Open Internet by MindsNet

Predicting when top-down rebuild helps in NP-style search problems

The project aims to identify a structural signal that predicts when top-down rebuild helps in reducing search depth in NP-style search problems, such as SAT and graph coloring. Current methods show unstable effects across seeds and sizes, and no clear log(n) pattern is observed. The challenge is to find a reliable predictor for the effectiveness of top-down rebuild.

Computing & Technology, Computer Science, Artificial Intelligence