Solutions to the Problem of K-SAT / K-COL Phase Transition Location

TL;DR

This paper extends concentration inequalities to more general cases, including branching random walks, and applies these results to solve longstanding open problems related to phase transition locations in K-SAT and K-COL problems.

Contribution

It introduces generalized concentration inequalities that relax previous conditions and applies them to analyze phase transitions in complex combinatorial problems.

Findings

Extended concentration inequalities to branching random walks.

Solved open problems on K-SAT and K-COL phase transition locations.

Abstract

As general theories, currently there are concentration inequalities (of random walk) only for the cases of independence and martingale differences. In this paper, the concentration inequalities are extended to more general situations. In terms of the theory presented in the paper, the condition of independence is $\frac { \partial y } {\partial t}=$ constant and martingale difference's is $\frac { \partial y } {\partial t} = 0 $. This paper relaxes these conditions to $\frac { \partial^2 y } {\partial u_i \partial t} \le L$; i.e. $\frac { \partial y } {\partial t}$ can vary. Further, the concentration inequalities are extended to branching random walk, the applications of which solve some long standing open problems, including the well known problems of K-SAT and K-COL phase transition locations, among others.