NLocalSAT: Boosting Local Search with Solution Prediction 2020-05-02

NLocalSAT: Boosting Local Search with Solution Prediction

Abstract

NLocalSAT ：address stochastic local search (SLS) 随机初始化的问题

NLocalSAT combines SLS with a solution prediction model, which boosts SLS by changing initialization assignments with a neural network.
NLocalSAT是使用神经网络增强SLS求解器的一种方法，离线方法, 使用神经网络来增强SAT求解器。
（NeuroCore在线预测，会给CDCL带来大量开销。而NLocalSAT以离线方式使用预测。 在这种方法中，对于每个SAT问题，只计算一次神经网络。）

Experiment on five SLS solvers (CCAnr, Sparrow, CPSparrow, YalSAT, and probSAT)
Experimental results show that NLocalSAT significantly increases the number of problems solved and decreases the solving time for hard problems.

Introduction

state-of-the-art SAT solvers :
CD