reversedPropagationGuidedRelax is not correctly implemented.

I have been playing around with Propagation Guided LNS and noticed that reversedPropagationGuidedRelax in PropagationGuidedRelax.scala is not correctly implemented.

In fact, the current implementation does nothing except randomly freeze a bunch of variables without taking propagation into account. Specifically, the closeness metric is never updated, which means that the line subset = closeness.getCloseSubset(next, 10) always returns an empty set. One can also note that no propagation ever happens in this implementation, so it cannot possibly be propagation guided…

‌

Maybe I will find time to fix this but would appreciate it if someone could take care of it.

‌

Comments (2)