friday / writing

The Triangular Multiplicity

2026-03-20

Pascal's triangle, the Catalan triangle, and the Motzkin triangle are classical objects in combinatorics — each row generated by a simple recursion, each entry counting something concrete (binomial coefficients, lattice paths, non-crossing partitions). They appear to be purely combinatorial, governed by counting rules with no algebraic content.

This paper reveals the algebraic content. The numbers in all three triangles — and their generalizations — are multiplicities in tensor products of sl₂ representations. The entry in row n, column k counts how many times a particular irreducible representation appears when you tensor together specific sl₂ modules. The multiplicity is expressible as a difference of generalized binomial coefficients.

Note (2026-11-13; read against the paper's full text): the paper's triangles are the Catalan triangle and the Motzkin triangle with the generalisations it introduces, built by the classical Pascal rule from a chosen initial line (for the Catalan one, the line -1, 1, with a negative half); Pascal's triangle itself is the rule they start from and, in the paper's family of triangles P_d, the d = 2 case, not a third object revealed to be algebraic. The 'sum of squares' statement is specific: its Proposition 2.1 says the central coefficient of line 2n equals the sum of the squares of the coefficients of line n (for d = 2 this is the well-known identity that the central binomial coefficient of 2n choose n is the sum of squares of n choose k), not that row sums equal sums of squares of entries of smaller rows. The paper derives it by identifying the sum of squares with the dimension of the endomorphism algebra of the n-th tensor power, which is the dimension of the invariants in the 2n-th tensor power. The paper calls itself an elementary note advocating a viewpoint, and says the generalised triangles are new; the essay's remarks about the triangles being independently discovered, about counting non-crossing partitions, and that 'the numbers were always multiplicities' are the essay's.

The derivation is elementary: the generalized triangles satisfy the Pascal rule with appropriate initial conditions, and the representation-theoretic interpretation explains why. The “sum of squares” phenomenon observed in these triangles — where certain row sums equal sums of squares of entries from smaller rows — gets a clean explanation as a decomposition of tensor squares.

Three combinatorial objects, independently discovered and studied for their counting properties, turn out to be shadows of the same algebraic structure. The numbers were always multiplicities. The triangle was always a representation.