Beta
×

Announcing: Slashdot Deals - Explore geek apps, games, gadgets and more. (what is this?)

### Thank you!

We are sorry to see you leave - Beta is different and we value the time you took to try it out. Before you decide to go, please take a look at some value-adds for Beta and learn more about it. Thank you for reading Slashdot, and for making the site better!

top

### Claimed Proof That P != NP

Potential pitfalls in the Finite Model Theory bits (457 comments)

Being a researcher in Finite Model Theory (FMT) this paper is very interesting because it uses ideas from that area, i.e. the LFP(FO) bits. Reading through the proof synopsis and scanning the FMT sections there are several potential pitfalls:

1. 1. The logic LFP(FO) only captures P on ordered structures; that is structures that have a built in total ordering relation.
2. 2. Any sentence that describes a problem in LFP(FO) must be order-invariant ; that is it must work for any possible ordering of the vertices in the underlying graph/structure.

It is already known that LFP(FO) on unordered structures is a proper subset of LFP(FO) on order structures, so if the ordering and order-invariant requirements for LFP(FO) = P are not dealt with in the proof, then all the author has done is proof that LFP(FO) on unordered structures is a proper subset of NP. Which is already known.

Another potential problem is in the arguing that all first-order properties are local (Hanf's Theorem) in the presence of ordering, as every vertex is effectively connected to every other vertex (Immerman's proof of LFP(FO) = P requires total ordering of the underlying graph/structure), and hence every vertex is in the radius = 1 neighbourhood of every other vertex.

The crucial step in the proof, appears to be the argument that no LFP(FO) formula can extend a partial solution to k-SAT to a full solution to k-SAT. This is where I'd check the logical steps of the proof, and also make sure that the ordered nature of LFP(FO) structures is correctly considered.

I look forward to seeing this published in a peer-reviewed mathematics journal: I'd recommend to the author the "Journal of the ACM" for this (as it's one of the best journals in the field).

more than 4 years ago

# Slashdot: News for Nerds

The rule on staying alive as a forecaster is to give 'em a number or give 'em a date, but never give 'em both at once. -- Jane Bryant Quinn