The Setting
At the initial scale, the question is whether opposite displacements of sparse defects admit trace simulations with a prescribed error budget. Each source bit is deleted with requested probability and survives with probability . The construction theorems use the working rate and ; their parameter called is this working rate.
A finite probability law on a finite set is a nonnegative mass function of total mass one.✦ For a real function on and another law on , expectation, probability overlap, and total variation distance use the conventions in (1). (1) A point mass at an outcome of a finite set is a probability law on that assigns mass one to and zero elsewhere. For a family of laws on a common finite outcome set, indexed by in , the mixture of with this family is the probability law with mass at each outcome ; the pushforward is the mixture using the point mass at for each .✦
A binary word of length is a sequence of bits indexed by the zero-based positions .✦ A binary trace is a finite bit list whose surviving length is part of the observation; denotes the set of all such traces.✦ A deletion mask retains precisely the positions with .✦ The retained subsequence , deletion mask law , and complete trace law are specified in (2). For a length bound , the complete trace space contains all traces of length at most ; a bounded trace is an element of this space.✦ The notation denotes the complete trace expectation of a real test . (2) The complete trace overlap is the overlap of and on their common space with bound .✦ The repeated word is the concatenation of copies of .✦
Approximate trace simulation is the condition in (3) on a finite law and a decoder into complete traces. For a real score on , a positive integer , and a zero-based level index , the probability-weighted squared even moment norm in the same display is taken under , with even score exponent . (3) An approximate oracle pair for with parameters is a pair of finite laws on a common finite outcome set, with a decoder and a real likelihood score satisfying (4). Write for the class of these pairs. (4)✦
A concrete level pair at scale , zero-based level , and half moment order is a pair of distinct binary words , finite laws with a common decoder , a real score , a nonnegative error , and a natural repetition count satisfying (5). Its parameters are the error allowance , decay exponent , overlap threshold , and count multiplier ; write for this class. (5)✦
The Initial Construction
For a natural number , the alternating primitive run is the binary word .✦ For natural numbers , the single defect period is the binary word in (6). For defect slots indexed by , the concatenated defect periods form the binary word in the same display. (6) A Laurent polynomial is a finitely supported real coefficient family indexed by . For a natural motif order , the Prouhet polynomial and motif support indicator in (7) specify a signed displacement . For a motif anchor , a displacement sign , and an interior slot with , the displaced slot and motif defect word use the zero-based period index . (7)
Write and for the floor and ceiling truncated below at zero. At scale , the primitives per period , buffer count , period length , and period capacity are given in (8). The base side count , block count , and period count are given in (8). For , the middle defect slot has the value in that display; the zero-based motif anchor is an integer. (8) The periodic reference word and signed initial word are defined in (9); the motif order is . (9)
The rank Schur constant and primitive period constant are the absolute constants in (10). The initial overlap lower bound and source overlap constant depend on the working rate. (10) The initial moment ratio and initial error bound are the functions in (11). The base primitive score weight specializes this bound to the source initial error , using the parameters of (8). (11) The initial construction geometry is the condition on given by the inequalities in (12). (12)✦
The Smallest Scale
For a block size , slack , and problem size , the induction depth and scale slack are defined by (13). (13) The warmup block count , source block dimension , block gain , and scale exponent are given in (14). The block index is zero-based, so is the dimension of block in the paper; is the zero-based scale index within a block. (14) The scale exponent product defines the backwards target scale in (15). The total score exponent and half moment order use the same block dimensions. (15) A real sequence has power law exponent relative to a scale sequence when it satisfies (16). (16)✦
The Claim
Theorem 1. Let be real sequences and natural numbers. Suppose , for every , and , for all sufficiently large . Assume that has power law exponent zero relative to , that , and that eventually. Put and abbreviate the signed words by and the reference word by . The assertions in (17) hold for all sufficiently large . (17)✦
Fix and . Let and for every , with . Evaluate , , , and at using (13) and (15), and put . The level-zero existence statement (18) holds for all sufficiently large . (18)✦
In the paper, the result reads: for the global parameters, a level-zero pair exists with simulation error and common constants as in Definition 2.2 with , and one may choose with absolute and with .
What Is Checked
The oracle assertion is stated over the signed initial words and their periodic reference, and the level-zero assertion over distinct equal-length binary words with finite decoded laws satisfying the simulation and score conditions. Deletion mask probabilities are nonnegative for ✦, and their sum is one for every real .✦