \(2\nmid s\), \(p\nmid s\) and \(\overline {\eta }(-s)^{m+e-2}(-1)^{\frac {(p-1)^{2}(m+e)}{8}}=-1\). From Corollary 3.18, the code \(\mathcal {C}_{D}\) has parameters \([p^{m-2e}+p^{\frac {m-3e}{2}}, s, (p^{e}-1)p^{m-3e}]\). Thus
Hence, the equation \(p^{e(s-2)}+1=p^{e(s-2)}+p^{\frac {e(s-3)}{2}}\) gives that s = 3. Consequently, we have \(\overline {\eta }(-3)=-1\). Therefore, the code \(\mathcal {C}_{D}\) is MDS with parameters [pe + 1,3,pe − 1].
Further, consider that \(2\nmid s\), \(p\nmid s\) and \(\overline {\eta }(-s)^{m+e-2}(-1)^{\frac {(p-1)^{2}(m+e)}{8}}=1\). In the same manner, we can show that \(\mathcal {C}_{D}\) is MDS with parameters [pe − 1,3,pe − 3] when s = 3 and \(\overline {\eta }(-3)=1\). Other cases may similarly be verified.
Thus, it follows from Lemma 2.6 that the code \(\mathcal {C}_{D}\) is optimal achieving the Griesmer bound provided that s = 3. □
It is well known that error-correcting capability of a linear code \(\mathcal {C}=[n, k, d]\) depends on the relative minimun distance \(\delta (\mathcal {C})\)(see [15]). A linear code \(\mathcal {C}\) is MDS if and only if \(\delta (\mathcal {C})+\mathcal {R}(\mathcal {C})=1\), where \(\mathcal {R}(\mathcal {C})\) is the transmission rate of \(\mathcal {C}\). The Remark 3.21 and Tables 9 and 10 conclude that codes determined in this paper have improved error-correcting capability and more close to MDS codes than the codes in [22].
In the Table 10, ∗ shows that the complete weight enumerator of the code is not inlcluded in the present paper due to large presentation.
Remark 3.21
The MDS code [3,3,1], which is unique, obtained in [22] has zero relative minimum distance while our MDS codes [3e,3,3e − 2] have nonzero relative minimum distance for every e ≥ 2. In addition, it can easily be checked that
Hence, one can conclude that our MDS codes have better error-correcting capability than the MDS codes in Corollary 2 of [22].
3.2 The dual code of the code \(\mathcal {C}_{D}\)
In this subsection, we study the dual code of the code \(\mathcal {C}_{D}\). In the following theorem, we have determined bounds on the minimum distance of the dual code.
Theorem 3.22
Let the symbols have the same meanings as before, and let d⊥ denote the minimum distance of the dual code \(\mathcal {C}_{D}^{\perp }\) of the code \(\mathcal {C}_{D}\) defined in (1). Then
-
1.
if 2∣s and p∣s, we have 3 ≤ d⊥≤ 4.
-
2.
if 2∣s, \(p\nmid s\) and e ≥ 2, we have 3 ≤ d⊥≤ 4.
-
3.
if \(2\nmid s\) and e ≥ 2, we have 3 ≤ d⊥≤ 4. In particular, d⊥ = 4 if m = 3e.
Proof
We only give the proof of first part since the proofs of other parts are similar to the proof of first part. It can easily be checked that \(\mathcal {C}_{D}^{\perp }\) has no codeword of weight 1. Next, suppose to the contrary that there exists a codeword \(c\in \mathcal {C}_{D}^{\perp }\) such that wt(c)= 2. Then, for all \(a\in \mathbb {F}_{p^{m}}\), we have
Consequently, we have that ci(di − dj) = 0. This is contradictory to the facts that ci≠ 0 and di≠dj. Hence, we do not have any vector \(c\in \mathcal {C}_{D}^{\perp }\) of weight 2. The upper bound is directly follows from the sphere-packing (or Hamming ) bound. Thus, the result is established. □
Example 3.23
Let (p, m, s, e) = (5,6,3,2). Then the code \(\mathcal {C}_{D}^{\perp }\) has parameters [24,21,4], and it is optimal with respect to the Griesmer bound. In addition, it is MDS code.
Example 3.24
Let (p, m, s, e) = (3,9,3,3). Then the code \(\mathcal {C}_{D}^{\perp }\) has parameters [27,24,4], and it is optimal with respect to the Griesmer bound. In fact, it is MDS code.
4 Application to constant composition codes
In this section, we construct some optimal constant composition codes employing the complete weight enumerators of the linear code \(\mathcal {C}_{D}\). The code \(\mathcal {C}_{D}\) can be found in Theorems 3.7, 3.10, 3.14 and 3.17.
It is well known that one can easily construct constant composition codes from the complete weight enumerators of the linear codes. Let r be a positive interger, and let A = {a0, a1,…,ar− 1} be a code alphabet. Any subset C ⊂ Sn of size M and minimum distance d such that each codeword has the same composition (t0, t1,…,tr− 1) is known as (n, M, d,(t0, t1,…,tr− 1)) constant composition code over S. There are many applications of constant composition codes in communications engineering [3,19]. The following LFVC bound of constant composition codes is given in [17].
Lemma 4.1
Let (n, M, d,(t0, t1,…,tr− 1)) be a constant composition code over S with \(nd-n^{2}+({t_{0}^{2}}+{t_{1}^{2}}+\cdots +t_{r-1}^{2})>0\). Then
A constant composition code which meets the LFVC bound is known as optimal constant composition code. We construct some optimal constant composition codes from the complete weight enumerators of \(\mathcal {C}_{D}\) in the following theorems:
Theorem 4.2
Consider the CWE of the Theorem 3.10. If m = 4e, then there exists an \((n, M, d,(t_{0},t_{1},\ldots ,t_{p^{e}-1}))\) optimal constant composition code \(\tilde {\mathcal {C}}\) over \(\mathbb {F}_{p^{e}}\), where \(n=p^{m-2e}-p^{\frac {m-2e}{2}},~M=p^{\frac {m-2e}{2}}-1,~d=p^{m-2e}-p^{m-3e},~t_{0}=p^{m-3e}-p^{\frac {m-2e}{2}},~t_{i}=p^{m-3e} \text { for }i\neq 0\).
Proof
Since, we have pm− 2e − 1 codewords of desired parameters n, t0 and ti(1 ≤ i ≤ pe − 1). So, any codeword can be re-arranged as
Fix all zero symbols, and consider all same symbols as a single symbol. Now, we take pe − 1 cycles of the nonzero symbols of the above re-arranged codeword. Define \(\tilde {\mathcal {C}}\) as the collection of all constructed cycles. It is obvious that the minimum distance of the code \(\tilde {\mathcal {C}}\) is pm− 2e − pm− 3e.
One can easily check that \(nd-n^{2}+({t_{0}^{2}}+{t_{1}^{2}}+\cdots +t_{p^{e}-1}^{2})=P^{\frac {3m-6e}{2}}-P^{\frac {3m-8e}{2}}>0 \) and \(nd=(p^{m-2e}-p^{\frac {m-2e}{2}})(p^{m-2e}-p^{m-3e})\). Then we have
Hence, by Lemma 4.1, \(\tilde {\mathcal {C}}\) is an optimal constant composition code. Thus, the result is established. □
Theorem 4.3
Consider the CWE of the Theorem 3.17. If m = 3e and \(\overline {\eta }(-3)=1\), then there exists an \((n, M, d,(t_{0},t_{1},\ldots ,t_{p^{e}-1}))\) optimal constant composition code \(\tilde {\mathcal {C}}\) over \(\mathbb {F}_{p^{e}}\), where \(n=p^{m-2e}-p^{\frac {m-3e}{2}},~M=p^{\frac {m-e}{2}}-1,~d=p^{m-2e}-p^{m-3e},~t_{0}=p^{m-3e}-p^{\frac {m-3e}{2}},~t_{i}=p^{m-3e} \text { for } i\neq 0\).
Proof
Following the arguments used in the proof of previous theorem, we can construct \(p^{\frac {m-e}{2}}-1\) codewords of desired parameters \((n, M, d,(t_{0},t_{1},\ldots ,t_{p^{e}-1}))\).
One can easily check that \(nd-n^{2}+({t_{0}^{2}}+{t_{1}^{2}}+\cdots +t_{p^{e}-1}^{2})=P^{\frac {3m-7e}{2}}-P^{\frac {3m-9e}{2}}>0 \) and \(nd=(p^{m-2e}-p^{\frac {m-3e}{2}})(p^{m-2e}-p^{m-3e})\). Then, we have
By Lemma 4.1, \(\tilde {\mathcal {C}}\) is an optimal constant composition code. This completes the proof. □
5 Concluding remarks
In this paper, a class of linear codes over arbitrary finite fields with their complete weight enumerators by giving two restrictions in the defining set has been presented. In addition, the weight distributions of the codes are also determined. The codes presented in this paper have improved relative minimum distance as we have shown in Table 10 which improves error-correcting capability. In Corollary 3.20, we have found a MDS code [3e,3,3e − 2] for any positive integer e. Furthermore, in Sections 3.2 and 4, we have found some optimal dual codes and constant composition codes respectively. Lastly the work presented in the paper may further be extended to find more applicable codes.
References
Ahn, J., Ka, D., Li, C.: Complete weight enumerators of a class of linear codes. Des. Codes Cryptogr. 83(1), 83–99 (2017)
Calderbank, A.R., Goethals, J.M.: Three-weight codes and association schemes. Philips J. Res. 39(4-5), 143–152 (1984)
Chu, W., Colbourn, C.J., Dukes, P.: On constant composition codes. Discrete Appl. Math. 154(6), 912–929 (2006)
Ding, C.: Optimal constant composition codes from zero-difference balanced functions. IEEE Trans. Inf. Theory 54(12), 5766–5770 (2008)
Ding, K., Ding, C.: Binary linear codes with three-weights. IEEE Commun. Lett. 18(11), 1879–1882 (2014)
Ding, K., Ding, C.: A class of two-weight and three-weight codes and their applications in secret sharing. IEEE Trans. Inform. Theory 61(11), 5835–5842 (2015)
Ding, C., Helleseth, T., Kløve, T., Wang, X.: A generic construction of Cartesian authentication codes. IEEE Trans. Inf. Theory 53(6), 2229–2235 (2007)
Ding, C., Wang, X.: A coding theory construction of new systematic authentication codes. Theoret. Comput. Sci. 330(1), 81–99 (2005)
Ding, C., Yin, J.: A construction of optimal constant composition codes. Des. Codes Cryptogr. 40(2), 157–165 (2006)
Griesmer, J.H.: A bound for error-correcting codes. IBM J. Res. Develop. 4(5), 532–542 (1960)
Heng, Z., Yue, Q.: Two classes of two-weight linear codes. Finite Fields Appl. 38, 72–92 (2016)
Li, C., Bae, S., Ahn, J., Yang, S., Yao, Z.: Complete weight enumerators of some linear codes and their applications. Des. Codes Cryptogr. 81(1), 153–168 (2016)
Li, F., Wang, Q., Lin, D.: A class of three-weight and five-weight linear codes. Discrete Appl. Math. 241, 25–38 (2018)
Lidl, R., Niederreiter, H.: Finite fields. Cambridge University Press, New York (1997)
Ling, S., Xing, C.P.: Coding theory. A first course. Cambridge University Press, Cambridge (2004)
Liu, Y., Liu, Z.: Complete weight enumerators of a new class of linear codes. Discrete Math. 341(7), 1959–1972 (2018)
Luo, Y., Fu, F., Vinck, A.J.H., Chen, W.: On constant-composition codes over \(\mathbb {Z}_{q}\). IEEE Trans. Inf. Theory 49(11), 3010–3016 (2003)
MacWilliams, F.J., Slone, N.J.A.: The theory of error-correcting codes. North-Holland Publishing, Amsterdam (1977)
Milenkovic, O., Kashyap, N.: On the design of codes for DNA computing lecture notes in computer science, vol. 3969. Springer, Berlin (2006)
Storer, T.: Cyclotomy and difference sets. Markham Publishing Company, Markham, Chicago (1967)
Wang, Q., Li, F., Lin, D.: Complete weight enumerators of two classes of linear codes. Discrete Math. 340(4), 467–480 (2017)
Yang, S., Kong, X., Tang, C.: A construction of linear codes and their complete weight enumerators. Finite Fields Appl. 48, 196–226 (2017)
Yang, S., Yao, Z.: Complete weight enumerators of a family of three-weight linear codes. Des. Codes Cryptogr. 82(3), 663–674 (2017)
Yang, S., Yao, Z., Zhao, C.: A class of three-weight linear codes and their complete weight enumerators. Cryptogr. Common. 9(1), 133–149 (2017)
Acknowledgements
The authors would like to thank the reviewers and the Editor for their valuable comments that considerably helped to improve the presentation and quality of the paper. The present research is supported by University Grants Commission, New Delhi, India, under JRF in Science, Humanities & Social Sciences scheme, with Grant number 11-04-2016-413564.
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
Kumar, P., Khan, N.M. A class of linear codes with their complete weight enumerators over finite fields. Cryptogr. Commun. 13, 695–725 (2021). https://doi.org/10.1007/s12095-021-00496-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12095-021-00496-w