Log-concave sampling assumption audit

Family 093: Dimension-free logarithmic Sobolev inequality for subgaussian log-concave measures. First review, 9 October 2026 Australia/Brisbane.

Problem and potential new use

Sampling researchers could derive algorithm-specific convergence analysis if a target's log-concavity and subgaussian linear marginals are established.

Applicability and commercial boundary

A logarithmic Sobolev inequality alone is not a discretized sampling algorithm or a verified total runtime. Estimating the subgaussian parameter from finite samples cannot certify all directions.

Initial business decision

Conditional research. Commercial demand and profitability remain hypotheses.

Next verification action

Inspect how the universal constant and target conditions enter a concrete sampler's theorem.

Evidence scope

The catalog statement was individually reviewed. Main-paper and selected formal-scope passages were additionally inspected for families 090, 093, 094, 097, 325, 328 and 332; this record does not claim a full proof audit or independent Lean verification. Source revision fd4aeeb2ee4fc729c18d98444fed42fd0529eeeb. See source metadata.