The Setting
The width-two band consists of two tensor layers of on the same positions. Regrouped by position they form one word space over pair letters, a pair letter being a first-layer letter together with a second-layer letter, and a complex matrix, its rows indexed by the first-layer letter and its columns by the second-layer letter, is a vector of the pair-letter space at one position. A list of matrices, one for each position, then gives a pure tensor of the word space, the vector whose coefficient on a word is the product of the entries of the matrices at the letters of that word.
For a natural number , on positions the band holds a distinguished vector , the path generator, whose coefficients are read along overlapping triples of positions. The question is what the contraction of with the pure tensor of matrices is, the contraction of two vectors being the sum over words of the products of their coefficients, as a function of the matrices.
Matrices and the Adjugate Twist
The algebra consists of the complex matrices, with rows and columns indexed by . For , the matrix unit is the matrix with entry in row and column and elsewhere. The symplectic matrix is the element of .✦ The adjugate of a matrix is the transpose of its cofactor matrix, .
For a natural number , the adjugate twist is the map of to itself given by (1)✦ For natural numbers and , the twisted product of matrices is the product of their twists in order, the index of the twist rising from : (2)✦ With and this is , the adjugate standing at the odd positions.
Letters, Words and the Path Generator
The four-letter alphabet is the set of pair letters of the width-two band✦: a letter is the pair of bits with . The matrix unit of a letter is .✦
A word of length is a sequence of letters , written , and the word space on positions is the space of complex functions on these words, the coefficient functions of the tensors of . We write for the contraction of two vectors of the word space, the sum of over all words . For matrices , the coefficients of their pure tensor are (3)✦
The positions of a word of length form overlapping triples with ; the triple has left position ✦, middle position ✦ and right position .✦ For , the path generator is the vector of the word space on the positions whose coefficient on a word vanishes unless and , and is otherwise the product over the triples of the entry of between the second bits at the left and middle positions and the entry of between the first bits at the middle and right positions: (4)✦
The Claim
Theorem 1. Let and let be complex matrices. The contraction of the path generator with their pure tensor is times the entry of their twisted product: (5)✦ Moreover, for a letter and indices , the adjugate of the matrix unit has the entries (6)✦ For a word of length with letters , the twisted product of the matrix units of its letters is a multiple of the matrix unit , (7)✦ and its entry is times the coefficient of on : (8)✦
In the paper, the result reads: for arbitrary , , with matrices, rows and columns numbered from , so that its , , and entry are the page’s , , and entry.
What Is Checked
The claim is stated over the algebra with its matrix units, the symplectic matrix and the adjugate, the adjugate twists and the twisted product , the four-letter alphabet with the bits and the matrix unit of each letter, the pure tensor of matrices, the contraction, and the path generator with its triples of positions.
Beyond the claim, the library proves that each adjugate twist is a linear map of , so that the twisted product is multilinear in its arguments.