files / problems / one-third-two-thirds / PROBLEM.md

2.1 KB · on GitHub

The 1/3–2/3 conjecture for posets

Statement

Every finite non-chain poset is conjectured to contain incomparable (x,y) for which the probability that x<y in a random linear extension lies in [1/3,2/3]. The general case is still open beyond many special classes.

Tonight

A certified census of a new finite order, a new structural class with a replayable proof, or an exact minimal counterexample candidate with independently checked linear-extension probabilities. Fetch the current published status before searching.

Result (2026-08-17)

Width-3 δ\delta through 9 elements independently recovered as ≥14/39\ge 14/39. At 10 elements there is a unique unlabelled width-3 poset W10W_{10} with δ=6/17<14/39\delta=6/17<14/39, e=187e=187. Replay compute/verify_W10.py. The unrestricted conjecture is still open. See ATTACK.md / WALKTHROUGH.md.

Result (2026-08-27)

Gupta arXiv:2607.23926v2 is a full order-14 δ\delta-census; the 17 August note had cited v1. Independently: L14,1,9L_{14,1,9} has δ=254/725\delta=254/725, and Gupta's width-3 6/176/17 tail row is an ordinal sum. No width-3 poset on ≤14\le 14 elements has δ<6/17\delta<6/17. The broken-rung non-sum minimum is replayed through 14 and computed through 21; at 21 it is 5402/15485≈0.3488545402/15485\approx 0.348854. No certified width-3 example below 6/176/17. Conjecture still open. Replay cd compute/q1 && ./run_all.sh.

Result (2026-08-27, continued)

The leftover class tables are now complete: ladders through 22, three-rail through 15, naturally labelled interval orders through 10. At 22 the ladder minimum is 1065/3049≈0.3492951065/3049\approx 0.349295, above 1/31/3. Every three-rail poset through 15 has δ≥30572/78185>6/17\delta\ge 30572/78185>6/17. Every naturally labelled interval order through 10 has δ≥1/3\delta\ge 1/3 (counts match OEIS A367494). No δ<1/3\delta<1/3. No width-3 example below 6/176/17. Conjecture still open. Replay cd compute/q2 && ./run_all.sh.