Theory of Computing ------------------- Title : Optimal Composition Theorem for Randomized Query Complexity Authors : Dmytro Gavinsky, Troy Lee, Miklos Santha, and Swagato Sanyal Volume : 19 Number : 9 Pages : 1-35 URL : https://theoryofcomputing.org/articles/v019a009 Abstract -------- For any relation f \subseteq \{0,1\}^n \times S and any partial Boolean function g defined on a subset of {0,1}^m, we show that R_{1/3}(f\circ g^n)\in\Omega(R_{4/9}(f)\cdot\sqrt{R_{1/3}(g)}), where \R_\epsilon(\cdot) stands for the _bounded-error randomized query complexity_ with error at most \epsilon, and f\circ g^n\subseteq (\{0,1\}^m)^n \times S denotes the composition of f with n instances of g. We show that the new composition theorem is optimal for the general case of relational problems: A relation f_0 and a partial Boolean function g_0 are constructed, such that R_{4/9}(f_0)\in\Theta(\sqrt{n}), R_{1/3}(g_0)\in\Theta(n) and R_{1/3}(f_0\circ g_0^n)\in\Theta(n). The theorem is proved via introducing a new complexity measure, _max-conflict complexity_, denoted by \bar\chi(\cdot). Its investigation shows that \bar\chi(g)\in\Omega(\sqrt{R_{1/3}(g)}) for any partial Boolean function g and R_{1/3}(f \circ g^n)\in\Omega(R_{4/9}(f)\cdot\bar\chi(g)) for any relation f, which readily implies the composition statement. It is further shown that \bar\chi(g)$ is always at least as large as the _sabotage complexity_ of g (introduced by Ben-David and Kothari in 2016). -------------------- A preliminary version of this paper appeared in the Proceedings of the 46th International Colloquium on Automata, Languages, and Programming (ICALP'19).