We consider a bipartite distance-regular graph Γ with diameter D ⩾ 4 , valency k ⩾ 3 , intersection numbers b i , c i , distance matrices A i , and eigenvalues θ 0 > θ 1 > ⋯ > θ D . Let X denote the vertex set of Γ and fix x ∈ X . Let T = T ( x ) denote the subalgebra of Mat X ( C ) generated by A , E 0 * , E 1 * , … , E D * , where A = A 1 and E i * denotes the projection onto the i th subconstituent of Γ with respect to x. T is called the subconstituent algebra (or Terwilliger algebra) of Γ with respect to x. An irreducible T-module W is said to be thin whenever dim E i * W ⩽ 1 for 0 ⩽ i ⩽ D . By the endpoint of W we mean min { i | E i * W ≠ 0 } . Assume W is thin with endpoint 2. Observe E 2 * W is a one-dimensional eigenspace for E 2 * A 2 E 2 * ; let η denote the corresponding eigenvalue. It is known θ ˜ 1 ⩽ η ⩽ θ ˜ d where θ ˜ 1 = - 1 - b 2 b 3 ( θ 1 2 - b 2 ) - 1 , θ ˜ d = - 1 - b 2 b 3 ( θ d 2 - b 2 ) - 1 , and d = ⌊ D / 2 ⌋ . To describe the structure of W we distinguish four cases: (i) η = θ ˜ 1 ; (ii) D is odd and η = θ ˜ d ; (iii) D is even and η = θ ˜ d ; (iv) θ ˜ 1 < η < θ ˜ d . We investigated cases (i), (ii) in MacLean and Terwilliger [Taut distance-regular graphs and the subconstituent algebra, Discrete Math. 306 (2006) 1694–1721]. Here we investigate cases (iii), (iv) and obtain the following results. We show the dimension of W is D - 1 - e where e = 1 in case (iii) and e = 0 in case (iv). Let v denote a nonzero vector in E 2 * W . We show W has a basis E i v ( i ∈ S ) , where E i denotes the primitive idempotent of A associated with θ i and where the set S is { 1 , 2 , … , d - 1 } ∪ { d + 1 , d + 2 , … , D - 1 } in case (iii) and { 1 , 2 , … , D - 1 } in case (iv). We show this basis is orthogonal (with respect to the Hermitian dot product) and we compute the square-norm of each basis vector. We show W has a basis E i + 2 * A i v ( 0 ⩽ i ⩽ D - 2 - e ) , and we find the matrix representing A with respect to this basis. We show this basis is orthogonal and we compute the square-norm of each basis vector. We find the transition matrix relating our two bases for W.
Read full abstract