层级只比较下界数值,不表示方法普遍给紧结果。one-sided LP 会忘掉消息顺序、round 与信息泄露;这也解释了它为何与Set Disjointness 的信息复杂度证明公理库Set Disjointness 的信息复杂度Information complexity of Set Disjointness · Information-statistics lower bound for DISJ以单坐标 AND 的条件信息和 transcript Hellinger 距离证明 Set Disjointness 的线性信息下界。形成方法对照,后者直接追踪 transcript 泄露与 Hellinger 距离。自然定义到 LP 定义的 转换也必须保留;若某证明只检验一族几何矩形而非全部组合矩形,不能作为本页可行 dual 证书。
参考资料
Rahul Jain and Hartmut Klauck, “The Partition Bound for Classical Communication Complexity and Query Complexity,” Proceedings of CCC, 2010, pp. 247–258.
Hartmut Klauck, “A Strong Direct Product Theorem for Disjointness,” Proceedings of STOC, 2010, pp. 77–86.
Amit Chakrabarti, Ranganath Kondapally, and Zhenghui Wang, “Information Complexity versus Corruption and Applications to Orthogonality and Gap-Hamming,” Proceedings of RANDOM, 2012, pp. 483–496.