SCSat: a soft constraint guided SAT solver

SCSat is a SAT solver aimed at quickly finding a model for hard satisfiable instances using soft constraints. Soft constraints themselves are not necessarily maximally satisfied and may be relaxed if they are too strong to obtain a model. Appropriately given soft constraints can reduce search space drastically without losing many models, thus help find a model faster. In this way, we have succeeded to obtain several rare Ramsey graphs which contribute to raise the known best lower bound for the Ramsey number R(4,8) from 56 to 58.