claimroutine

Coarse completeness in independent neighbor experiment

Strong labeling gives payoff ≥ (1−ζ)(1−2κ)kvloc(1-\zeta)(1-2\kappa)^k v_{\rm loc}
completeness in the independent-neighbor experiment
Claim·§A.2 The finite ordering transfer·inside propFinite ordering transfer from Unique Games

Statement (paper)

If a labeling LL strongly satisfies at least (1−ζ)∣A∣(1-\zeta)|A| left vertices, then the coarse rank map f(b,z)=zL(b)f(b,z)=z_{L(b)} satisfies V(f)≥(1−ζ)(1−2κ)kvlocV(f)\ge (1-\zeta)(1-2\kappa)^k v_{\mathrm{loc}}, where V(f)=E P0‾(f(w1),…,f(wk))V(f)=\mathbb E\,\overline{P_0}(f(w_1),\ldots,f(w_k)) is taken over the independent neighbor experiment with raw output positions wi=(bi,πa→biZ~i)w_i=(b_i,\pi_{a\to b_i}\widetilde Z_i).

In plain English

If a labeling LL strongly satisfies at least (1−ζ)∣A∣(1-\zeta)|A| left vertices, then the coarse rank map f(b,z)=zL(b)f(b,z)=z_{L(b)} achieves expected payoff V(f)≥(1−ζ)(1−2κ)kvlocV(f)\ge(1-\zeta)(1-2\kappa)^k v_{\mathrm{loc}} in the independent neighbor experiment.

Why it's here

This is the completeness half of the finite ordering transfer: a good Unique Games labeling yields a rank assignment whose value is close to the local value vloc≥1−ηv_{\mathrm{loc}}\ge1-\eta from the local instance. The loss factors reflect left vertices that are not strongly satisfied (the 1−ζ1-\zeta) and the noise perturbation of the kk sampled coordinates (the (1−2κ)k(1-2\kappa)^k). The later claim that conditions on distinct neighbors uses this lower bound to obtain the strict gap.
Classified routine:
This is the standard completeness analysis of a Khot–Kindler–Mossel–O'Donnell style dictatorship-test reduction from Unique Games: a dictator (long-code) encoding of a good labeling reads off the correct coordinate unless noise hits it. It is a direct calculation from the local instance's value.

Sources

PaperOpen
Section
§A.2 The finite ordering transfer
Page
—
Label
—
Line
appendices/finite-consequences.tex:655
Leannot aligned

Phase 2 aligns this node to a Lean declaration.