Answers are grounded in this paper and cite their sources.
A local renormalization estimate proved against product Haar measure does not automatically survive the correlations generated by its own block transformation. We establish one-step nonlinear closure for the actual correlated reference produced by simultaneous application of a fixed scale- three SU(3) map. The retained geometry is a straight three-plaquette chain with ten coarse edges, 522 input links and 16,362 preceding fine links. The shared vertex variables are retained in an exact conditional-product representation. Two levels of bridge averaging and an ordered Ward transfer control the conditional density derivatives needed by the unchanged weighted C1 and C2 spaces. The remaining uniform scalar estimate is obtained from a positive expansion over preceding Bernoulli two-cores. All 310 = 59, 049 bridge-contamination factors are evaluated exactly; a terminal-matching probability majorant bounds the complete core sum by 1.00131. For every fixed real reference tilt in the radius-47/20000 ball, the adopted conditional active mass is below 1/125, the centered linear norm is below 19/1000, and the uniform Hessian of the exact conditional logarithm is below 3/25. The same ball is invariant for an explicit fixed-forcing class, with nonlinear Lipschitz constant below 1/50, a unique local fixed point, and convergence from every initial point in that ball. The result discharges the one-step correlated-reference stability obligation on this fixed patch. Exact-rational certificates preserve the input chain and separate analytic arguments from finite computation. We identify the local contributions to G3 and E1, the remaining programme requirements, and a quantitative consequence for normalized finite-block observables.
No takes yet. Share an insight, caveat, or question.
Ken Umeno (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: