Nguyen Anh Tin
`Outcome-free protocol project for an observation-safe, matched-compute comparison of PC-PIMC-PUCT, SO-IS-PUCT, APB-PUCT and AR-SOIS-PUCT against a frozen neural-policy control under a locked custom asymmetric 9x10 Jieqi ruleset. APB tactical reranking is disabled in the primary comparison. The project contains the endpoint, resource caps, actor configuration hash, A/B compatibility receipt, power sensitivity plan, analysis plan, and source-disjointness gates. It contains no arena outcomes and does not authorize a superiority or training claim before the registered selection and confirmation gates pass