One-Dimensional Affine Association Schemes

• One-dimensional affine association schemes are translation schemes on groups like (𝔽₍q₎, +) or ℤₘ, defined by additive difference relations.
• They serve as foundational cases for higher-dimensional constructions, with eigenstructures governed by affine q-Krawtchouk and dual q-Hahn polynomials.
• These schemes integrate affine geometry, coding theory, and combinatorial designs through translation invariance and attenuated-space formulations.

Searching arXiv for the cited papers and closely related work on association schemes, affine schemes, and attenuated spaces. One-dimensional affine association schemes are most naturally understood as translation schemes on an abelian group identified with an affine line, typically $(\mathbb{F}_q,+)$ or $\mathbb{Z}_m$, with relations determined by additive differences. In the finite-geometric literature they also appear as the $m$-flat association schemes with $m=1$, and hence as a special case of association schemes based on attenuated spaces. These schemes occupy a central position between affine geometry, translation-invariant Bose–Mesner algebras, orthogonal-polynomial eigenstructures, and coding theory; they also serve as base objects for higher-dimensional constructions such as generalized Hamming schemes and, at the opposite extreme, illuminate the small-valency boundary cases isolated in quasi-thin classification results (Godsil, 2010, Kurihara, 2011, Muzychuk et al., 2010).

1. Definitions and scope

An association scheme on a finite set $X$ is a partition

$X\times X=\bigsqcup_{i=0}^d R_i$

such that $R_0$ is the identity relation, the relations are closed under transpose, and for all $i,j,k$ there exist intersection numbers $p_{ij}^k$ for which the number of $z\in X$ satisfying $\mathbb{Z}_m$0 and $\mathbb{Z}_m$1 depends only on the relation $\mathbb{Z}_m$2 containing $\mathbb{Z}_m$3. Equivalently, the adjacency matrices $\mathbb{Z}_m$4 span a Bose–Mesner algebra closed under both ordinary matrix multiplication and Schur product (Muzychuk et al., 2010, Kodalen, 2019).

In the standard language used for affine schemes, a translation scheme is an association scheme whose vertex set is an abelian group $\mathbb{Z}_m$5 and whose relations are invariant under translations by elements of $\mathbb{Z}_m$6. Usually one identifies the vertex set with $\mathbb{Z}_m$7 itself and defines relations by a partition $\mathbb{Z}_m$8 with

$\mathbb{Z}_m$9

Within this framework, a one-dimensional affine association scheme is the case where $m$0 is viewed as a one-dimensional affine object, most commonly $m$1 or a finite cyclic group (Godsil, 2010).

A second, equivalent geometric realization appears in the $m$2-flat literature. There, the one-dimensional affine case is the $m$3-flat association scheme with $m$4: its vertices are affine lines, that is, cosets of $m$5-dimensional subspaces of $m$6. Kurihara’s attenuated-space formulation shows that these schemes are not isolated examples but sit inside a larger $m$7-geometric family (Sosna, 2011).

2. Translation schemes on the affine line

The translation-scheme viewpoint is the basic affine model. If $m$8 is a translation scheme relative to an abelian group $m$9, then the additive action of $m=1$0 on itself is regular, and every relation is determined entirely by a difference class. This makes the adjacency algebra amenable to Fourier analysis on $m=1$1, and it is the finite-scheme analogue of replacing Euclidean invariance by additive invariance on a finite affine line (Godsil, 2010).

This perspective also explains the duality properties that repeatedly appear in the literature. For an abelian group $m=1$2, a translation association scheme defined by the orbits of the group on $m=1$3 has a dual scheme on the character group $m=1$4, with the first and second eigenmatrices interchanged. In the cometric literature this is presented as a basic source of formal duality, and cyclic schemes are described as canonical examples (Kodalen, 2019).

The translation model is minimal in dimension but structurally rich. It already contains the single-class scheme on $m=1$5, whose $m=1$6-fold symmetric extension is the classical Hamming scheme $m=1$7, and it supplies the natural base case for more elaborate affine constructions. In this sense, one-dimensional affine association schemes are the primitive translation-scheme building blocks from which higher-dimensional affine and product-type schemes are assembled (Godsil, 2010).

3. Realization as $m=1$8-flat schemes and attenuated-space schemes

The $m=1$9-flat association scheme $X$0 is defined on the set $X$1 of all $X$2-flats of $X$3, that is, cosets $X$4 of $X$5-dimensional subspaces $X$6. Its relations are indexed by pairs $X$7, where

$X$8

according to the dimension of $X$9 and the condition $X\times X=\bigsqcup_{i=0}^d R_i$0 or $X\times X=\bigsqcup_{i=0}^d R_i$1. For $X\times X=\bigsqcup_{i=0}^d R_i$2, the vertices are affine lines, and this is exactly the one-dimensional affine association scheme in the sense used in the attenuated-space literature (Kurihara, 2011).

Kurihara proves that the $X\times X=\bigsqcup_{i=0}^d R_i$3-flat association scheme is isomorphic to an attenuated-space scheme: $X\times X=\bigsqcup_{i=0}^d R_i$4 Here $X\times X=\bigsqcup_{i=0}^d R_i$5 consists of $X\times X=\bigsqcup_{i=0}^d R_i$6-dimensional subspaces of $X\times X=\bigsqcup_{i=0}^d R_i$7 disjoint from a fixed $X\times X=\bigsqcup_{i=0}^d R_i$8-dimensional subspace. In particular, the one-dimensional affine case $X\times X=\bigsqcup_{i=0}^d R_i$9 is realized as

$R_0$0

and the paper explicitly notes that when $R_0$1, the attenuated space is just an affine space (Kurihara, 2011).

This realization places one-dimensional affine schemes next to two major families. When $R_0$2, attenuated-space schemes reduce to Grassmann schemes; when $R_0$3, they reduce to bilinear forms schemes. The one-dimensional affine case therefore appears as part of a continuum linking affine spaces, Grassmannians, and bilinear forms through the attenuated-space formalism (Kurihara, 2011, Bernard et al., 2024).

4. Eigenvalues and polynomial structure

The character tables of attenuated-space schemes are given explicitly in terms of generalized Eberlein and generalized Krawtchouk polynomials. If $R_0$4 and $R_0$5 denote the first and second eigenmatrices of $R_0$6, then Kurihara obtains

$R_0$7

and

$R_0$8

For $R_0$9, these formulas specialize to the one-dimensional affine case, so the eigenstructure is already controlled by the same polynomial objects that govern the broader attenuated family (Kurihara, 2011).

A complementary formulation is given in the recent study of attenuated-space schemes with bivariate $i,j,k$0- and $i,j,k$1-polynomial structures. There, one-dimensional affine schemes are described as the univariate limit of the bivariate theory: in the one-dimensional affine case the eigenvalues are governed by affine $i,j,k$2-Krawtchouk polynomials, which satisfy both a three-term recurrence in the degree index and a second-order $i,j,k$3-difference equation in the spectral variable. When the attenuated-space domain collapses to a line, the bivariate polynomial system reduces to the familiar univariate affine $i,j,k$4-Krawtchouk or dual $i,j,k$5-Hahn picture (Bernard et al., 2024).

This spectral description is not merely formal. It identifies one-dimensional affine schemes with the rank-one end of a larger bispectral hierarchy. In the univariate case the adjacency algebra is generated by a single distance-like operator, while in the more general attenuated-space setting the same mechanism becomes bivariate and is expressed through recurrence and difference operators associated with dual $i,j,k$6-Hahn and affine $i,j,k$7-Krawtchouk polynomials (Bernard et al., 2024).

5. Derived constructions and extensions

One-dimensional affine schemes are also the natural base objects for generalized Hamming schemes. If $i,j,k$8 is any association scheme on a vertex set $i,j,k$9, the generalized Hamming scheme $p_{ij}^k$0 is defined on $p_{ij}^k$1 by recording, for each pair $p_{ij}^k$2, the relation-type vector $p_{ij}^k$3 counting how many coordinates lie in each relation of $p_{ij}^k$4. The resulting matrices $p_{ij}^k$5 form an association scheme, and if $p_{ij}^k$6 is a translation scheme relative to an abelian group $p_{ij}^k$7, then $p_{ij}^k$8 is a translation scheme relative to $p_{ij}^k$9. In particular, a one-dimensional affine scheme on $z\in X$0 extends canonically to an $z\in X$1-dimensional affine translation scheme on $z\in X$2 (Godsil, 2010).

The eigenvalues of $z\in X$3 are encoded by generating functions built from the eigenmatrix $z\in X$4 of the base scheme, and the same formalism yields a MacWilliams theorem: $z\in X$5 For additive subsets $z\in X$6 in a translation scheme, this specializes to a generalized MacWilliams identity for the dual code $z\in X$7. Thus one-dimensional affine association schemes feed directly into a uniform coding-theoretic mechanism extending the classical Hamming situation (Godsil, 2010).

Another line of construction starts from affine resolvable designs over finite fields. In the work on twin prime powers, Fermat primes, and Mersenne primes, auxiliary matrices $z\in X$8 derived from the additive group of $z\in X$9 are incidence matrices of an affine resolvable design, and together with Latin squares from $\mathbb{Z}_m$00 they produce commutative translation association schemes. The paper states that it does not use the phrase “one-dimensional affine association schemes” explicitly, but that it constructs translation association schemes whose affine structure is fundamentally $\mathbb{Z}_m$01-dimensional over finite fields (Kharaghani et al., 2017). These examples show that affine-line data can be combined with additional field structure to generate higher-rank translation schemes with explicitly computed eigenmatrices (Kharaghani et al., 2017).

6. Quasi-thin and low-valency boundary cases

The quasi-thin classification gives a very different, low-valency perspective on affine behaviour. A quasi-thin association scheme is one in which every basic relation has valency $\mathbb{Z}_m$02 or $\mathbb{Z}_m$03. Muzychuk and Ponomarenko prove that any Kleinian quasi-thin scheme arises from exactly one of three small geometries: a near-pencil on $\mathbb{Z}_m$04 points, an affine plane of order $\mathbb{Z}_m$05, or a projective plane of order $\mathbb{Z}_m$06. In this classification, the affine plane of order $\mathbb{Z}_m$07 is the only affine object that survives the restriction to valencies at most $\mathbb{Z}_m$08 (Muzychuk et al., 2010).

The same paper shows that any non-Kleinian quasi-thin scheme is schurian and separable, whereas the exceptional non-schurian and non-separable phenomena occur only in Kleinian schemes of index $\mathbb{Z}_m$09 or $\mathbb{Z}_m$10, corresponding to the affine plane of order $\mathbb{Z}_m$11 and the projective plane of order $\mathbb{Z}_m$12. The affine plane case is therefore a boundary case: it is not a general one-dimensional affine line scheme, but it is the only genuinely affine geometry present in the complete quasi-thin classification (Muzychuk et al., 2010).

A plausible implication is that one-dimensional affine association schemes, in the ordinary translation-scheme sense, occupy a much less rigid part of the theory than quasi-thin schemes do. Once valencies are forced down to $\mathbb{Z}_m$13 and $\mathbb{Z}_m$14, affine geometry survives only in the degenerate small-order form of the affine plane of order $\mathbb{Z}_m$15; outside that regime, affine-line schemes are better understood through translation invariance, attenuated spaces, and coding-theoretic constructions than through quasi-thin structure (Muzychuk et al., 2010).

7. Group-theoretic and categorical organization

Group-theoretic work on $\mathbb{Z}_m$16-schemes provides another affine-related viewpoint. Under the hypothesis

$\mathbb{Z}_m$17

the orbital scheme on $\mathbb{Z}_m$18 has valencies in $\mathbb{Z}_m$19, and the paper explicitly notes that this is the kind of setup that appears when a transitive permutation group acts on a one-dimensional affine space over $\mathbb{Z}_m$20 or $\mathbb{Z}_m$21. Its main theorem yields

$\mathbb{Z}_m$22

and in the schurian case this gives

$\mathbb{Z}_m$23

These inequalities place strong restrictions on affine-like schemes with thin residue of valency $\mathbb{Z}_m$24 and non-thin valency $\mathbb{Z}_m$25 (Abbas et al., 2021).

A categorical organization of finite association schemes is supplied by French’s subcategory $\mathbb{Z}_m$26 of admissible morphisms. In that setting every finite association scheme, including affine and translation schemes, is an object; admissible morphisms have kernels that are normal closed subsets, and the first isomorphism theorem takes the form

$\mathbb{Z}_m$27

Moreover, the thin radical $\mathbb{Z}_m$28, the thin quotient $\mathbb{Z}_m$29, and the adjacency algebra $\mathbb{Z}_m$30 become functorial on the quotient category $\mathbb{Z}_m$31 (French, 2012).

For group-based affine schemes this framework is especially natural. The translation part of an affine scheme is represented on the scheme side by thin data, while quotient constructions are encoded by closed subsets and thin quotients. The paper does not single out one-dimensional affine schemes, but its functorial treatment of radicals, quotients, and adjacency algebras gives a precise abstract setting in which finite affine translation schemes can be compared, factored, and represented (French, 2012).