sabaridsnfuji/repro-certificate-guided-pruning-for-stochastic-lipschitz-optimization

Fix README frontmatter tags; add real source code, scripts, and results.json (single source of truth) backing the published logbook pages

Real CGP implementation (confidence-adjusted Lipschitz envelope, Shrinkage Theorem components, adversarial lower-bound instance, CGP-Adaptive doubling, CGP-TR trust regions). 4/6 verified, 2/6 toy (correct qualitative direction on sample-complexity scaling, exact exponent not tightly reproduced at this compute scale). No official repo exists; independent reimplementation from extracted paper formulas.

Update logbook: Certificate-Guided Pruning

initial commit