The Setting
The boundary operator for word disks over a group combines a disk successor, a band pairing and inverse geometric corner labels. Let and be finite nonempty sets, with indexing coefficient positions and indexing disks of each sign. Write for a permutation of , for the coefficient family and for the exponent family. The word disk slots form the set in (1); the subscripts distinguish two disjoint copies of . (1)✦
For and , the signed disk successor , signed incidence coefficients and traversal sign are the functions on slots specified in (2). The values of are inverse geometric corner labels. (2)
The Operators and Their Phase
For a finite coordinate set , the matrix Hilbert space consists of square-summable complex families with the usual inner product.✦ A bounded operator on is a continuous complex-linear endomorphism of .✦ For and , the delta vector in is the family that is at and elsewhere.✦ The matrix unit on , for , transfers coordinate to coordinate , leaving the coordinate unchanged; is the coordinate projection. For a complex matrix and a scalar family , the scalar embedding and the scalar diagonal operator on are defined in (3). (3)
For a bijection of coordinate sets, the linear coordinate permutation sends a family to .✦ Given a permutation of and a family , the group-labelled coordinate permutation , its labelled shift and the weighted labelled shift are given by (4), where is a scalar weight family. (4)
The scalar rotation and fractional argument have the conventions in (5). On , is the fractional part of the principal argument of divided by . (5)
For a permutation of and a real family , set . The weighted permutation matrix on , its scalar operator on and the boundary operator on are defined by (6), with matrix indices ; the permutation applies first. (6)
For a complex matrix , the coefficient shift on , scalar mixing matrix on and scalar mixer on are specified in (7). Each block of acts on the positive and negative coordinates at one position . (7)
Let be a normal bounded operator on a complex Hilbert space and a vector. For a compactly supported continuous real function on the spectrum of , its complexification is the same function viewed as complex-valued.✦ The spectral vector functional and its finite representing measure satisfy (8); is positive and real-linear. (8)
For a normal bounded operator on a complex Hilbert space and a vector in that space, the vector spectral phase is the integral in (9).✦ For a normal bounded operator on a complex Hilbert space, the summed vector phase of a finite family of vectors in that space is the sum of their vector spectral phases for .✦ For a normal bounded operator on , the unnormalized phase trace is the sum in (9) of the vector spectral phases of the delta vectors at the identity . (9)
The fixed space of a complex matrix on is the subspace in (10), where is the identity matrix on ; its fixed space dimension is its complex dimension. (10)
The Claim
Theorem 1. For every unitary matrix , the phase of the product of the coefficient shift and the scalar mixer is given by (11). (11)✦ Let each be or , with sum or . Suppose that is an involution reversing traversal signs, so that , and that is antisymmetric, so that . There is a unitary matrix satisfying (12). (12)✦ For any finite set and unitary complex matrices with , the scalar phase estimate is (13). (13)✦ Under the stated conditions on , and , the boundary operator satisfies (14). (14)✦
In the paper, the result reads: for a connected disk-and-band assembly, every antisymmetric choice of defines and with , where here denotes the identity operator, and a scalar mixing involution can be chosen so that has at least fixed directions, , , and .
What Is Checked
The mixing and boundary claims (11), (12) and (14) are stated over arbitrary coefficient groups, finite nonempty position and disk index sets, and arbitrary position permutations, with the indicated conditions on signs, pairing and weights. The scalar estimate (13) concerns finite complex matrices and a unitary conjugacy between a matrix and its adjoint.
Coordinate permutations associated with bijections preserve square-summability✦ and inner products, and their linear reindexing maps are unitary. Scalar embedding preserves unitarity✦, and every scalar rotation is unitary.✦ A weighted group-labelled shift is unitary when all its scalar weights have modulus one.✦ The coefficient shift is unitary✦, as is for every unitary .✦