The Setting
The signed walk is built by a linear filter that takes vectors in one at a time, a finite set of coordinates. The filter keeps a real matrix of coefficients: its rows are the coordinates, and its columns are a state block, a copy of , followed by one noise coordinate for each vector. Each vector moves the matrix by one step, and just before that step the filter reads off a row of coefficients, the predictor row of the vector.
The Constants
The scale that normalizes the predictor rows is built from a chain of numerical constants. The gain constant is ; the epoch factor , the state rate , the body rate and the body scale are (1) and the number is the slab radius.
The Definition
Definition 1. Let be a finite set and a set, and let be a real matrix with rows indexed by and columns by . A filter step on such matrices is a triple (2)✦ A filter step acts on a real matrix with rows indexed by and columns by by (3)✦ where , with the -th standard basis vector, is the matrix whose column is and whose other columns vanish. A filter step is good if is invertible and (4)✦ The run of through a list of filter steps applies the steps in the order of the list, (5)✦ and is itself for the empty list. The matrix has fresh columns for if every column into which a step of writes is zero in : (6)✦ When is finite, the state matrix of for weights on its columns is the matrix (7)✦
The mass of a vector is its squared Euclidean norm (8)✦ its normalizer is the real number (9)✦ and its symmetric contraction is the matrix (10)✦ The symmetric step of into a column is the filter step (11)✦
The raw predictor row of at is the row vector indexed by (12)✦ The predictor rows of along a list of filter steps, with , are the raw predictor rows of the vectors at the matrices their steps meet: with , (13)✦
For a set of noise coordinates, the coefficients are the state block followed by the noise block . The initial matrix on the coefficients is the matrix (14)✦ The noise part of a vector is the vector that agrees with on the noise block and vanishes on the state block.✦ For vectors in and , the family of symmetric steps writes each into its noise coordinate : (15)✦ The predictor row of is the noise part of the -th predictor row of along , (16)✦ The signing scale is the real number (17)✦ and the normalized predictor row of is the row vector (18)✦
In the paper, the result reads: , , and , for the paper’s vectors , with the matrix that the step of meets and a vector of coefficients.
What Is Checked
The definitions are stated over real matrices with rows indexed by a finite set and columns by a set , and, for the signed walk, over vectors in and the numerical constants and .