Cycle Relevant Matrix (CRM) Overview

• Cycle Relevant Matrix (CRM) is a family of matrices that encode cycle constraints across different domains, including LDPC codes, cycle sets in QYBE, and graph cycle detectors.
• In coding theory, the CRM (or cycle consistency matrix) transforms cycle conditions in Tanner graphs into modular linear equations, guiding the design of SCB LDPC codes with reduced error floors.
• Alternate CRM constructions in quantum and graph theories utilize specific algebraic rules or set-matrix products, illustrating how domain-specific operations yield practical cycle-detection or structural insights.

“Cycle Relevant Matrix” (CRM) is not a single standardized object across the arXiv literature. In the supplied nomenclature, it names several matrices whose entries encode cycle-relevant structure in distinct domains: the cycle consistency matrix used to characterize absorbing sets in separable circulant-based LDPC codes, the cycle matrix associated with non-degenerate cycle sets and involutive set-theoretic solutions of the Quantum Yang–Baxter equation, and the set matrix $S_k$ whose diagonal detects $k$-cycles in directed graphs. The same acronym also appears independently as “Convergent r-Matrix,” which is unrelated to cycle encoding and instead concerns finite-dimensional matrix encodings of the position operator (Wang et al., 2012, Kanrar et al., 2023, 0709.4273, Song et al., 2024).

1. Terminological scope and domain-specific meanings

The term CRM is introduced in the supplied material as a nomenclature mapping rather than as a canonical term native to all cited papers. In coding theory, the paper “The Cycle Consistency Matrix Approach to Absorbing Sets in Separable Circulant-Based LDPC Codes” identifies the cycle consistency matrix (CCM) with the CRM: it is the matrix of modular cycle constraints that must vanish for a candidate absorbing-set topology to be realizable in a separable circulant-based code (Wang et al., 2012). In the Quantum Yang–Baxter literature, the paper “Cycle matrices: A combinatorial approach to the set-theoretic solutions of the Quantum Yang–Baxter Equation” uses “cycle matrix” for an $n\times n$ matrix $M=[m_{ij}]$ encoding the operation $i\cdot j=m_{ij}$ on $U_n=\{1,\dots,n\}$, and the supplied mapping identifies that object with the CRM (Kanrar et al., 2023). In graph algorithms, “Set Matrices and The Path/Cycle Problem” does not use the term CRM explicitly; the supplied mapping defines the CRM for cycle length $k$ to be the paper’s diagonal cycle-detection matrix $S_k$ (0709.4273).

Literature	Matrix identified as CRM	Function
SCB LDPC codes	Cycle consistency matrix (CCM)	Encodes modular cycle constraints for absorbing sets
Cycle sets / QYBE	Cycle matrix $M=[m_{ij}]$	Encodes the cycle operation $i\cdot j$
Graph path/cycle problem	Set matrix $k$0	Detects $k$1-cycles through diagonal entries
Position-operator theory	Convergent r-Matrix	Unrelated acronym usage

This plurality of meanings is substantive rather than cosmetic. In each case, the matrix records a different kind of cycle information: congruence constraints in Tanner graphs, binary operations satisfying the cycle law, or existence of graph cycles of fixed length. The shared label therefore denotes a family resemblance at the level of function, not a single algebraic formalism.

2. CRM as cycle consistency matrix in SCB LDPC codes

In separable circulant-based (SCB) LDPC coding, the CRM is the cycle consistency matrix attached to a candidate absorbing-set subgraph. The underlying setting is an LDPC Tanner graph $k$2, where an $k$3 absorbing set is a variable-node subset $k$4 such that $k$5, $k$6, and every variable node in $k$7 has strictly fewer neighbors in the odd-degree checks $k$8 than in the even-degree checks $k$9. A fully absorbing set imposes the analogous strict inequality for every variable node outside $n\times n$0. The paper treats absorbing sets as a principal cause of the error floor for LDPC codes under limited-precision message-passing decoding on AWGN channels, with small absorbing sets tending to dominate because they are more easily activated (Wang et al., 2012).

SCB codes form a large subclass of regular quasi-cyclic LDPC codes. Their parity-check matrices are block-circulant with circulant size $n\times n$1, typically prime, and take the form

$n\times n$2

with $n\times n$3 the $n\times n$4 circulant permutation matrix and separability

$n\times n$5

All SCB codes over a fixed $n\times n$6 derive from the mother matrix $n\times n$7 with $n\times n$8. Variable nodes are labeled $n\times n$9, check nodes $M=[m_{ij}]$0, and the graph obeys Bit Consistency, Check Consistency, and Pattern Consistency. The cycle constraints are captured by

$M=[m_{ij}]$1

for any length-$M=[m_{ij}]$2 cycle with variable-group labels $M=[m_{ij}]$3 and check-group labels $M=[m_{ij}]$4.

For an absorbing set with variable-group labels $M=[m_{ij}]$5, the paper defines

$M=[m_{ij}]$6

and rewrites the cycle equations as linear congruences in the vector $M=[m_{ij}]$7: $M=[m_{ij}]$8 Choosing one row per linearly independent cycle produces the CRM/CCM $M=[m_{ij}]$9, so that

$i\cdot j=m_{ij}$0

Its rows correspond to linearly independent cycles, its columns to variable-node column-group differences, and an efficient construction uses exactly $i\cdot j=m_{ij}$1 rows, where $i\cdot j=m_{ij}$2 is the dimension of the binary cycle space. When all check degrees in the absorbing-set subgraph are at most $i\cdot j=m_{ij}$3, the paper passes to the VN graph and computes $i\cdot j=m_{ij}$4 as the nullity of the VN-graph incidence matrix over $i\cdot j=m_{ij}$5.

The central existence criterion is stated as Theorem 1. For a proposed absorbing-set graph $i\cdot j=m_{ij}$6 with all variables in at least one cycle, and with prescribed row-group and column-group labels inherited from the SCB mother matrix, the following are necessary for the absorbing set to exist in every daughter SCB code selecting those groups: the CRM does not have full column-rank over $i\cdot j=m_{ij}$7, equivalently it has a nontrivial null space; the variable nodes satisfy Check Consistency and produce a difference vector $i\cdot j=m_{ij}$8 in that nullspace; and each check satisfies Bit Consistency. If the VN graph is non-extensible, these conditions are also sufficient.

The paper gives explicit CRMs for several small absorbing-set topologies. For the $i\cdot j=m_{ij}$9 absorbing set,

$U_n=\{1,\dots,n\}$0

with determinant

$U_n=\{1,\dots,n\}$1

For the $U_n=\{1,\dots,n\}$2 absorbing set, an efficient CRM is

$U_n=\{1,\dots,n\}$3

and its determinant factors as

$U_n=\{1,\dots,n\}$4

For the $U_n=\{1,\dots,n\}$5 absorbing set at $U_n=\{1,\dots,n\}$6,

$U_n=\{1,\dots,n\}$7

These formulas make the CRM a direct mechanism for converting graphical feasibility into modular linear algebra.

3. CRM-guided code design, multiplicity, and performance in LDPC constructions

Because absorbing-set existence requires a nontrivial CRM nullspace, the design principle in SCB LDPC codes is to choose the row-selection function $U_n=\{1,\dots,n\}$8, the column-selection function $U_n=\{1,\dots,n\}$9, and the circulant size $k$0 so that the relevant CRMs have trivial nullspace, equivalently full column-rank, for the targeted small topologies (Wang et al., 2012). The paper emphasizes that this is especially useful in the high-rate regime, where the variable-node degree $k$1 is small, such as $k$2 or $k$3, and small absorbing sets are otherwise prevalent. Lower-rate SCB codes often have larger $k$4, which increases the number of satisfied checks per variable and tends to make avoidance easier.

For $k$5, row selection alone can eliminate several dominant absorbing sets. To avoid $k$6 absorbing sets, the determinant expressions associated with the admissible row-group labelings must remain nonzero modulo $k$7. Explicit examples are given: the RSF $k$8 avoids all $k$9 absorbing sets for $S_k$0, and $S_k$1 avoids all $S_k$2 absorbing sets for $S_k$3 and certain other primes. For $S_k$4, the RSF $S_k$5 avoids $S_k$6 for $S_k$7 except the finite set $S_k$8. Theorem 7 states that for $S_k$9 SR-SCB codes, with proper RSF and sufficiently large $M=[m_{ij}]$0, all $M=[m_{ij}]$1, $M=[m_{ij}]$2, and $M=[m_{ij}]$3 absorbing sets can be precluded. The paper also lists good RSFs such as $M=[m_{ij}]$4 for $M=[m_{ij}]$5, $M=[m_{ij}]$6, $M=[m_{ij}]$7, $M=[m_{ij}]$8, and $M=[m_{ij}]$9, each with a minimal $i\cdot j$0 threshold beyond which all three targeted structures are avoided.

For $i\cdot j$1, the dominant structures are $i\cdot j$2 absorbing sets, and some configurations cannot be eliminated by row selection alone. The prescribed remedy is shortening through column selection: one chooses a subset of column groups $i\cdot j$3 so that no dangerous $i\cdot j$4-tuple of column-group labels remains in the solution set of $i\cdot j$5. The paper describes two greedy procedures, column-cutting and column-adding, to enforce $i\cdot j$6, where $i\cdot j$7 is the set of allowable $i\cdot j$8-tuples of column labels and $i\cdot j$9 is the set producing the absorbing sets. It also notes that shortening reduces rate but does not introduce smaller absorbing sets.

The CRM framework also yields multiplicity information. In EAB codes, small absorbing sets often occur with multiplicity scaling as $k$00. Corollary 2 states that $k$01 absorbing sets exist in every $k$02 EAB code and the total multiplicity is $k$03; Corollary 5 gives the same $k$04 scaling for $k$05 in $k$06 EAB codes. Certain $k$07 configurations likewise occur with multiplicity $k$08, although many contain $k$09 subgraphs and disappear once the smaller topology is eliminated.

The search space for good RSFs is reduced by equivalence. RSFs are equivalent under addition of a constant modulo $k$10, multiplication by a nonzero constant modulo $k$11, and equivalence of difference matrices including antidiagonal reflections. Any RSF is therefore equivalent to one of the form $k$12. For $k$13, the paper reports $k$14 such equivalence classes, of which $k$15 class leaders avoid all $k$16, $k$17, and $k$18 absorbing sets.

The computational pipeline is correspondingly explicit: specify the absorbing-set topology, identify linearly independent cycles, assemble the CRM from the cycle equations, enforce Bit and Check Consistency, and compute the rank of $k$19 over $k$20. For small absorbing sets, both $k$21 and $k$22 are small, so rank computation is an $k$23 operation on small matrices. Simulation results connect this algebraic elimination to decoder behavior. Under limited-precision SPA and SXOR decoding, including Q4.2 quantization and $k$24 iterations, CRM-optimized codes exhibit steeper error-floor slopes and “one order of magnitude of improvement in the low FER region.” For $k$25 codes at $k$26, the SR-SCB code with RSF $k$27 eliminates all $k$28 absorbing sets and outperforms EAB and Huang et al. constructions under both SPA and SXOR. For $k$29 codes at $k$30, the SR-SCB choice $k$31 eliminates $k$32, $k$33, and $k$34, with corresponding BER improvement.

4. CRM as cycle matrix in cycle sets and set-theoretic QYBE

In the cycle-set literature, the CRM is the cycle matrix $k$35, where $k$36 and the matrix defines a binary operation by $k$37. The matrix qualifies exactly when $k$38 is a non-degenerate cycle set: for each $k$39, the map $k$40 is bijective; the cycle law

$k$41

holds for all $k$42; and the diagonal map $k$43 is bijective. Square-free cycle sets are those with $k$44. Writing the $k$45-th row as a permutation $k$46, where $k$47, the cycle law is equivalent to

$k$48

for all $k$49 (Kanrar et al., 2023).

This yields a complete criterion for determining whether a given matrix is a CRM in this sense: each row must be a permutation of $k$50; the diagonal entries must themselves form a permutation of $k$51; and the row permutations must satisfy the cycle-law identity above. The paper gives an algorithmic verification cost of $k$52 for the row-permutation test, $k$53 for the diagonal permutation test, and worst-case $k$54 for checking the cycle law.

The matrix is structurally tied to involutive set-theoretic solutions of the Quantum Yang–Baxter equation through Rump’s correspondence between non-degenerate cycle sets and non-degenerate involutive solutions. The explicit solution attached to the CRM is

$k$55

equivalently

$k$56

The cycle matrix therefore serves not only as a storage device for the operation $k$57 but as a constructive object from which the QYBE solution is recovered.

The paper develops several structural operations on these matrices. If $k$58 and $k$59 are cycle sets, then $k$60 with

$k$61

is again a cycle set, and under the relabeling $k$62, the corresponding cycle matrix is the tensor product $k$63. The retraction construction identifies elements $k$64 when $k$65, forms $k$66, and defines the multipermutation level as the least $k$67 with $k$68. Using block cycle matrices, the paper constructs level-$k$69 solutions and, by an iterative doubling scheme beginning with $k$70, constructs a multipermutation solution of level $k$71 for every $k$72.

The matrix also supports classification statements. There is an $k$73-action on the set $k$74 of cycle matrices given by

$k$75

and two cycle matrices determine isomorphic solutions if and only if they lie in the same orbit. The stabilizer of a matrix is the automorphism group $k$76. The paper further states that if the permutation group generated by the rows acts intransitively, then the determinant of the symbolic matrix vanishes; consequently, a non-zero determinant implies indecomposability, although the converse fails. It also proves that every finite abelian group occurs as the permutation group of a non-degenerate involutive solution obtained by these cycle-matrix constructions.

5. CRM as a set-matrix detector of graph cycles

In the graph-theoretic framework of set matrices, the CRM for cycle length $k$77 is the matrix $k$78, whose diagonal entries detect the existence of $k$79-cycles attached to individual vertices. The paper works with a directed multigraph $k$80, where $k$81 is also the universal set used inside the matrix entries. A set matrix is a matrix $k$82 whose entries are subsets of $k$83, and the key multiplication for path filtering is

$k$84

Because of the special treatment of the diagonal, this multiplication is explicitly non-associative (0709.4273).

Two adjacency-derived set matrices are defined. The finish-label matrix $k$85 stores $k$86 in position $k$87 when there is an arc $k$88 with $k$89, and $k$90 otherwise. The start-label matrix $k$91 stores $k$92 under the analogous condition and $k$93 otherwise. Their recursively defined powers $k$94 and $k$95 support fixed-length path detection. The key theorem for paths is

$k$96

Cycles require a different multiplication that reads only the diagonal: $k$97 The base cycle matrix $k$98 records loops by setting $k$99 when $n\times n$00, and $n\times n$01 otherwise. For $n\times n$02,

$n\times n$03

Under the CRM identification, $n\times n$04.

The central cycle theorem is then

$n\times n$05

Only the diagonal entries of the CRM matter, since the off-diagonal entries are set to $n\times n$06 by definition. The matrix therefore does not enumerate cycles in the usual adjacency-matrix sense; instead, it acts as a set-theoretic filter whose nontrivial diagonal entries certify cycle existence of a fixed length.

The computational profile is also explicit. Naively computing the path powers costs

$n\times n$07

and computing a single cycle matrix $n\times n$08 from $n\times n$09 and $n\times n$10 costs $n\times n$11, dominated by the path-power step. For all paths and cycles of lengths up to $n\times n$12, the worst-case total time is $n\times n$13 on complete graphs. The paper notes a Hamiltonian-cycle optimization: by computing only a single row or column of $n\times n$14, detection can be reduced to $n\times n$15. The CRM in this setting is therefore a diagonal diagnostic for fixed-length cycle membership, embedded in a non-associative set-matrix calculus.

6. Common structural themes and acronym ambiguity

Across these three cycle-centered usages, the CRM is always a matrix that compresses cycle information into a form amenable to algebraic testing, but the algebra being used changes completely from one domain to another. In SCB LDPC coding, the entries lie in modular arithmetic and the decisive event is a nontrivial nullspace of $n\times n$16 over $n\times n$17. In cycle-set theory, the matrix entries lie in $n\times n$18, each row is a permutation, and admissibility is controlled by the cycle law $n\times n$19. In graph algorithms, the entries are subsets of a vertex set, the multiplication is non-associative, and cycle existence is read off from whether diagonal entries differ from the universal set. The shared function is thus representational: each CRM turns a cycle-existence or cycle-compatibility problem into a matrix condition.

The papers also differ in what the matrix certifies. The LDPC CRM is local to a prescribed absorbing-set topology and is used to prove presence, absence, multiplicity scaling, and code-design constraints. The QYBE CRM is global: it defines an entire algebraic structure $n\times n$20, from which one derives the associated involutive solution, retractions, tensor products, and automorphisms. The graph-theoretic CRM is length-indexed and vertex-indexed: $n\times n$21 answers whether a $n\times n$22-cycle is attached to a given vertex, without serving as a presentation of an independent algebraic object.

The acronym itself is ambiguous beyond cycle-related contexts. The paper “Position operators in terms of converging finite-dimensional matrices: Exploring their interplay with geometry, transport, and gauge theory” uses CRM to mean “Convergent r-Matrix,” a finite-dimensional encoding of the position operator $n\times n$23. That CRM is explicitly not a finite-dimensional representation of the Weyl algebra, because finite-dimensional matrices cannot satisfy $n\times n$24 owing to the trace obstruction. Its role is to replace the divergent diagonal term in the conventional $n\times n$25-matrix by a finite construction on a discretized product space $n\times n$26, with the reduced matrix

$n\times n$27

and inhomogeneous transformation law characteristic of differential operators rather than Lie-algebra matrices (Song et al., 2024).

The literature therefore contains both a family of cycle-centered CRM mappings and an unrelated acronymic use. In the cycle-centered cases, CRM is best understood not as a single universal object but as a domain-specific matrix formalism whose common purpose is to encode cycle-relevant constraints, operations, or existence tests in a tractable algebraic language.