Generalized Labeled Codes

• Generalized labeled codes are code families where each coordinate is endowed with algebraic, combinatorial, or geometric labels that determine local constraints.
• They incorporate diverse frameworks—ranging from surface singularity codes and GHFP-codes to graph-indexed and quantum stabilizer codes—to address complex decoding and recovery needs.
• Labels in these codes influence weight structure, deformation behavior, and recovery schemes, unifying methods across algebra, geometry, and combinatorics.

Searching arXiv for recent and foundational uses of “generalized labeled codes” and adjacent terminology. “Generalized labeled codes” does not denote a single universally standardized code family across coding theory. In the literature, the phrase and closely adjacent constructions refer to several distinct frameworks in which coordinates, symbols, or local constraints carry additional structure beyond the uniform coordinate model of a classical linear code. These include codes labeled by local singularity data on algebraic surfaces (Catanese, 22 Aug 2025), group-labeled nonlinear codes arising from generalized Hadamard matrices and cocycles (Armario et al., 2019), graph-indexed vertex labelings on Johnson and Hamming graphs (Duursma et al., 2019), group-algebra-labeled stabilizer codes generalizing Haah’s cubic code (Tian et al., 2019), and block-structured generalized algebraic-geometric evaluation codes built from higher-degree points and inner codes (Calderini et al., 2012). A common theme is that labels are not merely decorative metadata: they determine admissible local symbols, symmetry actions, recovery neighborhoods, or the algebraic form of parity constraints.

1. Terminological scope and principal meanings

In the most explicit recent usage, a generalized labeled code is a coding-theoretic object attached to a normal surface with isolated singularities, where the ambient coordinate group is a direct sum of dual local homology groups labeled by singularity type (Catanese, 22 Aug 2025). This replaces the classical binary nodal code, which lives in $(\mathbf Z/2)^s$, by an embedding

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$

with each summand determined by the local singularity at $x$ (Catanese, 22 Aug 2025).

A different, older line of work studies group-labeled nonlinear codes derived from generalized Hadamard matrices over the additive group of a finite field. There, labels are entries of a generalized Hadamard matrix, codewords are row-labelings up to addition of constant vectors, and the code structure is governed simultaneously by additive field structure, cocycles, and extension-group multiplication (Armario et al., 2019). The associated family is called generalized Hadamard full propelinear codes, or GHFP-codes (Armario et al., 2019).

Another meaning appears in graph-based coding. Johnson graph codes are linear subspaces of labelings of the vertices of a Johnson graph such that the labeling is uniquely determined by its restriction to a graph-defined generalized neighborhood $B_r(A)$ (Duursma et al., 2019). The coordinates are thus indexed by $v$-subsets of an $n$-set rather than by consecutive positions, and information sets are determined by shell structure in the graph (Duursma et al., 2019).

A further neighboring usage occurs in fracton stabilizer codes. The generalized Haah construction specifies a code by a finite group $G$, an integer $q$, two $q$-tuples $A=(A_1,\dots,A_q)$ and $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$0 of subsets of $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$1, and commuting $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$2 matrices over $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$3; these data label the supports of stabilizer generators on a $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$4-qubit-per-site lattice $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$5 (Tian et al., 2019). In that setting, the code is “labeled” by subset and matrix data rather than by a uniform translational template (Tian et al., 2019).

A plausible implication is that “generalized labeled codes” is best treated as an umbrella description for code families in which the ambient coordinates are indexed by algebraic, combinatorial, or geometric labels carrying operational significance, rather than as the name of a single canonical construction.

2. Geometric generalized labeled codes from normal and ADE surfaces

The most direct formal definition is given for normal surfaces. Let $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$6 be a finite set of labeled finitely generated abelian groups, let $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$7 be a finite set, let $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$8 assign to each $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$9 a local group $x$0, and let

$x$1

be a surjection. Dualizing yields

$x$2

and $x$3 is the generalized labeled code (Catanese, 22 Aug 2025).

This formalism is motivated by the topology of a normal compact complex surface $x$4 with isolated singularities. If $x$5, there is a natural local-to-global map

$x$6

and under the stated global hypotheses this map is surjective (Catanese, 22 Aug 2025). The strict generalized labeled code is the dual image

$x$7

for a projective normal surface $x$8 (Catanese, 22 Aug 2025). The extended code is

$x$9

where $B_r(A)$0 is a smooth hyperplane section (Catanese, 22 Aug 2025).

For ADE singularities, the local groups are the discriminant groups of the corresponding root lattices. The paper lists

$B_r(A)$1

as the local homology groups $B_r(A)$2 (Catanese, 22 Aug 2025). The geometric labeling is finer than the abstract isomorphism class of the group: distinguished generators are determined by the exceptional curves in the minimal resolution, and this affects refined weight counts (Catanese, 22 Aug 2025).

The extended code also measures lattice saturation. If $B_r(A)$3 is the minimal resolution and $B_r(A)$4, then

$B_r(A)$5

(Catanese, 22 Aug 2025). This ties the generalized labeled code directly to the non-primitivity of the lattice generated by exceptional divisors and polarization.

3. Weight structures, shortening, and ancestors in the geometric theory

The geometric theory requires a refined notion of weight because the local coordinate groups are heterogeneous. For

$B_r(A)$6

the paper defines the refined weight

$B_r(A)$7

the label weight

$B_r(A)$8

and the Hamming weight

$B_r(A)$9

(Catanese, 22 Aug 2025). This separates support size from local character type.

Several congruence restrictions are then proved. For order $v$0 vectors, a weighted sum involving contributions from $v$1, $v$2, and $v$3 is divisible by $v$4, and by $v$5 if $v$6 is divisible by $v$7 (Catanese, 22 Aug 2025). For order $v$8,

$v$9

is divisible by $n$0 (Catanese, 22 Aug 2025). For almost simple vectors of order $n$1,

$n$2

is divisible by $n$3 (Catanese, 22 Aug 2025). These are direct analogues of classical divisibility restrictions for binary nodal codes, but now expressed in label-sensitive form.

A central innovation is the extension of code shortening. An elementary shortening replaces one local group $n$4 by a quotient $n$5 together with a refined labeling of the replacement local pieces, and simultaneously quotients the global group by the image of the discarded part (Catanese, 22 Aug 2025). Dually,

$n$6

for the shortened ambient group (Catanese, 22 Aug 2025). In geometric terms, shortening is the algebraic counterpart of partial smoothing of singularities.

The model example is $n$7, where deleting one exceptional curve in the Dynkin chain corresponds to deforming

$n$8

and induces

$n$9

at the local-group level (Catanese, 22 Aug 2025). The paper’s realization theorem states that, for unobstructed ADE surfaces, geometric driven shortenings are exactly realized by partial smoothings, and conversely every partial smoothing yields a geometric driven shortening (Catanese, 22 Aug 2025).

This leads to the notion of an ancestor: a maximally singular model from which others arise by shortening or smoothing (Catanese, 22 Aug 2025). The paper identifies explicit ancestors among cubic and K3 surfaces, including the Cayley cubic, a cubic with $G$0 singularities, a cubic with $G$1, a cubic with $G$2, the Kummer surface, and the nine-cusp K3 surface (Catanese, 22 Aug 2025). This suggests a deformation-theoretic genealogy organized by generalized labeled codes.

4. Group-labeled nonlinear codes from generalized Hadamard matrices

In the generalized Hadamard framework, labels are elements of the additive group of $G$3, viewed multiplicatively when needed for cocycles and extensions (Armario et al., 2019). A normalized generalized Hadamard matrix $G$4 over $G$5 defines the row set $G$6 and the generalized Hadamard code

$G$7

with

$G$8

(Armario et al., 2019). These codes are typically nonlinear (Armario et al., 2019).

The algebraic backbone is an orthogonal cocycle

$G$9

satisfying the cocycle identity and the orthogonality condition that the cocyclic matrix $q$0 is a generalized Hadamard matrix (Armario et al., 2019). Such a cocycle determines a central extension

$q$1

and the paper proves the equivalence between orthogonal cocycles, cocyclic generalized Hadamard matrices, and central relative $q$2-difference sets (Armario et al., 2019).

The propelinear structure is then introduced. A generalized Hadamard code that is full propelinear is called a generalized Hadamard full propelinear code, or GHFP-code (Armario et al., 2019). In this setting, symbol actions are translations by the label vector, while nonconstant codewords induce fixed-point-free coordinate permutations (Armario et al., 2019). The group of coordinate permutations $q$3 satisfies

$q$4

(Armario et al., 2019).

The paper proves that GHFP-codes are equivalent to cocyclic generalized Hadamard matrices and to central relative difference sets (Armario et al., 2019). This makes the labeled nature explicit: codewords correspond to extension-group elements, coordinate permutations come from the regular action of $q$5, and labels combine through cocycle multiplication. A plausible implication is that GHFP-codes provide one of the clearest algebraic models of “generalized labeled codes” in the nonlinear finite-alphabet setting.

5. Graph-indexed labelings and group-algebra-labeled stabilizer codes

Johnson graph codes shift the notion of a code coordinate from positions in $q$6 to vertices of a graph. For the Johnson graph $q$7, coordinates are indexed by all $q$8-subsets of an $q$9-set, so the length is

$q$0

(Duursma et al., 2019). For a $q$1-subset $q$2, the generalized neighborhood $q$3 is a union of vertex neighborhoods, and a Johnson graph code $q$4 is defined so that every $q$5 is an information set (Duursma et al., 2019). The dimension is therefore

$q$6

(Duursma et al., 2019). The neighborhood size is given explicitly by

$q$7

(Duursma et al., 2019).

The main construction starts from a length-$q$8 MDS code $q$9 and the determinantal coordinate map

$A=(A_1,\dots,A_q)$0

for $A=(A_1,\dots,A_q)$1 matrices $A=(A_1,\dots,A_q)$2 having at least $A=(A_1,\dots,A_q)$3 rows in $A=(A_1,\dots,A_q)$4 (Duursma et al., 2019). The resulting code is a Johnson graph code with

$A=(A_1,\dots,A_q)$5

(Duursma et al., 2019). The dual is again a Johnson graph code: $A=(A_1,\dots,A_q)$6 (Duursma et al., 2019). In distributed storage, these codes serve as outer constraints linking layered exact-repair constructions and enabling recovery from $A=(A_1,\dots,A_q)$7 nodes while preserving layered repair structure (Duursma et al., 2019).

Generalized Haah codes provide a different graph- or group-labeled perspective. Here the lattice is replaced by a finite group $A=(A_1,\dots,A_q)$8, the Hilbert space is

$A=(A_1,\dots,A_q)$9

and the code is labeled by $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$00, $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$01, and commuting $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$02 matrices $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$03 over $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$04 (Tian et al., 2019). The stabilizers are

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$05

(Tian et al., 2019). Pairwise-commuting matrices guarantee commutation of the stabilizer family (Tian et al., 2019). This construction recovers the Haah A-code for $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$06 and the Haah B-code for $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$07 with a specific symmetric mixing matrix (Tian et al., 2019).

These two lines of work share an essential feature: labels determine recovery or commutation structure through combinatorial neighborhoods or translated supports, not just through scalar parity-check coefficients.

6. Block-structured generalized evaluation codes and adjacent generalized families

Generalized algebraic-geometric evaluation codes provide another labeled framework, now with blocks indexed by evaluation points of varying degrees. For points $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$08 of degrees $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$09 on a curve, and $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$10-linear isomorphisms

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$11

onto inner codes $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$12, the generalized AG code is

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$13

with

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$14

(Calderini et al., 2012). The code length is

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$15

(Calderini et al., 2012). In the extended affine-variety and order-domain formulation, the same block-structured construction yields minimum-distance bounds

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$16

with $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$17 (Calderini et al., 2012). In this line, the “label” of a coordinate block is the residue-field degree and the chosen inner code.

Related generalized code notions from the literature show the breadth of labeled or heterogeneous coordinate structures. Generalized concatenated codes partition information into several outer components over possibly different extension fields, all fed into one inner code (Blomqvist et al., 2020). Generalized quasi-cyclic codes are module codes over a product of quotient rings

$$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$18

and decompose into constituent codes over mixed alphabets through the Chinese remainder theorem (Güneri et al., 2017). Generalized product and extended product codes arrange coordinates in arrays with local row constraints, global column constraints, and extra parity layers indexed by protection level (Blaum et al., 2016). These constructions are not called generalized labeled codes in their source papers, but they instantiate the same structural principle: code coordinates belong to heterogeneous blocks, layers, or local alphabets, and those labels determine admissible decoding operations and distance behavior.

A final observation is that the term also appears in modern extremal and algebraic coding contexts only tangentially. For example, generalized constacyclic, twisted Reed–Solomon, and LCD constructions study generalized evaluation or extension mechanisms (Liu et al., 2022, Zhao et al., 23 Jan 2026, Galindo et al., 2017), but the “label” is not the central organizing concept there. This suggests that the most faithful uses of the phrase are the geometric, group-labeled, graph-indexed, and stabilizer-labeled theories.

7. Conceptual unification and recurring structural themes

Across these disparate settings, several recurrent ideas emerge. First, the coordinate space is rarely a uniform $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$19 with anonymous positions. Instead it is a direct sum of local character groups (Catanese, 22 Aug 2025), a union of cosets of row-labelings from a generalized Hadamard matrix (Armario et al., 2019), a space of graph-vertex labelings (Duursma et al., 2019), a stabilizer system determined by group-algebra subsets (Tian et al., 2019), or a product of inner-code blocks attached to higher-degree evaluation points (Calderini et al., 2012).

Second, labels control reconstruction. In the surface theory, the local labels determine the refined weights, shortening rules, and deformation behavior (Catanese, 22 Aug 2025). In GHFP-codes, labels determine the propelinear group law and the coordinate permutation action (Armario et al., 2019). In Johnson graph codes, a neighborhood $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$20 is an information set by definition (Duursma et al., 2019). In generalized Haah codes, labels specify stabilizer supports and commutation relations (Tian et al., 2019). In generalized AG evaluation codes, labels determine block length and local distance through the chosen inner code $$\mathcal K \subset \bigoplus_{x\in \Sigma} H_1(x)^\vee,$$21 (Calderini et al., 2012).

Third, many of these theories include a notion of inheritance or reduction. Geometric shortenings correspond to partial smoothings (Catanese, 22 Aug 2025). Johnson graph codes have explicit dual and complement symmetries (Duursma et al., 2019). GHFP-codes are equivalent to cocyclic generalized Hadamard matrices and relative difference sets (Armario et al., 2019). Generalized AG evaluation codes recover one-point GAG codes exactly in the order-domain setting (Calderini et al., 2012). This suggests that generalized labeled codes are often best understood not as isolated constructions but as organizing frameworks connecting algebra, geometry, and combinatorics.

A plausible implication is that the phrase “generalized labeled codes” is most useful when it denotes a structural viewpoint: a code family in which coordinates carry algebraic or combinatorial labels that affect local constraints, recovery sets, symmetry, or deformation. Within that viewpoint, the geometric ADE theory (Catanese, 22 Aug 2025) is the most explicit formalization, while GHFP-codes (Armario et al., 2019), Johnson graph codes (Duursma et al., 2019), generalized Haah codes (Tian et al., 2019), and generalized AG evaluation codes (Calderini et al., 2012) provide complementary realizations in nonlinear coding, graph coding, stabilizer coding, and algebraic-geometric coding.