A^sMDS Codes: Controlled Defect in MDS Theory

• A^sMDS codes are linear codes defined by a controlled defect (s) from the Singleton bound, generalizing traditional MDS codes.
• Geometric methods and construction techniques such as generalized Reed–Solomon and subgroup coset designs control parameters like hull dimension and code length.
• These codes are pivotal for quantum error-correction, symbol-pair code design, and optimal repair in distributed storage, with ongoing research on extremal parameters.

A$^s$MDS codes are families of linear codes defined by achieving a specific, controlled defect $s$ with respect to the Singleton bound—that is, their minimum distance $d$ satisfies $d = n-k+1-s$ for some $s \geq 0$, where $n$ is the block length, $k$ is the dimension, and $q$ is the size of the finite field. The classical MDS (Maximum Distance Separable) codes correspond to $s=0$; $A^1$MDS and $A^s$MDS codes for $s>0$ generalize this, facilitating systematic study of codes of prescribed defect, their extremal parameters, geometric representations, and quantum extensions.

1. Definitions and the Singleton Defect

Given a linear $[n,k,d]_q$ code $C$ over the finite field $\mathbb{F}_q$, the Singleton bound asserts $d \leq n-k+1$. The Singleton defect $S(C)$ is defined as $S(C) = n-k+1-d$.

• An MDS code is characterized by $S(C)=0$.
• A code is $A^s$MDS if $S(C)=s>0$, i.e., $d = n-k+1-s$.

In projective geometry language, $A^s$MDS codes are encoded as projective systems of "defect $s$," and they can be analyzed through geometric approaches concerning the arrangement of points and hyperplanes in projective space (Alderson et al., 27 Apr 2025).

2. Geometric and Projective System Perspective

Every linear code's generator matrix yields, up to column scaling and permutation, a multiset $G$ of $n$ points (possibly with multiplicities) in projective space $\Sigma = \mathrm{PG}(k-1,q)$. Key geometric correspondences are:

• The minimum distance $d$ is determined by the maximal number of $G$ lying in a hyperplane: $n-d = \max_{H} |G \cap H|$.
• The dual code's minimum distance $d^\perp$ corresponds to the smallest support $Q\subset G$ with $|Q|-\dim\langle Q\rangle=1$.

Codes with no zero coordinates (non-degenerate) correspond to projective systems that avoid a hyperplane. Degeneracy in $C$ or its dual is precluded to study maximal extremal lengths.

3. Parameters and Extremal Quantities

Three pivotal parameters are introduced to chart $A^s$MDS code families (Alderson et al., 27 Apr 2025):

Parameter	Definition
$m^s(k,q)$	Max length $n$ of non-degenerate $[n,k,d]_q$ $A^s$MDS code
$m^s_t(k,q)$	Max length $n$ s.t. both $C$ is $A^s$MDS and $C^\perp$ is $A^t$MDS
$\kappa(s,q)$	Largest $k$ for $n=(s+1)(q+1)+k-2$ length-maximal $A^s$MDS code

Boundaries for these parameters are derived through hyperplane-counting and quotient-shortening (deletion-projection) techniques.

4. Length and Dimension Bounds

Extensive upper and lower bounds for $A^s$MDS codes are established via geometric, combinatorial, and arithmetic techniques (Alderson et al., 27 Apr 2025):

• General Upper Bound: $n \leq (s+1)(q+1) + k - 2$ for non-degenerate $A^s$MDS codes.
• Refined Bound (for $0< s$(s+2,q)\neq(2^r,2^u)$): $n \leq (s+1)(q+1) + k - 4$.
• Planar Arc-Based Bound: For suitable $q,s$, $n \leq q(s+1) + k - 2$.
• Quotient-Shortening Lower Bounds: For $k=2$, $m^s(2,q) = (s+1)(q+1)$; more generally $m^s(k,q)\geq k+s$.
• Dual Defect Constraints: $m^s_t(k,q) \leq s(q+1)+k-1$ when $t>1$.

Length-maximal examples and cap-theoretic constructions are fully explicit only for small $k$ or maximal $s$ ($s=q-1$), and bounds quickly become strict as $k$ increases.

5. Duality and Self-Defect Conditions

For a code $C$, duality properties of the defect are subtle:

• Dually-$A^s$MDS (or NMDS) codes have both $C$ and $C^\perp$ with defect $s$.
• Classical MDS codes are always dually-MDS, but for $s>0$ the property can fail.

A sufficient criterion for dual self-defect is:

• If $1 < s < q-1$, $(s+1,q)\neq(2^e,2^h)$, $k > (s-1)(q+1)-1$, and $n > s(q+1)+k-3$, then $C^\perp$ is also $A^s$MDS [(Alderson et al., 27 Apr 2025), Theorem 5.5].

Quotient techniques are used to reduce dimensionality and enforce projectivity constraints, leading to tight requirements on $n$ and $k$.

6. Construction Techniques and Explicit Families

Infinite families of $A^s$MDS codes are constructed as follows (Luo et al., 2018):

• Generalized Reed–Solomon (GRS) Approach: For $q=2^m$, $1 < n \leq q$, and $1 \leq s \leq k \leq n$, appropriate choices of evaluation points and scalar multipliers $v_i$ yield $[n,k]$ MDS codes with $\dim \text{Hull}(C)=k-s$.
• Odd $q$, extended GRS constructions: When $q>3$, codes of length $q+1$ and dimension $k \leq(q+1)/2$ can have $\dim \text{Hull}(C)$ freely assigned over the allowable range.
• Subgroup coset and additive constructions: Specific group-theoretic selections of support and multipliers force the hull dimension to a prescribed value.

Summary of families:

Field/Construction	Parametric Family	Hull Dimension / Defect Control
Even $q$, $1 < n \leq q$	$[n,k]$ MDS, any $s$	$\dim \text{Hull}(C)=k-s$
Odd $q>3$, $n=q+1$	$[q+1,k]$ MDS, $k\le(q+1)/2$	$\dim \text{Hull}(C)=k-1-s$
Multiplicative subgroup	$[n,k]$ MDS, $n|q-1$, $k\le n/2$	$\dim \text{Hull}(C)=k-1-s$

These constructions enable the realization of MDS codes with arbitrary hull dimension, an essential ingredient for quantum error-correction applications.

7. Applications and Extensions

$A^s$MDS codes play a vital role in:

• Quantum codes: By using hull-dimension–tunable $A^s$MDS codes, entanglement-assisted quantum error-correcting codes (EAQECCs) are constructed with parameters $[[n, k-s, d=n-k+1; c=n-k-s]]_q$, with the entanglement cost $c$ flexibly chosen in the permitted range (Luo et al., 2018).
• Symbol-pair codes: $A^s$MDS principles underpin the design of AMDS (Almost MDS) symbol-pair codes, where the symbol-pair Singleton-type bound is $d_p \leq n - k + 2$, and explicit families with $d_p=n-k+1$ are constructed using repeated-root cyclic codes (Ma, 2022).
• Distributed storage: In exact repair for storage, asymptotically optimal repair bandwidth MDS codes (including A$^s$-families in the technical sense of (Chowdhury et al., 2017)) exploit the structure of the defect and the partitioning/subpacketization schemes that these codes admit.

Boundaries of existence, duality, and extremality remain open for many parameters, especially the maximal dimensions $\kappa(s,q)$ attained by length-maximal $A^s$MDS codes. Conjectures state that typically $\kappa(s,q)\leq4$, unless $q$ is even and $(s+2)$ divides $q$, or $s\in\{0,q-2\}$; these cases may admit larger $k$.

8. Open Problems and Extremal Results

Current research directions involve:

• Sharp classification of possible parameters for length-maximal $A^s$MDS codes;
• Existence and construction of extremal dually-$A^s$MDS codes beyond the known ranges;
• Combinatorial and arithmetic constraints ensuring the non-existence of codes in certain parameter domains (e.g., integrality of specific binomial quotients for length-maximal codes);
• Extensions to multivariate or algebraic-geometric code settings aimed at preserving high defect with additional structure, especially relevant for quantum and pair-metric coding contexts (Luo et al., 2018, Ma, 2022).

These inquiries underscore the foundational role of $A^s$MDS codes within both classical and quantum coding theory, revealing deep geometric and algebraic interconnections that inform code design and theory.