Novel Atomic-Set Computation Based On SAT Solving

[h3ssto]

Recently, we evaluated the performance of our novel atomic-set computation algorithm [1] against the one implemented in FeatJar. We found that FeatJar, on average, is about 40x slower. Now we are curious to understand whether this difference stems from the algorithm or simply FeatJar's implementation in Java and its use of SAT4J.

As our algorithm should be easy to implement, we are hopeful that this question can be answered without much effort.

[1] https://arxiv.org/abs/2501.12490