MathIdeasResearch in progressRepository ↗
← Research catalogOriginal Markdown ↓

Matching counts: literal schedule exclusion and bounded fragility reference

Family 113; selected manuscript/construction and formal-scope addendum. Source revision fd4aeeb2ee4fc729c18d98444fed42fd0529eeeb.

The FPRAS main statement and selected introduction, quantitative sampling guarantee, annealing/counting parameters, amplification and prescribed-degree corollary were read. The literal nonempty routine has S=ceil(10 sigma^-2(n+K+1)), with sigma<=1/(10^7D) and large enlarged-graph parameters. At two vertices and error 1/10, confidence failure 1/20, its general schedule requests about 4.52*10^48 fresh runs per stage. This is an exact declared-parameter calculation, not a runtime observation or impossibility for a different method. Empty/infeasible branches and practical special cases remain separate. The prescribed-degree/fixed-size sampler promises total-variation approximation, not exact or pointwise almost-uniform sampling.

The companion entropy introduction (lines 1–240) and selected deterministic algorithm (lines 1–155) were read. Entropy bounds require a feasible perfect-matching marginal vector, with general-graph odd cuts. The deterministic count guarantee uses factor 512^n, not fixed relative-error accuracy, and explicitly warns its polynomial degree is large. The selected family scope covers the FPRAS, entropy/count bounds and triangle-expansion relation, but excludes the sharp global face-dimension bound. No independent formal check was run.

Defer the literal production counter. Dossier 006 now describes a conventional bounded exact subset-DP reference, rank replay and single-edge failure counts. Its 93 finite checks do not verify the source FPRAS or buyer value. Full proof, sampler implementation and deletion reduction remain pending.

Exact source files read:

See source metadata, parameter calculations, and the exact utility.