claimroutineinferred_dep
Copy normalization to unit incident weight
Copy normalization gives unit incident weight with NO value ≤
equalizes left weights before rounding
Claim·§A.1 A simple strong-left form of Unique Games·inside lemUnique Games with constant left degree and strong satisfaction
Statement (paper)
Let be the set of copies, where each with has copies carrying local edge weights . Then and the incident weight at each copy is one. On a NO input the copy game's normalized value is at most . On a YES input there is a labeling of the copy game in which more than a fraction of has local satisfied weight at least .
In plain English
Replace each left vertex of positive weight by copies with local edge weights . The resulting copy set satisfies , each copy has incident weight one, the NO-case normalized value is at most , and in the YES case a labeling exists where more than a fraction of copies have local satisfied weight at least .
Why it's here
This turns the weighted game into one where every left vertex has the same total weight, which is needed before the weights can be rounded into an unweighted multigraph in the next claim. Flooring with the factor loses at most a constant factor in size and soundness, giving , which the parameter inequalities from the earlier claim absorb. In the YES case, the Unique Games Theorem's completeness is converted into a high fraction of copies that are almost fully satisfied, using the constants and .
Classified routine:
Standard vertex-duplication/weight-normalization bookkeeping (as in regularization reductions for Unique Games, e.g. Khot–Regev style left-regularization), combined with a Markov-type averaging of the UG theorem's completeness guarantee.
Uses
Used by
Sources
PaperOpen
- Section
- §A.1 A simple strong-left form of Unique Games
- Page
- —
- Label
—- Line
appendices/finite-consequences.tex:69
Leannot aligned
Phase 2 aligns this node to a Lean declaration.
Friction
- infoinferred_depoutline: "Copy normalization to unit incident weight" rests on claim-lem-simple-strong-left-1
- infoinferred_depoutline: "Copy normalization to unit incident weight" rests on thm:ugc