The computational hardness threshold conjecture for weighted independent sets

About 19 years old · traced to

Let d≥3d\geq 3 be the maximum degree, let ℓ\ell be the activity, and let ℓc(d)=(d−1)d−1/(d−2)d\ell_c(d)=(d-1)^{d-1}/(d-2)^d be the critical activity for decay of correlations on the infinite dd-regular tree. Consider the weighted sum of independent sets, with each independent set II having weight ℓ∣I∣\ell^{|I|}. Computational hardness threshold conjecture. For every d>3d>3 and all ℓ>ℓc(d)\ell>\ell_c(d), unless RP=NPRP=NP, there does not exist a fully polynomial approximation scheme for counting weighted independent sets in graphs of maximum degree dd with activity ℓ\ell. This conjecture identifies the computational threshold with the uniqueness threshold; below ℓc(d)\ell_c(d) a fully polynomial approximation scheme is known, while hardness above the threshold remains open.

References

Primary source

Elchanan Mossel, Dror Weitz and Nicholas Wormald, “On the hardness of sampling independent sets beyond the tree threshold”, arXiv:math/0701471 (2007).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.