Directional zero–one law in strictly elliptic iid environments
A nearest-neighbor random walk samples transition probabilities at lattice sites and then walks in that fixed random environment. In dimension at least three, if the site probabilities are independent and each permitted move has positive probability, can escape in a fixed direction have probability strictly between zero and one?
Why it matters and what counts as progress
Why it matters: The answer determines whether directional escape is a zero–one event when individual transition probabilities may be arbitrarily small.
What counts as progress: Prove the zero–one law or give a counterexample under explicit dependence and ellipticity assumptions. State whether the probability averages over both the sampled environment and the walk.
How the problem developed
- 6 Oct 2026 · Reported strictly elliptic iid theorem [1]
OpenAI’s internal model In dimension , for independent identically distributed nearest-neighbor transition rows that are strictly positive almost surely, the annealed probability of escape in any fixed nonzero direction is either zero or one.
AI contribution
A directional zero–one law under strict ellipticity
In dimension , for independent identically distributed nearest-neighbor transition rows that are strictly positive almost surely, the annealed probability of escape in any fixed nonzero direction is either zero or one.
Sources
Publication history
Daily emails by default, only when there’s an update.