claimroutine
Block functions are simplex-valued, low influence
Block functions are simplex-valued with mean , low influence if pseudorandom
puts blocks in the form needed by rounding
Statement (paper)
For each left vertex , the block functions defined in satisfy for every , and for every . If is pseudorandom at threshold , then for every block index and every .
In plain English
For each left vertex , the block functions take values in the simplex , and each coordinate has mean exactly . If is pseudorandom at threshold , then every noised block function has all coordinate influences , , at most .
Why it's here
This puts the block functions of a left vertex in the form needed to apply the Gaussian/noise-stability machinery: simplex-valued, with equal means , and low influence when the vertex is pseudorandom. The local payoff bound for pseudorandom vertices uses it to compare the vertex's ordering payoff with the optimum of the -block problem. The mean comes from the blocks having equal size , which follows from .
Classified routine:
Direct verification from the definition of the block functions: the simplex values and the 1/q means follow from the blocks partitioning the ordering into equal sizes (q | h), and the influence bound is the definition of pseudorandomness at threshold t, as in standard Unique Games dictatorship-test reductions.
Uses
Nothing else in the paper
Used by
Sources
PaperOpen
- Section
- §A.2 The finite ordering transfer
- Page
- —
- Label
—- Line
appendices/finite-consequences.tex:707
Leannot aligned
Phase 2 aligns this node to a Lean declaration.