Many Non-Reed-Solomon Type MDS Codes From Arbitrary Genus Algebraic Curves (Hao Chen)

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Many Non-Reed-Solomon Type MDS

Codes From Arbitrary Genus


arXiv:2208.05732v3 [cs.IT] 26 Jul 2023

Algebraic Curves
Hao Chen ∗

July 27, 2023

Abstract
It is always interesting and important to construct non-Reed-Solomon
type MDS codes in coding theory and finite geometries. In this paper,
we prove that there are non-Reed-Solomon type MDS codes from ar-
bitrary genus algebraic curves. It is proved that MDS algebraic geom-
etry (AG) codes from higher genus curves are not equivalent to MDS
AG codes from lower genus curves. For genus one case, we construct
MDS AG codes of small consecutive lengths from elliptic curves. New
self-dual MDS AG codes over F2s from elliptic curves are also con-
structed. These MDS AG codes are not equivalent to Reed-Solomon
codes, not equivalent to known MDS twisted Reed-Solomon codes and
not equivalent to Roth-Lempel MDS codes.

Hence many non-equivalent MDS AG codes, which are not equiv-


alent to Reed-Solomon codes and known MDS twisted-Reed-Solomon
codes, can be obtained from arbitrary genus algebraic curves. It is
interesting open problem to construct explicit longer MDS AG codes
from maximal curves.

Index terms: MDS AG code, Reed-Solomon code, Twisted Reed-


Solomon code, Self-dual MDS code.

Hao Chen is with the College of Information Science and Technology/Cyber Security,
Jinan University, Guangzhou, Guangdong Province, 510632, China, [email protected].
The research of Hao Chen was supported by NSFC Grant 62032009.

1
1 Introduction and Preliminaries
The Hamming weight wt(a) of a vector a ∈ Fnq is the number of non-zero
coordinate positions. The Hamming distance d(a, b) between two vectors a
and b is defined to be the Hamming weight of wt(a−b). For a code C ⊂ Fnq ,
its minimum Hamming distance

d(C) = min{d(a, b), a ∈ C, b ∈ C}


a6=b

is the minimum of Hamming distances d(a, b) between any two different


codewords a and b in C. For a linear code C ⊂ Fnq , its minimum Hamming
distance is its minimum Hamming weight. For a linear [n, k, d]q code, the
Singleton bound asserts d ≤ n − k + 1. When equality holds, this code is
an MDS code. We refer to [37, 48] for the theory of error-correcting codes.
In [48, Page 317], the Chapter 11 ”MDS codes” is called ”one of the most
fascinating chapters in all of coding theory”.

The main conjecture of linear MDS codes proposed in [58] claims that the
length of a linear MDS code over Fq is at most q + 1, except some exceptional
cases. In [3] the main conjecture was proved for linear MDS codes over prime
fields. Some classification results about general MDS codes over small fields
were given in [42]. An (n, M = q n−d , d)q code is called almost MDS. A linear
almost MDS code C satisfying that the dual C⊥ is also almost MDS is called
near MDS code. It is well-known that AG codes from elliptic curves are near
MDS codes. The main conjecture of near MDS codes was proposed in [44].
For counting the number of MDS linear codes, we refer to [29, 41].

We say that two codes C1 and C2 in Fnq are equivalent if C2 can be


obtained from C1 by a permutation of coordinates and the multiplication of
a Hamming weight n vector v = (v1 , v2 , . . . , vn ) ∈ Fnq on coordinates, where
vi 6= 0 for i = 1, . . . , n. That is

C2 = {c = (c1 , . . . cn ) : (c1 , . . . , cn ) = (v1 x1 , . . . , vn xn ), x ∈ P erm(C1 )},

where P erm(C1 ) is the code obtained from C1 by a coordinate permutation.


Equivalent codes have the same code length, the dimension and the mini-
mum Hamming distance.

2
Reed-Solomon codes proposed in the 1960 paper [55] are well-known MDS
codes. Let P1 , . . . , Pn be n ≤ q distinct elements in Fq . The Reed-Solomon
code RS(n, k) is defined by

RS(n, k) = {(f (P1 ), . . . , f (Pn )) : f ∈ Fq [x], deg(f ) ≤ k − 1}.

This is a [n, k, n − k + 1]q linear MDS codes, because a degree deg(f ) ≤ k − 1


polynomial has at most k − 1 roots. These codes were called ”the greatest
codes of them all” in [32, Chapter 5]. Reed-Solomon codes are AG codes
from the genus zero curve. It should be mentioned that Reed-Solomon
codes were widely used in secret sharing, secure multiparty computation,
see [13, 18, 20, 49, 59] and distributed storage systems, see [16].

MDS codes which are not equivalent to Reed-Solomon codes are called
non-Reed-Solomon type MDS codes. It is always interesting and impor-
tant to construct non-Reed-Solomon MDS codes in coding theory and fi-
nite geometries, we refer to [8, 10, 12, 56]. One method used to distinguish
MDS codes from Reed-Solomon codes is the calculation of dimensions of
their Schur squares, see [10, 12]. The componentwise product (star prod-
uct) of t vectors xj = (xj,1 , . . . , xj,n ) ∈ Fnq , j = 1, . . . , t, is x1 ⋆ · · · ⋆ xt =
(x1,1 · · · xt,1 , . . . , x1,n · · · xt,n ) ∈ Fnq . The componentwise product of linear
codes C1 , . . . , Ct in Fnq is defined by

C1 ⋆ · · · ⋆ Ct = Σci ∈Ci Fq c1 ⋆ · · · ⋆ ct .

When C1 = C2 = C, the componentwise product C ⋆ C is called the Schur


square of the linear code C. It is clear that Schur squares of equivalent
linear codes are equivalent. Hence the dimension and the minimum Ham-
ming distance of Schur squares are invariants of two equivalent linear codes.
The dimension of the Schur square of the Reed-Solomon [n, k, n − k + 1]q ,
2k ≤ n ≤ q + 1, is 2k − 1. Then any MDS [n, k, n − k + 1]q code C satisfying
dim(C ⋆ C) ≥ 2k, is a non-Reed-Solomon type MDS code, see [12].

The interest to construct non-Reed-Solomon type MDS codes comes from


the invention of the twisted Gabidulin codes in the rank-metric. Rank-metric
codes attaining the similar Singleton bound |C| ≤ q n(n−dr (C)+1) is called max-
imal rank distance (MRD) codes, where dr (C) is the minimum rank-distance
of the rank-metric code C. Gabidulin codes are well-known MRD codes and
can be thought as the rank-metric analogue of Reed-Solomon codes, see [27].

3
It is a great achievement that some twisted Gabidulin codes were found
in [60]. These rank-metric codes are MRD codes, which are not equivalent
to Gabidulin codes, see [52,60]. Then many papers, see e.g., [17,46,65], were
published for constructing new twisted Gabidulin codes.

Motivated by the progress of twisted Gabidulin codes, twisted Reed-


Solomon codes were introduced by P. Beelen, S. Puchsinger, J. Nielson and
J. Rosenkilde in [10, 12]. Some of these twisted Reed-Solomon codes are
MDS code, see Section 4 of [12] and [45], and MDS twisted Reed-Solomon
codes are non Reed-Solomon type MDS codes. However it seems that there
are strong restrictions on lengths of these MDS twisted Reed-Solomon codes,
see [12]. In fact, in an earlier paper [56], some non-Reed-Solomon type MDS
codes were constructed and studied. Twisted Hermitian codes from Hermi-
tian curves were introduced in a recent paper [2].

The (Euclid) dual of a linear code C ⊂ Fnq is C⊥ = {c = (c1 , . . . , cn ) :


Σni=1 ci xi = 0, ∀x = (x1 , . . . , xn ) ∈ C}. T
A linear code is called self dual if
C = C⊥ . In general the linear code C C⊥ is called the hull of C. The
Hermitian dual of a linear code C ⊂ Fnq2 is

C⊥h = {c = (c1 , . . . , cn ) : Σni=1 ci xqi = 0, ∀x = (x1 , . . . , xn ) ∈ C}.

It is clear C⊥h = (C⊥ )q , where

Cq = {(cq1 , . . . , cqn ) : (c1 . . . , cn ) ∈ C}.

The minimum distance of the Euclid dual is called the dual distance and is
denoted by d⊥ . The minimum distance of the Hermitian dual is the same as
d⊥ . A linear code C ⊂ Fnq2 is called Hermitian self-dual if C = C⊥h . The
intersection C C⊥h is called the Hermitian hull of this code C. We refer
T
to [22,23] and [37, Chapter 9] for earlier results about self-dual and Hermitian
self-dual codes over small fields. From the Calderbank-Shor-Steane (CSS)
construction of entanglement-assisted quantum error correction (EAQEC)
codes in [15], Euclidean and Hermitian self-dual MDS codes can be used to
construct MDS EAQEC codes.

The construction of new self-dual MDS codes or near MDS codes has
been a long active topic in coding theory, see [9, 26, 30, 31, 36, 38, 39, 53, 68].

4
On the other hand the construction of Hermitian self-orthogonal (or dual-
containing) MDS codes had been active for the purpose to construct MDS
quantum codes, see [4–6, 34, 40] and references therein. Since the introduc-
tion of twisted Reed-Solomon codes in [10,12], the construction of non-Reed-
Solomon self-dual MDS codes from twisted Reed-Solomon codes has been
given in [36, 64]. These codes are not equivalent to the Reed-Solomon codes
and can be thought as new self-dual MDS codes. From the view of coding
theory, it is always interesting to construct non-Reed-Solomon type MDS
codes and non-Reed-Solomon type self-dual MDS codes.

Let X be an absolutely irreducible, smooth and genus g curve defined


over Fq . Let P = {P1 , . . . , Pn } be the set of n distinct rational points of
X over Fq . Let G be a rational divisor over Fq of degree deg(G) satisfying
2g − 2 < deg(G) < n and
\
support(G) P = ∅.

Let L(G) be the function space associated with the divisor G, that is, L(G)
is the space of all rational functions f satisfying (f ) + G ≥ 0, where (f ) is
the divisor associated with f . The algebraic geometry code (functional code)
associated with G, P = {P1 , . . . , Pn } is defined by

C(P, G, X) = C(P1 , . . . , Pn , G, X) = {(f (P1 ), . . . , f (Pn )) : f ∈ L(G)}.

The dimension of this code is

k = deg(G) − g + 1

follows from the Riemann-Roch Theorem. The minimum Hamming distance


is
d ≥ n − deg(G).
Algebraic-geometric residual code CΩ (P1 , . . . , Pn , G, X) with the dimension
k = n − m + g − 1 and minimum Hamming distance d ≥ m − 2g + 2 can be
defined, we refer to [37,66] for the detail. It is the dual code of the functional
code of the dimension m − g + 1. The AG codes from elliptic curves satisfy
k + d ≥ deg(G) + n − deg(G) = n. Hence these elliptic curve codes are near
MDS codes.

5
A divisor G = Σmi Gi where Gi ’s are points of the curve, is called effective
if mi ≥ 0. Two effective divisor G1 and G2 are called linear equivalent if
there is a rational function f such that the divisor (f ) associated with f is
of the form
(f ) = G1 − G2 .
It is clear that for two linear equivalent divisors G1 and G2 , the AG codes
C(P1 , . . . , Pn , G1 , X) and C(P1 , . . . , Pn , G2 , X) are equivalent linear codes.

It is well-known that AG codes from elliptic curves are best examples


of linear codes with the Singleton defect n + 1 − d − k = 1. However it is
also well-known that MDS codes can be obtained from elliptic curve codes
if the evaluation points are carefully chosen. This is similar to many works
on Reed-Solomon codes in which evaluation points have to be determined
in a complicated pattern, see [32] and references therein. The main conjec-
ture for MDS codes from curves of genus 1 (elliptic) and 2 was proved in
old papers [50, 51], also see [67]. The main conjecture for MDS AG codes
arising from hyper-elliptic curves was proved independently in [19] and [14].
For a tighter upper bound on the maximal length of MDS codes from ellip-
tic curves, we refer to a recent paper [33]. The self-dual near MDS codes
from elliptic curves were constructed in [39] and some non-extendable near
MDS codes from elliptic curves were constructed in a recent paper [1]. We
also refer to [25] for the earlier work on near MDS codes. In [8] some MDS
[n, k, n − k + 1]q codes with the covering radius n − k − 1 were constructed as
extended codes of MDS codes from elliptic curves. It is well-known that the
covering radius of the Reed-Solomon [n, k, n−k + 1]q code is n−k. Therefore
these MDS codes in [8] are non-Reed-Solomon type MDS codes. It is inter-
esting and important to ask the following question about MDS AG codes.

Problem. Can MDS AG codes be obtained from arbitrary genus curves


by choosing evaluation points carefully? Are these MDS AG codes non-Reed-
Solomon type? Moreover are these MDS AG codes equivalent to each other?

In this paper, we prove that there are many non-Reed-Solomon type


MDS codes over Fq from arbitrary genus curves. It is proved that MDS
AG codes from genus g1 curves are not equivalent to MDS AG codes from
genus g2 curves, if g1 > g2 . In the range of lengths n satisfying n ≤ q 1/4
and gcd(n, q) = 1, non-Reed-Solomon type MDS elliptic curve codes can be

6
constructed for consecutive lengths. These codes can be constructed as AG
codes from elliptic curves E defined over Fq by choosing the evaluation point
set as a coset of a subgroup of E(Fq ). They are not equivalent to Reed-
Solomon codes, not equivalent to the known MDS twisted Reed-Solomon
codes and not equivalent to Roth-Lempel MDS codes in [56]. Moreover
there are non-Reed-Solomon type MDS elliptic curve codes with the same
length and the same dimension. Many new self-dual non-Reed-Solomon type
MDS codes with various lengths are also constructed. It is always good to
understand that algebraic geometry codes can give us more interesting non-
Reed-Solomon type MDS codes.

2 AG codes from elliptic curves


2.1 Elliptic curves over finite fields
In this subsection, we recall basic facts about elliptic curves defined over a
finite field, which are mainly from the paper [57].

Let E be an elliptic curve defined over Fq . It is well-known that when


q is not a power of 2 or 3, then elliptic curves over Fq can be realized as a
non-singular plane cubic curve. Let E(Fq ) be the set of all Fq -rational points
of E. The number |E(Fq )| of its rational points over Fq satisfies the Hasse
bound

|q + 1 − |E(Fq )|| ≤ 2 q.
For any positive real number x we set

x− = x + 1 − 2 x,

and √
x+ = x = 1 + 2 x.
If q = p is a prime number it follows from the result in [24, 57] that for
any positive integer N satisfying p− < N < p+ , there is an elliptic curve E
defined over Fp such that the number of Fp -rational points of E satisfying

|E(Fp )| = N.

7
It is well-known there is an Abelian group structure on E(Fq ). As a
group E(Fq ) ≃ Z/mZ × Z/mkZ for some positive integers m and k. We
refer to [57, 62] for the detail. More accurately the following two results were
proved in [57].

Theorem 2.1 (Theorem 1a in [57]). All the possible orders |E(Fq )|


of an elliptic curve E defined over Fq , where q = pn is a prime power, are
given by
|E(Fq )| = 1 + q − β,

where β is an integer with |β| ≤ 2 q satisfying one of the following condi-
tions:
(a) gcd(β, p) = 1;

(b) If n is even: β = ±2 q;

(c) If n is even and p 6= 1 mod 3: β = ± q;
n+1
(d) If n is odd and p = 2 or 3: β = ±p 2 ;
(e) If either (i) n is odd or (ii) n is even, and p 6= 1 mod 4: β = 0.

All possible group structures of elliptic curves over Fq were also deter-
mined in [57].

Theorem 2.2 (Theorem 3 in [57]). Let E be anQelliptic curve over


a finite field Fq with q = pn elements. Let |E(Fq )| = l lhl be the prime
factoring. Then all the possible groups E(Fq ) with the order |E(Fq )| are the
following, Y
Z/php Z × (Z/lal Z × Z/lhl −al Z),
l6=p

with
(a) In case (b) of Theorem 3.1: Each al is equal to h2l ;
(b) In cases (a), (c), (d), (e) of Theorem 3.1: al is an arbitrary integer
satisfying
hl
0 ≤ al ≤ min{vl (q − 1), [ ]},
2
where vl (q − 1) is the order of prime factor l in q − 1.

From Theorem 2.1 and 2.2, there are a lot of elliptic curves with dif-
ferent orders |E(Fq )| and different Abelian group structures. Therefore we

8
can construct many MDS elliptic curve codes from the following Theorem 2.3.

2.2 MDS codes from elliptic curves


The following basic facts about AG codes from elliptic curves are well-known,
for example, see [8,51]. Let E be an elliptic curve defined over the finite field
Fq . Let P0 , P1 , . . . , Pn be n+1 rational points. We take the divisor G = mP0 ,
where 0 < m ≤ n − 1. Then an AG code C(P1 , . . . , Pn , G, E) from this el-
liptic curve E is constructed. This is a linear [n, m, ≥ n − m]q code.

Theorem 2.3. 1) The minimum Hamming distance of C(P1 , . . . , Pn , G, E)


is n − m if and only if there are m distinct points Pi1 , . . . , Pim in the set
{P1 , . . . , Pn } such that the effective divisor Pi1 + · · · + Pim is linear equivalent
to the divisor mP0 .
2) If there is an effective rational divisor G1 satisfying that G1 + G is lin-
early equivalent to the effective divisor P1 + · · · + Pn . The the dual code of
C(P1 , . . . , Pn , G, E) is equivalent to the AG code C(P1 , . . . , Pn , G1 , E).

Proof. 1) is direct from the definition of AG code. 2) is direct from the


definition of residual code, see [51].

Notice that the Jacobian of the elliptic curve E is itself, so the condition
in Theorem 2.3 1) is equivalent to if Pi1 + · · · + Pim = mP0 is valid in the
group E(Fq ), we refer to [51].

2.3 The Schur squares of elliptic curve codes


Let E be an elliptic curve defined over Fq and P0 is a rational point. We
now discuss some properties of the linear space L(mP0 ) of rational functions
f satisfying (f ) + mP0 ≥ 0. The following property is well-known.

Proposition 2.1. L(P0 ) is the one dimension linear space of all constant
functions. For each m ≥ 2, there is one rational function f in L(mP0 )/L((m−
1)P0 ).

9
Proof. If there is a non-constant rational function f in the space L(P0 ),
then (f ) = Q − P for some rational point Q ∈ E(Fq ). That is, P0 is linearly
equivalent to another rational point Q ∈ E(Fq ). This is a contradiction to
the fact that E(Fq ) is the Jacobian

{G : deg(G = 0}/linear − equivalence

of E.

On the other hand from the Riemann-Roch theorem, dim(L(mP0 )) = m


when m ≥ 2g − 1 = 1. The second conclusion follows immediately.

Theorem 2.4. The dimension of the Schur square of an one-point AG


code C(P1 , . . . , Pn , mP0 , E) from an elliptic curve E is 2m if 6 ≤ 2m ≤ n.

Proof. Let f1 be the constant function in L(mP0 ), fi , i = 2, . . . , m be


the rational function with the i-th order pole at the point P0 . It is clear
that C(P1 , . . . , Pn , mP0 , E) = {(f (P1), . . . , f (Pn )) : f = Σai fI , ai ∈ Fq }. It
is clear that the Schur square of C(P1 , . . . , Pn , mP0 , E) is in the AG code
C(P1 , . . . , Pn , 2mP0 , E). On the other hand it is easy to verify that the eval-
uation codewords of f1 , f2 , . . . , fm are in the Schur square. The rational
function fi fj , where i ≥ 2, j ≥ 2 satisfying m + 1 ≤ i + j ≤ 2m, can have
w-th order pole at the point P0 for w = m + 1, . . . , 2m. Then the Schur
square of C(P1 , . . . , Pn , mP0 , E) is C(P1 , . . . , Pn , 2mP0 , E). The conclusion
is proved.

It is obvious that the minimum Hamming distance of the Schur squares


of an elliptic curve [n, k, ≥ n−k]q code is at least n−2k. Theorem 2.4 can be
generalized to one point AG codes from higher genus curves, see Theorem 4.1.

3 Non-Reed-Solomon type MDS codes from


elliptic curves
Let q be a prime power. In this section, the set P = {P1 , . . . , Pn } is the dis-
joint union of several cosets of a subgroup of E(Fq ). We show that there are
many MDS AG codes of consecutive lengths from elliptic curves E defined

10
over Fq . These MDS codes are not equivalent to the Reed-Solomon codes
from dimensions of their Schur squares. On the other hand, MDS twisted
Reed-Solomon codes have been only constructed for some special lengths n
satisfying that n is a divisor of q − 1 or gcd(n, q − 1) = n2 , or n is a divisor
of q, see [12]. Therefore it is obvious that there are more non-Reed-Solomon
MDS AG codes from elliptic curves than twisted Reed-Solomon codes. In
the following part, P0 is the zero element in the group E(Fq ).

Theorem 3.1. 1) Let E1 ⊂ E(Fq ) be a subgroup of the order n1 and


b ∈ E(Fq ) be a nonzero element such that the n2 order cyclic subgroup < b >
generated by b intersects E1 at the zero element. Then for any positive inte-
ger m ≤ n2 − 1, the elliptic curve code C(b + E1 , mP0 , E) is an MDS code.
2) Let E1 ⊂ E(Fq ) be a subgroup of the order n1 and bi ∈ E(Fq ), i = 1, . . . , t,
be t nonzero elements in E(Fq ) such that the set {m1 b1 +· · ·+mt bt : m1 +· · ·+
mt = m} intersects the subgroup E1 at the zero element. Set P be the union
of t cosets b1 + E1 , . . . , bt + E1 . Then the elliptic curve code C(P, mP0 , E) is
an MDS code.

Proof. The sum Pi1 + · · · + Pim of m rational points in the coset b + E1


is of the form mb + P , where P ∈ E1 . From the condition that the the order
n2 cyclic subgroup < b >⊂ E(Fq ) intersects E1 only at zero element, this
sum is not zero element in E(Fq ), since m ≤ n2 − 1. The conclusion follows
directly. The second conclusion can be proved similarly.

Example 3.1. Let q = p be a prime number, from the classical re-


sult in [24] for any given positive integer satisfying p− ≤ N ≤ p+ , there is
an elliptic curve E defined over Fp such that N = |E(Fp )|. Suppose that
N = p1 p2 be the product of two different prime numbers, where gcd(p, pi ) = 1
for i = 1, 2. There is an elliptic curve defined over Fp such that the group
structure of E(Fp ) is Z/p1 Z ⊕ Z/p2 Z from Theorem 2.2. Let E1 be the cyclic
subgroup Z/p1 Z × 0 of the order p1 . Then for each element of the form
(0, x) ∈ 0 × Z/p2 Z, the elliptic curve code C(b + E1 , mP0 , E) is an MDS code
for 1 ≤ m ≤ p1 .

Let b1 < · · · < bt be t(≤ p2 −1) distinct nonzero elements in 0×Z/p2 Z and
m be a positive integer satisfying mbt ≤ p2 . Then m1 b1 + · · · + mt bt ≤ mbt <
p2 , where m1 +· · ·+mt = m. Let P be the union of t cosets b1 +E1 , . . . , bt +E1 .
The elliptic curve code C(P, mP0 , E) is an MDS code.

11
It is obvious when t ≥ p22+1 , then m cannot bigger than or equal to two.
Hence if we want to construct a dimension 2 MDS elliptic curve code in this
example, the length is smaller than p12p2 ≤ q + 1.

On the other hand we can take E1 be the subgroup 0 × Z/p2 Z, then


for 1 × 0 = b, the elliptic curve code C(P, mP0 , E) is an MDS code when
m ≤ p1 − 1.

When p = 19, we can get the following non-Reed-Solomon type MDS


codes, an MDS [4, 2, 3]19 code from an elliptic curve with 12 rational points,
an MDS [5, 2, 4]19 code from an elliptic curve with 15 rational points, an
MDS [6, 2, 5]19 code from an elliptic curve with 18 rational points, an MDS
[5, 2, 4]19 code from an elliptic curve with 20 rational points, and an MDS
[6, 3, 4]19 code from an elliptic curve with 24 rational points.

Corollary 3.1. Let E be an elliptic curve with the order |E(Fq )| = l1 l2


where l1 < l2 are two positive integers satisfying gcd(li , q) = 1, i = 1, 2, and
gcd(l1 , l2 ) = 1. Then there exists an order l1 subgroup E1 of E(Fq ) and one
coset P of E1 , the elliptic curve code C(P, mP0 , E) is an MDS code for each
m satisfying 2 ≤ m ≤ l1 − 1.

Proof. We have an elliptic curve E defined over Fq such that E(Fq )


has a subgroup of the form Z/l1 l2 Z from Theorem 2.2 and the condition
gcd(l1 , q) = gcd(l2 , q) = 1. From the condition gcd(l1 , l2 ) = 1, the group
E(Fq ) is of the form Z/l1 Z ⊕ Z/l2 Z. In this case, by setting E1 be the cyclic
subgroup Z/l1 Z × 0 of the order l1 and b be the for 0 × x, where x is the
generator of Z/l2 Z, the conclusion follows from Theorem 3.1.

Corollary 3.2. Let n be a positive integer satisfying 6 ≤ n ≤ q 1/4 ,


gcd(n, q) = 1, and m be any positive integer satisfying 2 ≤ m ≤ n2 . Then
there is an MDS elliptic curve code over Fq with the length n and dimension
m. This MDS code is not equivalent to the Reed-Solomon [n, m, n − m + 1]q
code when m ≥ 3.

Proof. First of all, we can find an elliptic curve of the order nl where l is
a positive integer satisfying gcd(n, l) = 1. Then the conclusion follows from
Theorem 3.1 and Theorem 2.4 immediately.

12
Since the general MDS conditions about twisted Reed-Solomon codes are
restricted to subfields or subgroups as in [12], certainly many MDS codes
constructed in Corollary 3.2 are not equivalent to these MDS twisted Reed-
Solomon codes, or there is no known MDS twisted Reed-Solomon code with
the corresponding length. Therefore many new non-Reed-Solomon type MDS
codes from elliptic curves are constructed for consecutive lengths. We can
observe the following example of MDS twisted Reed-Solomon codes as in [12].

Let n and k be two positive integers satisfying n ≤ q − 1 and k ≤ n − 1.


Let α1 , . . . , αn be n distinct elements in the finite field Fq such that α =
{α1 , . . . , αn } is a n element subset Fq . Let η be a nonzero element of Fq . Set
g0 = 1 + ηxk , g1 = x, . . . , gk−1 = xk−1 . Let P(η, k) be the linear span over
Fq by g0 , . . . , gk−1 . The linear [n, k]q code Cα,η,k is the evaluation code of
these polynomials in P(η, k) at the above n elements in the subset α. The
dimension of the Schur square of Cα,η,k is at least 2k. Thus this code is not
equivalent to a Reed-Solomon code when 2k ≤ n. It is not hard to verify
that if η can not be represented as the product of any k elements in α, that
is, Y
η 6= α ij ,
1≤j≤k

for any k distinct αi1 , . . . , αik ∈ α, this code Cα,η,k is an MDS code.

The above condition about the set α is strong, if we want to construct


MDS twisted Reed-Solomon codes for large dimensions. From the construc-
tion in Theorem 3.1, the MDS condition for elliptic curve codes is not so
strong. Hence there are many MDS elliptic curve codes, which are not equiv-
alent to known MDS twisted Reed-Solomon codes, or there is no known MDS
twisted Reed-Solomon code with the corresponding length.

Corollary 3.3. Let p be an odd prime number. √


Then there are MDS
√ √ ⌊ p⌋ √
[⌊ p⌋, k, ⌊ p⌋ − k + 1]p codes for k = 2, . . . , 2 . There are MDS [⌊ p⌋ +
√ √
⌊ p⌋
1, k, ⌊ p⌋ − k + 2]p codes for k = 2, . . . , 2 . These codes are not equivalent
to Reed-Solomon codes when k ≥ 3.

Proof. Set n = ⌊ p⌋, it is clear that
√ √
p − p ≤ n(n + 1) ≤ p + p,

13
gcd(n, p) = gcd(n + 1, p) = 1. Then there is an elliptic curve E defined
over Fp with n(n + 1) rational points from Theorem 2.1 and 2.2, with the
group structure |E(Fp ) = Z/nZ ⊕ Z/(n + 1)Z. The conclusion follows from
Theorem 3.1.

Now we can use supersingular elliptic curves to construct non-equivalent


MDS codes. Let us recall some basic facts about supersingular elliptic curves.
From [62, page 152], the elliptic curve E defined over Fp by y 2 = x3 + 1 is
supersingular when p ≡ 2 mod 3, and the elliptic curve E defined over Fp by
y 2 = x3 + x is supersingular when p ≡ 3 mod 4. For a supersinglular elliptic
curve E defined over Ep it is known that

|E(Fpn )| = pn + 1,

if n is an odd positive integer, or


n n
|E(Fpn )| = (p 2 − (−1) 2 )2 ,

if n is an even positive integer, see page 155 of [62]. We have the following
result.

Corollary 3.4. Let p be an odd prime satisfying


n
√ pn≡ 2 mod 3 or p ≡ 3
mod 4. Let N be a factor of p + 1 satisfying N < p + 1, when n is odd,
n n n n
and N be a factor of (p 2 − (−1) 2 )2 satisfying N < p 2 − (−1) 2 when n is
even. Then there is an MDS [N, k, N − k + 1]pn linear code. This MDS code
is not equivalent to the Reed-Solomon code when k ≤ N2 − 1.

Proof. There is no factor pl in |E(Fpn )|, then we can take a(l) = 0 for
any prime factor l of |E(Fpn )| in Theorem 2.2. The conclusion follows from
Theorem 3.1 1) directly.

Some non-Reed-Solomon MDS codes were introduced in [56], it is not


hard to verify that the minimum Hamming distances of the Schur square
of Roth-Lemple MDS codes are very small. Hence many of our constructed
MDS elliptic curve codes are not equivalent to Roth-Lempel MDS codes.

Moreover from Theorem 2.3 and 2.4, we can construct some MDS ellip-
tic curve codes such that their Schur squares are MDS codes or not MDS

14
codes. Therefore some non-equivalent MDS elliptic curve codes from one el-
liptic curve or two different elliptic curves (with different elliptic curve group
structures) of the same length and the same dimension can be constructed.

4 Non-Reed-Solomon type MDS codes from


higher genus g ≥ 2 curves
In this section, we prove that MDS codes from a genus g > g ′ curve is not
equivalent to MDS codes from a genus g ′ curve, when m ≥ 4g. Then MDS
codes from genus g ≥ 2 curves are essentially new if the degree of G is bigger
than or equal to 4g. It is proved that at least short length new non-Reed-
Solomon type MDS codes from higher genus g ≥ 2 curves can be constructed.

The following result is a direct generalization of Theorem 2.4.

Theorem 4.1. Let X be a genus g curve defined over Fq , P0 , P1 , . . . , Pn


are n + 1 rational point of X, m is a positive integer satisfying 4g < m < n
and 2m < n. Then the dimension of the Schur square of the dimension k one
point function code C(P1 , . . . , Pn , mP0 , X) is exactly 2k +g −1. In particular,
MDS AG codes from a genus g1 curves are not equivalent to MDS AG codes
from a genus g2 curve if g1 > g2 .

Proof. We recall the well-known Weierstrass gap theorem, see Chapter


6 of [35], except g positive integers 1 = α1 < α2 < · · · < αg ≤ 2g − 1, there
is a rational function f ∈ L(mP0 ) such that the pole part of the divisor (f )
associated with the function f is exactly m′ P0 , where m′ 6= αi , i = 1, . . . , g.
Therefore for each positive integer m′ ≥ 2g we can find a rational function
f ∈ L(mP0 ) such that the pole part of f is exactly m′ P0 . Then the Schur
square of C(P1 , . . . , Pn , mP0 , X) is exactly C(P1 , . . . , Pn , 2mP0 , X) with the
dimension 2m − g + 1 = 2(m − g + 1) + g − 1 = 2k + g − 1.

Now we observe the MDS condition for AG codes from genus g curves.
In the case g = 1, this is just Theorem 2.3 1).

Theorem 4.2. Let X be a genus g curve defined over Fq , P0 , P1 , . . . , Pn

15
are n+1 rational point of X, m is a positive integer satisfying 2g−1 < m < n.
Then the one point function code C(P1 , . . . , Pn , mP0 , X) is MDS if and only
if the following MDS condition holds.

MDS condition for AG codes: For every m−g+1 = k different points


Pi1 , . . . , Pik among P1 , . . . , Pn , there is no degree g − 1 rational effective di-
visor G′ such that the divisor Pi1 + · · · + Pim−g+1 + G′ is linear equivalent to
the divisor mP0 .

Proof. We only need to prove that there is no weight w codeword, where


n − m ≤ w ≤ n − m + g − 1. This is obvious since there is no weight n − m + i
codeword in this function code for i = 0, . . . , g − 1, from the above MDS
condition.

The existence of such n rational points P1 , . . . , Pn satisfying the above


MDS condition can be proved by a counting argument, at least for small
n ≥ 5g. We can consider the above MDS condition for the projective imbed-
ding Φ of the curve X in Pm−g defined by the divisor mP0 . Let Φ(X) be the
image of this curve X in Pm−g . Then the above MDS condition is equivalent
to the following condition, also see [67].

MDS condition for embedding: To find n rational points P1 , . . . , Pn


in Φ(X), such that there is no m − g + 1 different points Pi1 , . . . , Pim−g+1
among them, which are in a hyperplane of Pm−g .

Therefore the following result is direct from a simple counting argument.


The existence of short length MDS codes from higher genus curves is proved.

Theorem 4.3. Let X be a genus g ≥ 2 curve defined over Fq , and m


be a positive integer satisfying m ≥ 4g. If Φ(X) ⊂ Pm−g defined by the
linear system mP0 is a non-singular curve with N rational points which are
images of rational
 points of X. If n is a positive integer satisfying m < n
n
and m · < N, then there exists a length n and dimension m − g + 1
m−g
MDS code from X. These MDS codes are not equivalent to Reed-Solomon
codes.

Proof. For each different m−g rational points in general position amonge

16
chosen evaluation points, we determine a hyperplane in Pm−g . This hyper-
plane interests Φ(X) at m points. Then the conclusion follows directly.

In particular, when m = O(g) is fixed then new non-Reed-Solomon MDS


1
codes of lengths n ≤ O(q m−g ) over Fq of the dimension m − g + 1 can be
constructed from Theorem 4.3, when q tends to the infinity.

It would be interesting to construct explicit longer MDS AG codes from


the Hermitian curve, recent constructed maximal curves in [11]. This could
give explicit new non-Reed-Solomon type MDS codes from higher genus
curves, which are longer than MDS codes from Theorem 4.3. It is also
easy to generalize Theorem 4.1 to two point AG codes, see [7, 43], and to
construct more new longer non-Reed-Solomon type MDS codes from higher
genus curves.

5 Self-dual MDS codes from elliptic curves


over finite fields F2s .
In this section we restrict to the elliptic curves E defined over a finite field
F2s of the characteristic 2. The set P = {P1 , . . . , Pn } is one coset of a
subgroup E1 ⊂ E(F2s ). We show that there are many self-dual MDS ellip-
tic curve codes from elliptic curves E defined over F2s . We always are in
the case a) of Theorem 2.1. That is, β is an odd positive integer satisfy-
ing that gcd(2 − β, 2s − 1) > 1 has an odd divisor. Then the group order
|E(F2s )| = 2s + 1 − β is an even number, and 2s + 1 − β = 2s − 1 + 2 − β,
gcd(|E(F2s )|, 2s − 1) > 1 has an odd divisor. Actually since β can be any
odd number in the case a) of Theorem 2.1, when q = 2s , if s is a composite
number s = s1 s2 it is obvious we can take 2 − β ≡ 0 mod (2s1 − 1). Hence
there are many such elliptic curves with the desired group orders.

Therefore we need the following conditions to construct self-dual MDS


elliptic curve codes.
1) q = 2s1 s2 , 2 − β ≡ 0 mod 2s1 − 1;
2) β ≡ 1 mod 8;
3) An elliptic curve defined over F2s1 s2 with 2s1 s2 + 1 − β rational points;

17
Since 2s1 s2 + 1 − β ≡ 0 mod 8, the exponent h2 of 2 in the prime factor
decomposition of
|E(F2s1 s2 )| = 2s1 s2 + 1 − β,
is at least h2 ≥ 3.

4) The above elliptic curve is of the group structure Z/2h2 Z ⊕ Z/LZ,


where h2 ≥ 3 and L is an odd positive integer.

From Theorem 2.1 and 2.2 there are many such an elliptic curve. Then
the group order of the elliptic curve is

|E(F2s1 s2 )| = 2h2 L,

where h2 ≥ 3 and L is an odd positive integer.

Theorem 5.1. Let q = 2s1 s2 as above and L′ be an odd divisor of L,


n = 2t L′ where t ≤ h2 − 1. There exists a self-dual [n, n2 , n2 + 1]2s1 s2 MDS
code equivalent to an elliptic curve code from E, which is not equivalent to
the Reed-Solomon code.

Proof. From Theorem 2.2, we can find elliptic curve such that for any
odd divisor L′ |L, there is an order L′ subgroup E2 ⊂ E(F2s1 s2 ). Therefore
we have an order 2t L′ subgroup E1 of the form (i2h2 −t θ) × E2 , where θ is the
generator of the cyclic subgroup Z/2h2 Z ⊂ E(F2s1 s2 ), and i = 0, 1, . . . , 2t − 1.

Then set b = 2h2 −1−t θ × 0 ⊂ E(F2s1 s2 ). The coset P = b + E1 has 2t L′


elements. The sum of arbitrary n2 different elements in P is of the form
2t−1 L′ b + P , where P is an element in the subgroup E1 . It is clear that
2t−1 L′ b + P is of the form

2h2 −2 L′ θ × 0 + P.

This is not zero. Therefore the elliptic curve code C(P, 2tL′ P0 , E) is an MDS
code from Theorem 2.3. Here P0 is the zero element of the group E(F2s1 s2 ).

The sum of all elements in the coset E1 is of the form −2h2 −1 L′ θ × 0,


since the sum of all elements in an order L′ subgroup, L′ odd, is zero, and

18
the sum of all elements in Z/2u Z is −2u−1 . Then the sum of all elements in
P is 2h2 −1 L′ θ × 0 − 2h2 −1 L′ θ × 0 is zero element P0 . Therefore from Theorem
2.3. 2), the dual code of C(P, 2t−1 L′ P0 , E) is equivalent to a linear code
C(P, 2t−1 L′ P0 , E). Suppose that the dual code is of the form
v·C(P, 2t−1L′ P0 , E) = {(v1 c1 , . . . , vn cn ) : (c1 , c2 , . . . , cn ) ∈ C(P, 2t−1L′ P0 , E)},
where v = (v1 , . . . , vn ) ∈ Fn2s1 s2 is a Hamming weight n vector.

Since this field is of characteristic 2, each element vi is a square, set


vi = vi′2 , i = 1, . . . , n. Set v′ = (v1′ , . . . , vn′ ), the equivalent code v′ ·
C(P, 2t−1 L′ P0 , E) is a self-dual code. Actually (v′ · C(P, 2t−1L′ P0 , E))⊥ =
1
v′
· C(P, 2t−1L′ P0 , E)⊥ = v′ · C(P, 2t−1 L′ P0 , E). The dimension of the Schur
square of this self-dual code is exactly n. It is not equivalent to a Reed-
Solomon [n, n2 , n2 + 1]2s1 s2 code, the dimension of its Schur square is n − 1.
The conclusion is proved.

Notice that L′ can be any odd divisor of the group order |E(F2s1 s2 )|, there
are indeed many self-dual MDS codes which are equivalent to elliptic curve
codes. Actually self-dual MDS elliptic curve codes over F2s of√ the length
s 2s ⌋
4L, where L is any odd positive number in the range [1, 2 +1+⌊2 8
], can be
constructed. Hence there are many new self-dual MDS elliptic curve codes
over the finite field F2s , which are not equivalent to self-dual Reed-Solomon
codes or self-dual twisted Reed-Solomon codes.

From the result in [21], it is easy to construct equivalent LCD MDS codes
from self-dual MDS elliptic curve codes.

From Theorem 5.1 and the CSS construction of EAQEC codes in [15],
the following results follows immediately.

Corollary 5.1. Let q = 2s1 s2 be an even prime power and |E(Fq )| = 2h2 ·L
as in above, n be a positive integer of the form 2t L′ where L′ is an odd divisor
of the group order E(F2s1 s2 ) and t ≤ h2 − 1, and k be a positive integer sat-
isfying n2 ≤ k ≤ n − 1, and h be a nonnegative integer satisfying 0 ≤ h ≤ n2 ,
there exists an MDS EAQEC [[n, k − h, n − k + 1, n − k − h]]2s1 s2 code.

Proof. From Theorem 5.1 we have an equivalent MDS self-dual elliptic


curve [n, n2 , n2 + 1]2s1 s2 code. From the result in [21] we have a linear MDS

19
[n, n2 , n2 + 1]2s1 s2 code with the h-dimension hull, where 0 ≤ h ≤ n2 . Then the
conclusion follows from the CSS construction of EAQEC codes immediately.

Notice that the MDS elliptic curve codes in Theorem 5.1 are not equiv-
alent to Reed-Solomon code, then these MDS EAQEC codes in Corollary
5.1 are new MDS EAQEC codes, comparing to previous constructed MDS
EAQEC codes from generalized Reed-Solomon codes.

6 Conclusion and unsolved problems


In this paper, we proved that there are many MDS AG codes from arbitrary
genus algebraic curves, and self-dual MDS AG codes from elliptic curves.
These MDS codes are not equivalent to Reed-Solomon codes, not equivalent
to known MDS twisted Reed-Solomon codes and not equivalent to Roth-
Lempel MDS codes. This showed that MDS AG codes can be obtained from
carefully chosen evaluation points in an arbitrary genus curve. The following
two problems about longer MDS AG codes seem interesting.

1) What is the range of lengths such that there are non-Reed-Solomon


type MDS AG codes. Ranges of lengths of MDS elliptic curve codes in Corol-
lary 3.3 and 3.4 are obviously not optimal. Are there non-Reed-Solomon type
MDS codes from elliptic curve over prime field Fp with any given length
n ≤ p3 ?

2) Can longer explicit MDS [n, k, n − k + 1]q codes be constructed as one


point AG codes from genus g ≥ 2 curves for each genus g ≥ 2? We proved
that there are very short length MDS codes from higher genus curves. It is
interesting to construct longer MDS AG codes from maximal curves, with
explicit evaluation points.

References
[1] A. Aguglia, L. Giuzzi and A. Sonnino, Near MDS codes from elliptic
curves, Des., Codes and Cryptogr., vol. 89, pp. 965-972, 2021.

20
[2] A. Allen, K. Blackwell, O. Fiol, R. Kshirsagar, B. Matsick, G. L.
Matthews and Z. Nelson, Twisted Hermitian codes, Mathematics Spe-
cial Issue” Algebra and its Application, vol. 9, no. 1, 2021.

[3] S. Ball, On large subsets of a finite vector space in which every subset
of basis size is a basis, J. EMS, vol. 14, pp. 733-748, 2012.

[4] S. Ball, Some constructions of quantum MDS codes, Des., Codes and
Cryptogra. vol. 89, pp. 811-821, 2021.

[5] S. Ball and R. Vilar, Determining when a truncated generalised Reed-


Solomon code is Hermitian self-orthogonal, IEEE Trans. Inf. Theory,
vol. 68, pp. 3796-3805, 2022.

[6] S. Ball and R. Vilar, The geometry of Hermitian orthogonal codes, Jour-
nal of Geometry, vol. 113, artical no. 7, 2022.

[7] E. Barelli, P. Beelen, M. Datta, V. Neiger and J. Rosenkilde, Two-point


codes for the generalized GK cuvres, IEEE Trans. Inf. Theory, vol. 64,
pp. 6268-6276, 2018.

[8] D. Bartoli, M. Giulietti and I. Platoni, On the covering radius of MDS


codes, IEEE Trans. Inf. Theory, vol. 61, no. 2, pp. 801-811, 2015.

[9] A. Bassa and H. Stichtenoth, Self-dual codes better than the Gilbert-
Varshamov bound, Des., Codes and Cryptogr., vol. 87, pp. 173-182,
2019.

[10] P. Beelen, S. Puchsinger and J. Nielsen, Twisted Reed-Solomon codes,


Proc Int. Symp. Inf. Theory, 2017.

[11] P. Beelen and M. Montanucci, A new family of maximal curves, J. Lond.


Math. Soc., vol. 18, pp. 573-592, 2018.

[12] P. Beelen, S. Puchinger and J. Rosenkilde, Twisted Reed-Solomon codes,


IEEE Trans. Inf. Theory, vol. 68, no. 5, pp. 3047-3061, 2022.

[13] M. Ben-Or, S. Goldwasser and A. Wigderson, Completeness theorems


for non-cryptographic fault-tolerant distributed computation, Proceed-
ings of STOC 1988, ACM Press, pp.1–10, 1988.

21
[14] M. A. de Boer, MDS codes from hyperelliptic curves, Proc. Arithmetic,
Geometry and Coding Theory, pp. 23-34, 1996.

[15] T. Brun, I. Devetak and Min-Hsiu Hsieh, Correcting quantum errors


with entanglemnent, Science, vol. 304 (5798), pp. 436-439, 2006.

[16] V. R. Cadambe, S. A. Jafar and H. Maleki, Asymptotic interference


alignment for exact repair in distributed storage systems, 2010 Con-
ference Record of the Forty Fourth Asilomar Conference on Signals,
Systems and Computers, 2010.

[17] B. Csajbok, G. Marino, O. Polverino and C. Zauelle, A new family of


MRD-codes, Linear Alegbrai and Appl., vol. 548, pp. 203-220, 2018.

[18] D. Chaum, C. Crépeau and I. Damgaard, Multi-party unconditionally


secure protocols, Proceedings STOC 1988, ACM Press, pp. 11–19, 1988.

[19] H. Chen, On the main conjecture of geometric MDS codes, International


Mathematics Research Notices, no. 8, pp. 313-318, 1994.

[20] H. Chen and R. Cramer, Algebraic geometric secret sharing schemes and
secure multiparty computation over small fields, Crypto 2006, LNCS
4117, pp. 521-536, 2006.

[21] H. Chen, On hull-variation problem of equivalent linear codes, IEEE


Trans. Inf. Theory vol. 69, no. 5, pp. 2911-2922, 2023.

[22] J. H. Conway, V. Pless and N. J. A. Sloane, Self-dual codes over GF (3)


and GF (4) of length not exceeding 16, IEEE Trans. Inf. Theory, vol. 25,
no. 3, pp. 312-322, 1979.

[23] J. H. Conway and N. J. A. Sloane, A new upper bound on the minimal


distance of self-dual codes. IEEE Trans. Inf. Theory vol. 36, pp. 1319-
1333, 1990.

[24] M. Deuring, Die Typen der Multiplikatorenringe elliptischer Funktio-


nenkorper, Abh. Math. Sem. Hansischen Univ., Vol. 14, pp. 197-272,
1941.

[25] S. Dodubekov and I. Landjev, On near MDS codes, J. Geometry, vol.


54, pp. 30-43, 1995.

22
[26] X. Fang, M. Liu and J. Luo, New MDS Euclidean self-orthogonal codes,
IEEE Trans. Inf. Theory, vol. 67, no. 1, pp. 130-137, 2021.

[27] E. M. Gabidulin, Theory of codes with maximum rank distances, Prob-


lems of Information Transmission, vol. 21, pp. 1-21, 1985.

[28] Y. Gao, Q. Yue, X. Huang and J. Zeng, Hulls of generalized Reed-


Solomon codes via Goppa codes and their applications to quamtum
codes, IEEE Trans. Inf. Theory, vol. 67, no. 10, pp. 6619-6626, 2021.

[29] S. R. Ghorpade and G. Lachaud, Hyperplane sections of Grassmannians


and number of MDS linear codes, Finite Fields Appl., vol. 7, pp. 468-506,
2001.

[30] G. Guo and R. Li, Hermitian self-dual GRS and entended GRS codes,
IEEE Commun. Lett., vol. 25, no. 4, pp. 1062-1065, 2021.

[31] T. A. Gulliver, J-L. Kim and Y. Lee, New MDS or near MDS codes,
IEEE Trans. Inf. Theory, vol. 54, no. 9, pp. 4354-4360, 2008.

[32] V. Guruswami, A. Rudra and M. Sudan, Essential coding the-


ory, https://cse.buffalo.edu/faculty/atri/courses/coding-theory/book/,
Jan., 2021.

[33] D. Han and Y. Ren, A tight upper bound for the maximal length of
MDS elliptic codes, IEEE Trans. Inf. Theory, vol. 69, no. 2, pp. 819-822,
2023.

[34] X. He, L. Xu and H. Chen, New q-ary quamtum MDS codes with dis-
tances bigger than q2 , Quantum Inf. Process., vol. 15, pp. 2745-2758,
2016.

[35] J. W. Hirschfeld, G. Korchmáros and F. Torres, Algebraic curves over


finite fields, Princeton University Press, Princeton, Oxford, 2008.

[36] D. Huang, Q. Yue, Y. Niu and X. Li, MDS or NMDS self-dual codes from
twisted generalized Reed-Solomon codes, Des., Codes and Cryptogra.,
vol. 89, pp. 2195-2209, 2022.

[37] W. C. Huffman and V. Pless, Fundamentals of error-correcting codes,


Cambridge University Press, Cambridge, U. K., 2003.

23
[38] L. Jin and C. Xing, New MDS self-dual codes from generalized Reed-
Solomon codes, IEEE Trans. Inf. Theory, vol. 63, no. 3, pp. 14-34-1438,
2017.
[39] L. Jin and H. Kan, Self-dual near MDS codes from elliptic curves, IEEE
Trans. Inf. Theory, vol. 65, no. 4, pp. 2166-2170, 2019.
[40] X. Kai and S. Zhu, New quantum MDS codes from negacyclic codes,
IEEE Trans. Inf. Theory, vol. 59, no. 2, pp. 1193-1197, 2013.
[41] K. V. Kaipa, An asymptotic formula in q for the number of [n, k] q-ary
MDS codes, IEEE Trans. Inf. Theory, Vol. 60, no. 11, pp. 7047-7057,
2014.
[42] J. I. Kokkala, D. Krotov and P. R. J. Österga̋rd, On the classification
of MDS codes, IEEE Trans. Inf. Theory, Vol. 61, pp. 6485-6492, 2015.
[43] L. Landi and L. Vicino, Two-ponit AG codes from Beelen-Montanucci
cuvres, Finite Fields Appl., vol. 50, 102009, 2020.
[44] I. Landjev and A. Rousseva, The main conjcture for near MDS code,
WCC 2015, in Proc. 9th Int. Workshop on Coding and Cryptography,
A. Canteaut, G. Leurent, M. Naya-Plasencia, Apr 2015, Paris, France,
hal-01276222, 2015.
[45] H. Liu and S. Liu, Construction of MDS twisted Reed-Solomon codes
and LCD MDS codes, Des., Codes and Cryptogra., vol. 89, pp. 2051-
2065, 2021.
[46] G. Lunardon, R. Trombetti and Y. Zhou, Generalized twisted Gabidulin
codes, J. Combinat. Theory A, vol. 159, pp. 79-106, 2018.
[47] G. Luo, X. Cao and X. Chen, MDS codes with hulls of arbitray dimen-
sions and their quantum error correction, IEEE Trans. Inf. Theory, vol.
65, no. 5, pp. 2944-2952, 2019.
[48] F. J. MacWilliams and N. J. A. Sloane, The Theory of error-correcting
codes, 3rd Edition, North-Holland Mathematical Library, vol. 16. North-
Holland, Amsterdam, 1977.
[49] R. J. McEliece and D. V. Sarvate, On sharing secrets and Reed-Solomon
codes. Commun. ACM, vol 24, no. 11, pp. 583-584, 1981.

24
[50] C. Munuera, On main conjecture on MDS geometric codes, IEEE Trans.
Inf. Theory, vol. 38, no. 5, 1992.

[51] C. Munuera, On elliptic MDS codes, Discrete Math., vol. 117, pp. 279-
286, 1993.

[52] A. Neri, S. Puchsinger, A.-L. Horlemann-Trautmann, Equivalence and


characterization of linear rank-metric codes based on invariants, Linear
Alg.Appl., vol. 603, no. 15, pp. 418-469, 2020.

[53] Y. Niu, Q. Yue, Y. Wu and L. Hu, Hermitian self-dual, MDS and gen-
eralized Reed-Solomon codes, IEEE Commun. Lett., vol. 23, no. 5, pp.
781-784, 2019.

[54] K. Otal and F. Özbudak, Additive rank-metric codes, IEEE Trans. Inf.
Theory, vol. 63, no. 1, pp. 164-168, 2017.

[55] I. Reed and G. Solomon, Polynomial codes over certain finite fields, J.
SIAM, vol. 8, pp. 300-304, 1960.

[56] R. M. Roth and A. Lempel, A construction of non-Reed-Solomon type


MDS codes, IEEE Trans. Inf. Theory, vol. 35, no. 3, pp. 655-657, 1989.

[57] H-G. Rück, A note on elliptic curves over finite fields, Math. Comput.,
Vol. 49, No. 179, pp. 301-304.

[58] B. Segre, Curve raizionali normali e k-archi negli spazi finiti, Ann. Mat.
Pura Appl., Vol. 39, pp. 357-378, 1955.

[59] A. Shamir, How to share a secret, Commun. ACM, vol. 22, no. 11, pp.
612-613, 1979.

[60] J. Sheekey, A new family of linear maximum rank distance code, Adv
Math. Commun., vol. 10, no. 3, pp. 475-488, 2016.

[61] P. W. Shor, Scheme for redcuing decoherence in quantum memory, Phys.


Rev. A, vol. 52, pp. R2493-2496, 1995.

[62] J. H. Silverman, The arithemtic of ellptic curves, Chapter V, 2nd Edi-


tion, Springer Dordrecht Heidelberg London New York, 2009

25
[63] A. M. Steane, Mutiple particle interference and quantum error correc-
tion, Proc. Roy. Soc. London, vol. 452, pp. 2551-2577, 1996.

[64] J. Sui, Q. Yue, X. Li and D. Huang, MDS, near MDS or 2-MDS self-dual
codes via twisted Reed-Solomon code, IEEE Trans. Inf. Theory, vol. 68,
no. 12, pp. 7832-7841, 2022.

[65] R. Trombetti and Y. Zhou, A new family of MRD codes in Fq2n×2n with
right and middle nuclei Fnq , IEEE Trans. Inf. Theory, vol. 65, no. 2, pp.
1054C1062, Feb. 2019.

[66] M. Tsfasman and S. G. Vlǎdut, Algebraic-geometric codes, Vol.58,


Springer Science and Business Media, 2013.

[67] J. L. Walker, A new approach to the main conjcture of algebraic-


geometric MDS codes, Des., Codes and Cryptogr., vol. 9, pp. 115-120,
1996.

[68] A. Zhang and K. Feng, A unified approch to construct MDS self-dual


codes via Reed-Soloon code, IEEE Trans. Inf. Theory, vol. 66, no. 6, pp.
3650-3656, 2020.

26

You might also like