Homogeneous metric and matrix product codes over finite commutative principal ideal rings

https://doi.org/10.1016/j.ffa.2020.101666Get rights and content

Abstract

In this paper, a necessary and sufficient condition for the homogeneous distance on an arbitrary finite commutative principal ideal ring to be a metric is obtained. We completely characterize the lower bound of homogeneous distances of matrix product codes over any finite principal ideal ring where the homogeneous distance is a metric. Furthermore, the minimum homogeneous distances of the duals of such codes are also explicitly investigated.

Introduction

Constantinescu and Heise [4] first introduced the notion of homogeneous weight on the residue ring of integers. This weight is a generalization of the Hamming weight on finite fields and the Lee weight on the residue ring of integers modulo 4. This generalized weight is the only function that has the properties of homogeneity that are required to construct codes in coding theory. Later, Greferath and Schmidt [8] extended homogeneous weights to arbitrary finite rings, proved the existence and uniqueness of the homogeneous weight on finite rings, and exhibited a formula to calculate the homogeneous weight. However, it is not easy to determine the exact value of the homogeneous weight on finite rings. Fan and Liu [7] obtained an explicit formula of the homogeneous weight on finite principal ideal rings. In general, the homogeneous distance on a finite principal ideal ring induced by homogeneous weight may not be a metric. It is proved that [4], [12] the homogeneous distance on ZN is a metric if and only if N is not divisible by 6. There is no further result about when the homogeneous distance on an arbitrary finite commutative principal ideal ring is a metric.

Matrix product codes over finite fields were introduced in [1]. There are many such constructions, for example, the (u|u+v)-construction and the (a+x|b+x|a+b+x)-construction which can be viewed as matrix product codes. The reference [13] showed that some quasi-cyclic codes can also be rewritten as matrix product codes. The reference [1] also exhibited a lower bound for the minimum Hamming distances of matrix product codes over finite fields constructed by non-singular by columns matrices.

Codes over finite rings have received much attention recently after it was proved that some important families of binary non-linear codes are images under a Gray map of linear codes over Z4 (see, for example, [2], [9], [15]). The references [17], [18] showed that only finite Frobenius rings are suitable for coding alphabets. In reference [6], the minimum Hamming distance of matrix product codes over finite Frobenius rings constructed with several types of matrices is bounded in different ways and the duals of matrix product codes are also explicitly described in terms of matrix product codes. The matrix product codes over some special finite Frobenius rings like finite chain rings and finite principal ideal rings have been investigated from many perspectives. In reference [16], matrix product codes over finite chain rings were studied and the lower bound on the minimum distance of matrix product codes by non-singular by columns matrices in [1] was extended to the minimum homogeneous distance. Let R be a finite commutative principal ideal ring with identity. Then R=R1×R2××Rs, where Rt is a finite chain ring and Jt is the unique maximal ideal of Rt. Suppose qt=|Rt/Jt| for 1ts, and q1q2qs. Reference [5] gave a lower bound for the minimum homogeneous distances of matrix product codes constructed by non-singular by columns matrices over the finite principal ideal ring R under the condition q2>q1+1 provided s2. Reference [5] also provided a lower bound for the minimum homogeneous distances of the duals of such codes constructed by square non-singular by columns matrices. Some other related results can be found in [3], [10], [11], [14].

In this paper, we first give a necessary and sufficient condition for the homogeneous distance on an arbitrary finite commutative principal ideal ring to be a metric. Then we completely characterize the lower bound of homogeneous distances of matrix product codes over any finite principal ideal ring where the homogeneous distance is a metric. As a consequence, the main results in reference [5] on the homogeneous distance lower bound when R satisfies q2>q1+1 provided s2 are special cases of our results, from which they can be obtained directly. The minimum homogeneous distances of the duals of such codes constructed by arbitrary non-singular by columns matrices are also explicitly investigated. It generalizes the previous results in reference [5] where only square non-singular by columns matrices are considered.

This paper is organized as follows. Some necessary background materials are given in Section 2. We provide a necessary and sufficient condition for the homogeneous distance on a finite commutative principal ideal ring to be a metric in Section 3. Then we obtain a lower bound for the minimum homogeneous distances of matrix product codes and their duals constructed with non-singular by columns matrices over all such rings where the homogeneous distance is a metric in Section 4 and Section 5.

Section snippets

Finite commutative principal ideal rings and homogeneous weights

In this paper, R is always a finite commutative principal ideal ring with identity. Then R is isomorphic to a product of finite chain rings, which means that there exists a ring isomorphismRR1×R2××Rs,r(r(1),r(2),,r(s)), where Rt is a finite commutative chain ring with identity and r(t)Rt for 1ts. With this isomorphism, we identify R with R1×R2××Rs and just write r=(r(1),r(2),,r(s)) in the following. For t=1,2,,s, let Jt denote the unique maximal ideal of Rt generated by γt and et

Homogeneous distances

Recall that RR1×R2××Rs, where Rt is a finite commutative chain ring with identity. For t=1,2,,s, Jt is the unique maximal ideal of Rt, Ft:=Rt/Jt is a finite field and qt=|Ft|. Furthermore, we also assume that q1q2qs.

By Remark 2.1, the homogeneous distance dh on R is not always a metric on R. In this section, we provide a necessary and sufficient condition for the homogeneous distance dh on R to be a metric on R. It is easy to see that the homogeneous distance dh on R is a metric on R if

Homogeneous distance of matrix product codes

A lower bound for the minimum homogeneous distances of matrix product codes over finite principal ideal rings with q2>q1+1 was obtained in the previous research [5]. In this section, we will completely characterize the lower bound of homogeneous distances of matrix product codes over any finite principal ideal ring.

A finite commutative principal ideal ring R=R1×R2××Rs can be divided into the following three types:

  • (T1)

    q1=q2==qs, where s1.

  • (T2)

    qt+1qt+2 provided q1==qt<qt+1qt+2qs, where 1ts1.

  • (T3)

    qt+

Homogeneous distances of the duals

Theorem 4.1 gives a lower bound for the minimum homogeneous distances of the matrix product codes constructed with non-singular by columns matrices over R which satisfies (T1) or (T2). In this section, we will provide a lower bound for the minimum homogeneous distances of the duals of such codes. In particular, the main result in reference [5] on the lower bound of the dual codes of matrix product codes is a special case of our results. Note that the dual code of a matrix product code

Acknowledgements

This work was supported by NSFC (Grant No. 11871025).

References (18)

There are more references available in the full text version of this article.

Cited by (3)

The main results in this paper have been reported at the 8th ICCM (June 2019) hosted by Yau Mathematical Sciences Center at Tsinghua University, Beijing, China.

View full text