The Finite Symmetric Lovasz Local Lemma
Abstract
A finite family of rare dependent events can be avoided simultaneously under the exponential symmetric local-lemma criterion.
Let Omega and I be finite types, with decidable equality on I. A real weight w on Omega is nonnegative and sums to one. For a set E, write P(E) for the sum of w over E. Let A_i be the bad events and G a simple graph on I with decidable adjacency. JointIndependence(w,A,G) means that for every index i and finite set S disjoint from i and all its neighbors, the probability of A_i intersected with all A_j for j in S equals P(A_i) times the probability of the latter intersection. Pairwise independence alone is insufficient.
Theorem 1.1 (Positive Probability of Avoiding Every Bad Event).
Proof. Machine-checked in Lean as D5/S3/Combinatorics/Probability/FiniteLovaszLocalLemma.lovasz_local_lemma_symmetric_finite (✓ std3). ∎
Citation. ATLAS contributors; AutoformBot (2026). ATLAS finite symmetric Lovasz local lemma. URL: https://github.com/facebookresearch/atlas-lean/blob/0b121a198307b6153181f5a1d9145dcda2f7bfee/MathlibExt/Probability/Combinatorics/LovaszLocalLemma.lean.
Commentary.
For a nonnegative real p and natural d, suppose every bad event has probability at most p, the maximum degree of G is at most d, and the stated joint independence condition holds. If exp(1) p (d+1) is at most one, the event that all bad events fail has strictly positive probability. The conclusion includes the zero-degree boundary case. It asserts simultaneous avoidance and does not provide an efficient search algorithm.
References
- Truth anchor:
D5/S3/Combinatorics/Probability/FiniteLovaszLocalLemma.lovasz_local_lemma_symmetric_finite