Cyclic Orbit Flag Codes

• Cyclic orbit flag codes are nested subspace codes generated by the cyclic group action on flags in a finite vector space, unifying constant-dimension and orbit code approaches.
• They employ algebraic invariants such as the best friend and best friend vector to establish key parameters like size, minimum distance, and optimality in network coding.
• Recent constructions leverage group-theoretic methods and block-diagonal embeddings to maximize cardinality and approach optimum distance, bolstering error-control capabilities.

A cyclic orbit flag code is a class of codes in the context of network coding, consisting of sequences of nested subspaces (flags) of a finite field vector space, constructed as the orbit of a flag under the action of a cyclic group or subgroup of the general linear group. These codes generalize both constant-dimension subspace codes and Galois-theoretic spread and orbit codes. Their systematic study connects algebraic combinatorics, group actions, finite geometry, and error-control coding.

1. Formal Definitions and Constructions

Let $\mathbb{F}_q$ denote a finite field of order $q$, and let $V = \mathbb{F}_q^n$ be the standard $n$-dimensional vector space over $\mathbb{F}_q$. A flag of type $t = (t_1, \ldots, t_r)$, where $1 \leq t_1 < \cdots < t_r < n$, is a strictly nested sequence of subspaces:

$$F = (F_1 \subset F_2 \subset \cdots \subset F_r), \qquad \dim F_i = t_i.$$

The set of all flags of type $t$ in $V$ is denoted $\mathcal{F}_q(t;n)$.

The flag distance between two flags $F, F'$ is given by

$d_f(F, F') = \sum_{i=1}^r d_s(F_i, F'_i),$

where $d_s(U, V) = \dim(U+V) - \dim(U\cap V)$ is the classical subspace distance.

The general linear group $\mathrm{GL}(n,q)$ acts on flags by componentwise action on their representative matrices. Given $G \leq \mathrm{GL}(n,q)$ and a flag $F$, the $G$-orbit is:

$\mathrm{Orb}_G(F) = \{ F \cdot g : g \in G \},$

where the action is defined on each subspace individually.

If $G$ is cyclic—typically a subgroup generated by a single invertible matrix or, under field isomorphism, an element of $\mathbb{F}_{q^n}^*$—$\mathrm{Orb}_G(F)$ is called a cyclic orbit flag code (Alonso-González et al., 2021, Navarro-Pérez et al., 2021). For $\beta \in \mathbb{F}_{q^n}^*$, the “$\beta$-cyclic orbit flag code” generated by $F$ is:

$$C = \mathrm{Orb}_{\langle \beta \rangle}(F) = \{ F\beta^i : 0 \leq i < \operatorname{ord}(\beta) \}.$$

The cardinality follows from the orbit-stabilizer theorem:

$$|C| = \frac{|\langle \beta \rangle|}{|\mathrm{Stab}_{\langle \beta \rangle}(F)|},$$

with $\mathrm{Stab}_{\langle \beta \rangle}(F)$ the subgroup of elements fixing $F$ (Alonso-González et al., 2021).

2. Algebraic Invariants: Best Friend and Best Friend Vector

A central invariant of a flag $F$ is its best friend: the largest subfield $\mathbb{F}_{q^m} \subset \mathbb{F}_{q^n}$ over which each $F_i$ is an $\mathbb{F}_{q^m}$-vector space. The best friend $m$ serves as a fundamental parameter for the cyclic orbit; it both determines code size and imposes divisibility constraints on the code’s distance (Alonso-González et al., 2021, Alonso-González et al., 2023).

The best friend vector $\mathbf{b} = (m_1, m_2, \ldots, m_r)$ records, for each $F_i$, the largest $m_i$ such that $F_i$ is an $\mathbb{F}_{q^{m_i}}$-subspace. The overall best friend is $\mathbb{F}_{q^{m}}$, where $m = \gcd(m_1, \ldots, m_r)$ (Alonso-González et al., 2023). The cardinality of $C = \mathrm{Orb}(F)$ then becomes

$|C| = \frac{q^n - 1}{q^{m} - 1},$

and for each projection,

$|C_i| = \frac{q^n - 1}{q^{m_i} - 1}.$

The best friend vector refines estimations for minimum distance, constrains admissible type vectors, and reflects the arithmetic structure required for realizability (Alonso-González et al., 2023).

3. Distance Metrics and Bounds

The minimum flag distance $d_f(C)$ satisfies general divisibility and bounding properties. If the best friend is $\mathbb{F}_{q^m}$, all nontrivial pairwise distances $d_f(F, F\cdot \beta^j)$ are multiples of $2m$:

$d_f(C) \geq 2m.$

Sharper lower bounds arise when exactly $j$ coordinates of the best friend vector equal $m$: $d_f(C) \geq 2m\,j$ (Alonso-González et al., 2023). The tightest upper bound is the absolute maximum

$$D_{\max}(t;n) = \sum_{i=1}^r \min\{2t_i, 2(n-t_i)\},$$

which is met by optimum distance codes (Alonso-González et al., 2021, Navarro-Pérez et al., 2021).

A code is termed optimum distance if $d_f(C) = D_{\max}(t;n)$. For cyclic orbit flag codes, this can only occur under severe restrictions on the type vector, which, for fixed best friend $m$, must be $(m)$, $(n-m)$, or $(m, n-m)$. These types correspond to spread codes and their “mixed” two-step analogues (Alonso-González et al., 2021).

4. Distinguished Families: Galois, Generalized Galois, and Extended Constructions

Galois cyclic orbit flag codes are constructed from flags whose subspaces are the intermediate subfields in a chain $$\mathbb{F}_{q^{t_1}} \subset \cdots \subset \mathbb{F}_{q^{t_r}} \subset \mathbb{F}_{q^n}$$, with $t_1 \mid t_2 \mid \cdots \mid t_r \mid n$. Such codes attain the strict minimum flag distance $d_f = 2t_1$ and are optimum distance by construction, with cardinality $(q^n-1)/(q^{t_1}-1)$ (Alonso-González et al., 2021, Alonso-González et al., 2021).

Generalized Galois flag codes extend this principle by allowing the flag to include at least one subspace that is not a subfield, while containing a Galois subflag as a subsequence. Their minimum distance is lower but can be analyzed in terms of the number and distribution of subfields among the coordinates (Alonso-González et al., 2021).

Cardinality-consistent flag codes with longer type vectors are provided by cyclic orbit constructions leveraging block-diagonal embeddings of cyclic subgroups, yielding families with strictly increasing type vectors—such as $$t_{\mathrm{long}}=(1,2,\ldots,k+h,2k+h,\ldots,(s-2)k+h, n-k,\ldots,n-1)$$—and maintaining both high minimum flag distance and maximum attainable cardinality $\sum_{i=1}^{s-1} q^{ik+h} + 1$ (Jia et al., 13 Jan 2026).

5. Algebraic and Group-Theoretic Constructions

The realization of cyclic orbit flag codes at scale depends upon the group-theoretic properties of the cyclic or Singer subgroup acting on the flag variety. Singer groups—cyclic groups of order $q^n-1$—enable the construction of Desarguesian spreads, providing the backbone for many full-type optimum distance flag codes (Navarro-Pérez et al., 2021).

The field-reduction technique (embedding of $\mathrm{GL}(s, q^k)$ into $\mathrm{GL}(sk, q)$) constructs flag codes whose projections are linked to spreads in the Grassmannian. Block-matrix generator constructions yield explicit flags whose orbits under Singer subgroups are cardinality-maximal and distance-optimal (Navarro-Pérez et al., 2021, Jia et al., 13 Jan 2026).

The interplay between the group action, the type vector, and the best friend vector is critical in ensuring that the constructed code is cardinality-consistent, optimum distance, or achieves prescribed tradeoffs.

6. Parameter Relations and Classification

Key parameters of cyclic orbit flag codes—best friend, type, size, and minimum distance—are interdependent. The following table summarizes main families (Alonso-González et al., 2021):

Family	Best Friend	Type	Size	Distance
Galois	$\mathbb{F}_{q^{t_1}}$	$(t_1, ..., t_r)$	$(q^n-1)/(q^{t_1}-1)$	$2t_1$
spread (optimum, $m$)	$\mathbb{F}_{q^m}$	$(m)$	$(q^n-1)/(q^m-1)$	$2m$
spread (optimum, $n-m$)	$\mathbb{F}_{q^m}$	$(n-m)$	$(q^n-1)/(q^m-1)$	$2(n-m)$
mixed optimum	$\mathbb{F}_{q^m}$	$(m, n-m)$	$(q^n-1)/(q^m-1)$	$2(n-m)$

For longer or full type vectors, cardinality-consistent cyclic orbit constructions maintain maximum code size within the arithmetic constraints specific to the field and type vector (Jia et al., 13 Jan 2026).

7. Refinements, Applications, and Recent Advances

The best friend vector invariant refines the best friend approach, providing finer control over size, distance, and realizability issues. The construction of cyclic orbit flag codes with prescribed best friend vectors is governed by number-theoretic divisibility and least common multiple/gcd relations between type coordinates and friend exponents (Alonso-González et al., 2023).

Recent work demonstrates that flag codes of maximal cardinality and distance can be achieved via intricate orbit constructions over block-diagonal cyclic subgroups, producing two infinite families of cardinality-consistent flag codes: one attaining flagged optimum distance for the admissible type $(1,2,\ldots,k, n-k,\ldots,n-1)$, and one with even longer type vectors while retaining near-optimal projection distances (Jia et al., 13 Jan 2026).

Cyclic orbit flag codes and their descendants play an essential role in network coding for error-resilience, extending the performance and structural properties of classical subspace codes to more intricate algebraic-geometry-inspired geometries.

---

Key references: (Alonso-González et al., 2021) (Alonso-González & Navarro-Pérez), (Navarro-Pérez et al., 2021) (Navarro-Pérez & Soler-Escrivà), (Alonso-González et al., 2023, Alonso-González et al., 2021, Jia et al., 13 Jan 2026).