Symmetric Chains of Tensor Products

• Symmetric chains of tensor products are explicit decompositions of posets that satisfy symmetric rank conditions and offer both combinatorial and linear analogues.
• The linearized BTK construction produces symmetric Jordan bases that enable the block-diagonalization of operators like the Terwilliger algebra in Boolean algebras.
• These structured decompositions facilitate constructive proofs and have applications in coding theory, algebraic combinatorics, and spectral analysis.

A symmetric chain of tensor products refers to a highly structured decomposition of posets formed by the product of chains, with consequential linear and representation-theoretic analogues. These decompositions manifest in both combinatorial and algebraic forms, having deep connections to symmetric Jordan chains, the construction of explicit orthogonal bases in Boolean algebras, and structures such as the Terwilliger algebra of the binary Hamming scheme (Srinivasan, 2010).

1. Combinatorial and Linear Definitions

A finite graded poset $P$ is characterized by a rank function $r:P\to \{0, 1, \dots, r(P)\}$, where covering relations increase rank by one. A symmetric chain in $P$ is a saturated chain $p_1 < \cdots < p_h$ such that $r(p_1) + r(p_h) = r(P)$ (for $h \geq 2$), or $2r(p_1) = r(P)$ if $h=1$. A symmetric chain decomposition (SCD) partitions $P$ into disjoint symmetric chains.

The linear analogue involves the complex vector space $V(P) = \bigoplus_{i=0}^{r(P)} \mathbb{C} P_i$ graded by rank, with the up-operator

$r:P\to \{0, 1, \dots, r(P)\}$0

acting as a nilpotent linear map. A graded Jordan chain with respect to $r:P\to \{0, 1, \dots, r(P)\}$1 consists of vectors $r:P\to \{0, 1, \dots, r(P)\}$2 such that $r:P\to \{0, 1, \dots, r(P)\}$3 for $r:P\to \{0, 1, \dots, r(P)\}$4 and $r:P\to \{0, 1, \dots, r(P)\}$5, with the symmetry property $r:P\to \{0, 1, \dots, r(P)\}$6. A symmetric Jordan basis (SJB) is a basis composed of disjoint symmetric Jordan chains.

For product posets, let $r:P\to \{0, 1, \dots, r(P)\}$7 be nonnegative integers and define

$r:P\to \{0, 1, \dots, r(P)\}$8

ordered componentwise. This is isomorphic to a product of chains, with rank $r:P\to \{0, 1, \dots, r(P)\}$9 and $P$0. Of particular interest are the uniform case $P$1 ($P$2) and the Boolean algebra $P$3.

2. Linearized de Bruijn–Tengbergen–Kruyswijk (BTK) Construction

The linearized BTK algorithm constructs an explicit SJB for $P$4, inductively reducing to lower-dimensional cases. The base case for $P$5 admits two families of homogeneous basis vectors:

• The "main" chain,

$P$6

• The complementary vectors,

$P$7

This set forms a basis of $P$8, breaking into one long symmetric chain and shorter chains inherited from $P$9.

For higher $p_1 < \cdots < p_h$0, the process decomposes $p_1 < \cdots < p_h$1 into direct sums indexed by the $p_1 < \cdots < p_h$2-th coordinate, constructing chains via shift maps and inductive application of the two-dimensional construction. All coefficients remain integral and the decomposition fully explicit.

3. Structure and Orthogonality in the Boolean Algebra Case

Specializing to $p_1 < \cdots < p_h$3, the symmetric Jordan basis arising from the BTK construction—denoted $p_1 < \cdots < p_h$4—is orthogonal under the standard inner product. Explicit singular value ratios govern the norm growth along a chain: $p_1 < \cdots < p_h$5 or equivalently,

$p_1 < \cdots < p_h$6

Chains with the same starting rank are parallel in the sense that their stepwise ratios coincide. This orthogonality arises both from direct computation and a representation-theoretic interpretation using $p_1 < \cdots < p_h$7.

4. Representation-Theoretic Interpretation and the Symmetric Gelfand–Tsetlin Basis

The action of the up-operator $p_1 < \cdots < p_h$8, the down-operator $p_1 < \cdots < p_h$9, and the grading operator $r(p_1) + r(p_h) = r(P)$0 on $r(p_1) + r(p_h) = r(P)$1 realizes an $r(p_1) + r(p_h) = r(P)$2-representation: $r(p_1) + r(p_h) = r(P)$3 Each irreducible $r(p_1) + r(p_h) = r(P)$4 summand yields a unique basis with unequivocal transition rules: $r(p_1) + r(p_h) = r(P)$5 These coincide with the symmetric Jordan chains defined combinatorially.

The symmetric group $r(p_1) + r(p_h) = r(P)$6 acts naturally by coordinate permutation, with multiplicity-free branching for $r(p_1) + r(p_h) = r(P)$7. Every irreducible $r(p_1) + r(p_h) = r(P)$8-component thus supports a canonical Gelfand–Tsetlin basis, characterized as eigenvectors of the Jucys–Murphy elements

$r(p_1) + r(p_h) = r(P)$9

The orthogonal SJB $h \geq 2$0 produced by the linear-BTK algorithm is, up to scaling, the unique symmetric Gelfand–Tsetlin basis (SGZB) for $h \geq 2$1, characterized by being an SJB for $h \geq 2$2 and simultaneous eigenvectors for the $h \geq 2$3.

5. Explicit Block-Diagonalization of the Terwilliger Algebra

The Terwilliger algebra $h \geq 2$4 of the binary Hamming scheme is defined as

$h \geq 2$5

A convenient basis consists of the matrices $h \geq 2$6 indexed by $h \geq 2$7, $h \geq 2$8, with

$h \geq 2$9

The dimension of $2r(p_1) = r(P)$0 is $2r(p_1) = r(P)$1.

The orthogonal change of basis $2r(p_1) = r(P)$2 given by $2r(p_1) = r(P)$3 blocks diagonalizes all $2r(p_1) = r(P)$4. Each $2r(p_1) = r(P)$5 maps to blocks indexed by $2r(p_1) = r(P)$6 (where $2r(p_1) = r(P)$7), with block sizes $2r(p_1) = r(P)$8 and multiplicities $2r(p_1) = r(P)$9. Explicitly, the $h=1$0–block of the $h=1$1th family is

$h=1$2

where $h=1$3 is the $h=1$4 matrix unit and

$h=1$5

These results coincide with Schrijver’s explicit block-diagonal form of $h=1$6 (Srinivasan, 2010).

6. Context and Significance

Symmetric chains of tensor products provide a rich synthesis of combinatorial, algebraic, and representation-theoretic structures. The linearized BTK construction yields explicit decompositions facilitating constructive block-diagonalization of the Terwilliger algebra—critical within coding theory and algebraic combinatorics. The identification of the orthogonal SJB with the symmetric Gelfand–Tsetlin basis establishes profound links between the combinatorics of poset products and the representation theory of symmetric and general linear groups. These results provide new constructive proofs for diagonalizability and explicit formulas for transition coefficients, with implications for spectral analysis in algebraic statistics, coding, and the theory of association schemes (Srinivasan, 2010).