1.
Introduction
Constructing quantum codes is an important subject in the quantum information. The CSS construction and the Hermitian construction were introduced to construct quantum codes from classical error-correcting codes in [1,2,3,4]. Constructing quantum codes needs Euclidean dual-containing or Hermitian dual-containing codes with respect to the Euclidean or Hermitian inner product using the CSS construction or Hermitian construction. Fan and Zhang [5] introduced the Galois inner products to generalize the Euclidean inner product and the Hermitian inner product. In [6], Hermitian LCD λ-constacyclic codes over Fq were addressed by studying the Galois inner product. Galois hulls of linear codes over finite fields were addressed by studying the Galois dual in [7]. In [8], σ-self-orthogonal constacyclic codes over Fpm+uFpm were studied by generalizing the notion of self-orthogonal codes to σ-self-orthogonal codes over an arbitrary finite ring. Fu and Liu [9] extended constacyclic codes to obtain Galois self-dual codes.
Now, many quantum codes have been constructed by studying the algebraic structure of cyclic and constacyclic codes over finite fields, finite chain rings and finite non-chain ring. Huang et al. [10] constructed quantum codes from Hermitian dual-containing codes by applying the Hermitian construction. In [11], some quantum codes were obtained from constacyclic over Fq[u,v]/⟨u2−γu,v2−δv,uv=vu=0⟩ by using the CSS construction. Gowdhaman et al. [12] studied the the structure of cyclic and λ-constacyclic codes over Fp[u,v]⟨v3−v,u3−u,uv−vu⟩ and constructed quantum codes over Fp using the CSS construction. Islam and Prakash [13] obtained quantum codes from cyclic codes over a finite non-chain ring Fq[u,v]/⟨u2−αu,v2−1,uv−vu⟩ using the CSS construction. Kong and Zheng obtained some quantum codes from constacyclic codes over Fq[u1,u2,⋯,uk]/⟨u3i=ui,uiuj=ujui⟩ [14] and Fq[u1,u2,⋯,uk]/⟨u3i=ui,uiuj=ujui=0⟩ [15] using the CSS construction. As an application, Galois inner product can be applied in the constructions of quantum codes. Some entanglement-assisted quantum codes were obtained from Galois dual codes in [16,17,18].
Motivated by these works, we define a new non-chain ring Rl,k by generalizing [15] and study the algebraic structure of σ-self-orthogonal and σ-dual-containing constacyclic codes over Rl,k based on the σ-inner product. Then, we give the necessary and sufficient conditions for λ-constacyclic codes over Rl,k to satisfy σ-self-orthogonal and σ-dual-containing. Finally, we obtain some new quantum codes from σ-dual-containing constacyclic codes over Rl,k using the CSS construction or Hermitian construction and compare these codes better with the existing codes that appeared in some recent papers.
2.
Preliminaries
Let R be a finite commutative ring, I is an ideal of R and I is generated by one element, then I is called a principal ideal. If all the ideas of R are principal, R is called a principal ideal ring. If R has a unique maximal ideal, R is called a local ring. If the ideals of R are linearly ordered by inclusion, R is called a chain ring.
Let Rl,k=Fpm[u1,u2,⋯,uk]/⟨uli=ui,uiuj=ujui=0⟩, where p is a prime, l is a positive integer, (l−1)∣(p−1) and 1≤i,j≤k. It is a commutative non-chain ring with pm(l−1)k+m elements. Since (l−1)∣(p−1), then
where αi∈Fpm for i=1,2,⋯,l.
Let
where 2≤i≤l and 1≤j≤k.
We can get that ∑ijςij=1, ς2ij=ςij and ςijςi′j′=0, where (i,j)≠(i′,j′).
Let e1=ς21,e2=ς22,⋯,ek=ς2k,⋯,e(l−1)k=ςlk,e(l−1)k+1=ς11. Let s=(l−1)k+1, thus, 1=e1+e2+⋯+es, e2i=ei and eiej=0, where i≠j and i,j=1,2,⋯,s. By the Chinese Remainder Theorem, then
For any element r∈Rl,k, it can be expressed uniquely as
where ri∈Fpm for i=1,2,⋯,s.
Let Aut(Rl,k) be the ring automorphism group of Rl,k. If σ∈Aut(Rl,k), then we can get a bijective map
Let a=(a0,a1,⋯,an−1) and b=(b0,b1,⋯,bn−1)∈Rnl,k, the σ-inner product [8] of a,b is defined as
When Rl,k=Fpm and σ is the identity map of Fpm, then σ-inner product is the Euclidean inner product. When Rl,k=Fpm and m is even, ∀a∈Fpm, if σ(a)=apm2, then σ-inner product is the Hermitian inner product. When Rl,k=Fpm and ∀a∈Fpm, σ(a)=apl, 0≤l≤m−1, then σ-inner product is the Galois inner product [5].
When ⟨a,b⟩σ=0, a and b are called σ-orthogonal, for any code C over Rl,k, the σ-dual code of C is defined as
If C⊆C⊥σ, C is σ-self-orthogonal, if C⊥σ⊆C, C is σ-dual-containing, and if C=C⊥σ, C is σ-self-dual.
Let θt be an automorphism of Fpm, θt: Fpm→Fpm defined by θt(a)=apt, where 0≤t≤m−1. We can define the automorphism of Rl,k as follows:
Let c=(c0,c1,⋯,cn−1)∈Rnl,k and λ be a unit of Rl,k, then the constacyclic shift δλ of c is defined as δλ(c)=(λcn−1,c0,⋯,cn−2). If δλ(C)=C, C is called a λ-constacyclic code of length n over Rl,k. In particular, when λ=1, C is called cyclic code and negative cyclic code when λ=−1.
Let f(x)=a0+a1x+a2x2+⋯+ar−1xr−1+xr be a monic polynomial, the reciprocal polynomial of f(x) is denoted by f∗(x)=xrf(x−1).
Each codeword (a0,a1,⋯,an−1)∈Rnl,k can be represented by a polynomial a0+a1x+⋯+an−1xn−1∈Rl,k[x]/⟨xn−λ⟩,
In polynomial representation, a λ-constacyclic code of length n over Rl,k is defined as an ideal of Rl,k[x]/⟨xn−λ⟩.
We define a Gray map ϕl,k:Rl,k→Fspm by a=∑si=1aiei↦(a1,a2,⋯,as), and we extend ϕl,k as
where ai=a1,ie1+a2,ie2+⋯+as,ies∈Rl,k, i=0,1,⋯,n−1.
∀r∈Rl,k, the Gray weight of r is defined as wG(r)=wH(ϕl,k(r)), where wH(ϕl,k(r)) is the Hamming weight of the image of r under ϕl,k.
∀x=(x1,x2,⋯,xn),y=(y1,y2,⋯,yn)∈Rnl,k, the Gray weight of x−y is defined as wG(x−y)=∑ni=1wG(xi−yi), the Gray distance of x,y is defined as dG(x,y)=wG(x−y), and the Gray distance of C is defined as dG(C)=min{dG(a,b),a,b∈C,a≠b}.
For a linear code C of length n over Rl,k. Let
where j=1,2,⋯,s.
Then, C1,C2,⋯,Cs are linear codes of length n over Fpm, C=⨁sj=1ejCj and ∣C∣=∏sj=1∣Cj∣.
Lemma 2.1. An element (λ1e1+λ2e2+⋯+λses)∈Rl,k is a unit in Rl,k if and only if λi is a unit in Fpm for i=1,2,⋯,s.
Proof. This proof is the same as Lemma 3.1 in [14]. □
3.
λ-constacyclic codes over Rl,k
In this section, let λ=∑sj=1λjej be a unit in Rl,k, then λ−1=∑sj=1λ−1jej, where s=(l−1)k+1, λj∈F∗pm for j=1,2,⋯,s.
Lemma 3.1. (See [8]) Let R be a finite commutative Frobenius ring with identity, σ,˜σ∈Aut(R) and C be a liner code of length n over R, then
(1) C⊥σ is a linear code over R.
(2) C⊥σ=σ−1(C⊥) and ∣C∣∣C⊥σ∣=∣R∣n.
(3) (C⊥σ)⊥˜σ=˜σ−1σ−1(C).
Theorem 3.1. C=⨁sj=1ejCj is a λ-constacyclic code of length n over Rl,k if and only if Cj is a λj-constacyclic code of length n over Fpm for j=1,2,⋯,s.
Proof. This proof is the same as Theorem 1 in [15]. □
Theorem 3.2. Let C=⨁sj=1ejCj be a linear code of length n over Rl,k. Then C⊥σ=∑sj=1ejC⊥σj, where C⊥σj is the σ dual code of Cj for j=1,2,⋯,s.
Proof. Let ˜C=∑sj=1ejC⊥σj, ∀x=∑sj=1ejxj∈C and ∀˜x=∑sj=1ej~xj∈˜C, then
where xj∈Cj, ~xj∈C⊥σj.
So
For Rl,k is a Frobenius ring, by Lemma 3.1,
Hence
So
□
Theorem 3.3. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k. Then C⊥σ=∑sj=1ejC⊥σj is a σ(λ−1)-constacyclic code of length n over Rl,k, and C⊥σj is a σ(λ−1j)-constacyclic code over Fpm for j=1,2,⋯,s.
Proof. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k. ∀x=(x0,x1,⋯,xn−1)∈C⊥σ and ∀y=(y0,y1,⋯,yn−1)∈C, then
and
We can get that
So
We have δσ(λ−1)(x)∈C⊥σ, so C⊥σ is a σ(λ−1)-constacyclic code.
By Lemma 2.1, λ−1=λ−11e1+λ−12e2+⋯+λ−1ses, it implies that C⊥σ is a σ(λ−1)-constacyclic code of length n over Rl,k. By Theorem 3.1, we can have C⊥j is a σ(λ−1j)-constacyclic code over Fpm for j=1,2,⋯,s. □
Theorem 3.4. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k, then there exists a polynomial g(x)∈Rl,k[x] such that C=⟨g(x)⟩, where g(x)=∑sj=1ejgj(x) is the divisor of xn−λ, gj(x)∈Fpm[x] is the generator polynomial of λj-constacyclic over Cj, and gj(x) divides xn−λj for j=1,2,⋯,s.
Proof. This proof is the same as Theorem 2 in [15]. □
Corollary 3.1. Let C=⟨g(x)⟩ be a λ-constacyclic code of length n over Rl,k. Then C⊥σ=⟨∑sj=1ejσ−1(f∗j(x))⟩ and ∣C⊥σ∣=pm∑sj=1deg(gj), where fj(x)gj(x)=xn−λj, j=1,2,⋯,s.
Proof. Let C⊥j=⟨f∗j(x)⟩. By Lemma 3.1, Theorem 3.3 and Theorem 3.4, we can have
and
where j=1,2,⋯,s.
Then
and C⊥σ has the form
Let
it is easy to see that, ˜D⊆C⊥.
On the other hand, for j=1,2,⋯,s,
Thus C⊥σ⊆˜D, it implies that,
□
Theorem 3.5. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k, then C is a σ-self-orthogonal code over Rk if and only if C1,C2,⋯,Cs are σ-self-orthogonal codes over Fpm.
Proof. By Theorem 3.2, C⊥σ=∑sj=1ejC⊥σj, so C⊆C⊥σ if and only if Cj⊆C⊥σj, it implies that C is a σ-self-orthogonal code over Rl,k if and only if C1,C2,⋯,Cs are σ-self-orthogonal codes over Fpm. □
Lemma 3.2. Let C be a λ-constacyclic code of length n over Fpm, its generator polynomial is g(x). Then C is a σ-self-orthogonal code if and only if f(x)σ−1(f∗(x)) is the divisor of xn−λ, λ=±1, where f∗(x) is the generator polynomial of C⊥.
Proof. We can assume that σ(λ−1)=λ.
Since C is a λ-constantcyclic code of length n over Fpm, and C⊥σ is a σ(λ−1)-constantcyclic code of length n over Fpm, so C is a σ-self-orthogonal code must satisfy the condition C⊆C⊥σ.
Let C⊥=⟨f∗(x)⟩, where f(x)g(x)=(xn−λ) and λ=±1. C⊥σ=σ−1(C⊥)=⟨σ−1(f∗(x))⟩, C is a σ-self-orthogonal code if and only if there exists a polynomial h(x)∈Fpm[x], such that σ−1(f∗(x))h(x)=g(x), if and only if f(x)σ−1(f∗(x))h(x)=f(x)g(x)=(xn−λ) if and only if f(x)σ−1(f∗(x)) is the divisor of xn−λ. □
By Theorem 3.5 and Lemma 3.2, it is easy to get the following theorem.
Theorem 3.6. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k, then C is a σ-self-orthogonal code over Rl,k if and only if fj(x)σ−1(f∗j(x)) is the divisor of xn−λj, where f∗j(x) is the generator polynomial of C⊥j and σ(λ−1j)=λj for j=1,2,⋯,s.
4.
Quantum codes from σ-dual-containing constacyclic codes over Rl,k
Theorem 4.1. Let C=⨁sj=1ejCj be a linear code of length n over Rl,k, which order ∣C∣=pmk′ and the minimum Gray distance of C is dG, where s=(l−1)k+1. Then ϕl,k(C) is a linear code [sn,k′,dG] and ϕl,k(C)⊥σ=ϕl,k(C⊥σ). If C is a σ-self-orthogonal code over Rl,k, then ϕl,k(C) is a σ-self-orthogonal code over Fpm. Specifically, if C is a σ-self-dual code over Rl,k, then ϕl,k(C) is a σ-self-dual code over Fpm.
Proof. By the definition of ϕl,k, it is easy to know that ϕl,k is both a distance preserving map and a linear map from Rnl,k to Fsnq, it implies that ϕl,k(C) is a linear code [sn,k′,dG].
If C is a σ-self-orthogonal code, ∀x=∑sj=1ejxj∈C, ∀y=∑sj=1ejyj∈C, where xj,yj∈Fnpm, because C⊆C⊥σ, so the σ-inner product of x,y is ⟨x,y⟩σ=∑sj=1ej⟨xj,yj⟩σ=0, which implies ⟨xj,yj⟩σ=0, so
So ϕl,k(C) is a σ-self-orthogonal code over Fpm.
Let a=(a0,a1,⋯,an−1)∈C and b=(b0,b1,⋯,bn−1)∈C⊥σ, where aj=∑si=1ai,jei and bj=∑si=1bi,jei∈Rl,k for j=0,1,2,⋯,n−1, a(i)=(ai,0,ai,1,⋯,ai,n−1) and b(i)=(bi,0,bi,1,⋯,bi,n−1) for i=1,2,⋯,s.
Then
So ⟨a(i),b(i)⟩σ=0 for i=1,2,⋯,s.
Since ϕl,k(a)=(a(1),a(2),⋯,a(s)) and ϕl,k(b)=(b(1),b(2),⋯,b(s)). It follows that
So we have
As ϕl,k is a bijection, and ∣C∣=∣ϕl,k(C)∣.
Then
We have
Suppose C is a σ-self-dual code and C=C⊥σ, then
Therefore, ϕl,k(C) is a σ-self-dual code over Fpm. □
Lemma 4.1. Let C be a λ-constacyclic code of length n over Fpm, whose generator polynomial is g(x). Then, C is a σ-dual-containing code if and only if xn−λ is the divisor of σ−1(f∗(x))f(x), where f∗(x) is the generator polynomial of C⊥ and σ(λ−1)=λ.
Proof. Let C⊥=⟨f∗(x)⟩, where f(x)g(x)=(xn−λ) and σ(λ−1)=λ. Then C⊥σ=σ−1(C⊥)=⟨σ−1(f∗(x))⟩, so C is a σ-dual-containing code if and only if there exists a polynomial h(x)∈Fpm[x], such that σ−1(f∗(x))=h(x)g(x), if and only if σ−1(f∗(x))f(x)=h(x)g(x)f(x)=h(x)f(x)g(x)=h(x)(xn−λ) if and only if xn−λ is the divisor of σ−1(f∗(x))f(x). □
Theorem 4.2. (CSS construction [4]) Let C=[n,k,d] be a linear codes over Fq with C⊥⊆C, then there exists a quantum code [[n,2k−n,d]]q.
Theorem 4.3. (Hermitian construction [4]) Let C=[n,k,d] be a linear code over Fq2 with C⊥H⊆C, then there exists a quantum code [[n,2k−n,d]]q.
By Lemma 4.1 and Theorem 3.3, we can have the following theorem.
Theorem 4.4. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k, where s=(l−1)k+1. Then C⊥σ⊆C if and only if xn−λj is the divisor of σ−1(f∗j(x))fj(x), where σ(λ−1j)=λj and f∗j(x) is the generator polynomial of C⊥j for j=1,2,⋯,s.
By Lemma 4.1 and Theorem 4.4, we can have the following corollary and theorem.
Corollary 4.1. Let C=⨁sj=1ejCj be a λ-constacyclic code of length n over Rl,k. Then C⊥σ⊆C if and only if C⊥σj⊆Cj for j=1,2,⋯,s.
Theorem 4.5. Let C=⨁sj=1ejCj be a a λ-constacyclic code of length n over Rl,k, Cj is λj-constacyclic code over Fpm and C⊥σj⊆Cj, where s=(l−1)k+1, σ(λ−1j)=λj for j=1,2,⋯,s. Then ϕl,k(C)⊥σ⊆ϕl,k(C). If θt is the identity map of Fpm, then there exists a quantum code [[sn,2k′−sn,dG]]pm. If m is even, ∀a∈Fpm, then σ(a)=apm2, then there exists a quantum code [[sn,2k′−sn,dG]]pm/2, where dG is the minimum Gray weight of code C, and k′ is the dimension of the linear code ϕl,k(C).
Proof. Let C⊥σj⊆Cj and σ(λ−1j)=λj. By Corollary 4.1, we have C⊥σ⊆C, so ϕl,k(C⊥σ)⊆ϕl,k(C). By Theorem 4.1, ϕl,k(C)⊥σ=ϕl,k(C⊥σ), therefore ϕl,k(C)⊥σ⊆ϕl,k(C), and by Theorem 4.1, ϕl,k(C) is a linear code [sn,k′,d].
If θt is the identity map of Fpm, so σ-inner product is the Euclidean inner product, by Theorem 4.2, there exists a quantum code [[sn,2k′−sn,dG]]pm.
If m is even, ∀a∈Fpm, then σ(a)=apm2, so σ-inner product is the Hermitian inner product. By Theorem 4.3, there exists a quantum code [[sn,2k′−sn,dG]]pm/2. □
Example 4.1. Let n=8 and R2,2=F17[u1,u2]/⟨u21=u1,u22=u2,u1u2=u2u1=0⟩. In F17[x],
If θt is the identity map of Fpm, then σ-inner product is the Euclidean inner product. Let C be an (e1+e2+(−1)e3)-constacyclic code of length 8 over R2,2. Let g(x)=e1g1+e2g2+e3g3 be the generator polynomial of C, where g1(x)=(x+2)(x+4)(x+8),g2(x)=g3(x)=(x+3), then C1=⟨g1(x)⟩ is a cyclic code of length 8 over F17, C2=⟨g2(x)⟩ and C3=⟨g3(x)⟩ are negacyclic codes of length 8 over F17.
By Theorem 4.1, ϕ2,2(C) is a linear code over F17 with parameters [24,19,4]. By Theorem 4.5, we have C⊥⊆C, we get a quantum code [[24,14,4]]17.
Example 4.2. Let n=45 and R2,2=F5[u1,u2]/⟨u21=u1,u22=u2,u1u2=u2u1=0⟩. In F5[x],
Let C be an (e1+(−1)e2+(−1)e3)-constacyclic code of length 45 over R2,2. Let g1(x)=x2+x+1, g2(x)=g3(x)=x+1, then C1=⟨g1(x)⟩ is a cyclic code of length 45, C2=⟨g2(x)⟩ and C3=⟨g3(x)⟩ are negacyclic codes of length 45 over F5.
By Theorem 4.1, ϕ2,2(C) is a linear code over F5 with parameters [135,131,3]. By Theorem 4.5, we have C⊥⊆C, we can get a quantum code [[135,127,3]]5, which has larger dimension than [[135,63,3]]5 in [12].
Example 4.3. Let n=30 and R2,2=F5[u1,u2]/⟨u21=u1,u22=u2,u1u2=u2u1=0⟩. In F5[x],
Let C be an (e1+(−1)e2+(−1)e3)-constacyclic code of length 30 over R2,2. Let g1(x)=x2+x+1, g2(x)=g3(x)=x+3, then C1=⟨g1(x)⟩ is a cyclic code of length 30, C2=⟨g2(x)⟩ and C3=⟨g3(x)⟩ are negacyclic codes of length 30 over F5.
By Theorem 4.1, ϕ2,2(C) is a linear code over F5 with parameters [90,86,3]. By Theorem 4.5, we have C⊥⊆C, we get a quantum code [[90,82,3]]5, which has larger dimension than [[90,68,3]]5 in [11].
Example 4.4. Let n=8 and R1,2=F9[u1]/⟨u21=u1⟩, σ(a)=a3. In F9[x],
Let C be an (e1+e2)-constacyclic code of length 8 over R1,2. Let g1(x)=(x+w)(x+w2) and g2(x)=x+w3, then C1=⟨g1(x)⟩ and C2=⟨g2(x)⟩ are cyclic codes of length 8 over F9.
By Theorem 4.1, ϕ1,2(C) is a linear code over F9 with parameters [16,13,3]. By Theorem 4.5, we have C⊥σ⊆C, we can get a quantum code [[16,10,3]]3 satisfying n−k+2−2d=2.
In Table 1, we provide some new quantum codes (in the eighth column) and compare the existing codes (in the ninth column) better (by means of larger code rate or larger distance) than [11,12,13]. Further, the fifth column gives the value of units (λ1,⋯,λs), the sixth column gives the generator polynomials ⟨g1(x),⋯,gs(x)⟩, where gi(x)=anxn+an−1xn−1+⋯+a1x+a0 is denoted by anan−1⋯a1a0, e.g., "11" represents the polynomial "x2+x", the seventh column gives parameters of ϕl,k(C).
5.
Conclusions
In this article, we construct quantum codes by studying the algebraic structure of σ-self-orthogonal constacyclic codes over a new finite non-chain ring Rl,k, and our results will enrich the code sources of constructing quantum codes. As an application, we obtain some new quantum codes from σ-dual-containing constacyclic codes over Rl,k using the CSS construction or Hermitian construction and compare these codes better with the existing codes that appeared in some recent references.
Use of AI tools declaration
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
Acknowledgments
This work was supported in part by the Zhengzhou Special Fund for Basic Research and Applied Basic Research under Grant ZZSZX202111, in part by the Postgraduate Education Reform and Quality Improvement Project of Henan Province under Grant YJS2023JD6, and in part by the Soft Science Research Project of Henan Province under Grant 232400411122.
We would like to thank the referees and the editor for their careful reading the paper and valuable comments and suggestions, which improved the presentation of this manuscript.
Conflict of interest
We declare no conflicts of interest.