Maximin Spacing — beat the optimal gap-packer

Scatter N+1 random points on a line, pick M of them, and try to maximize the smallest gap between your picks. That's exactly the same problem as grouping the N gaps between points into M contiguous blocks and maximizing the smallest block sum — the paper shows this is equivalent to counting reset-cycles of a threshold-resetting random walk, and derives exact distributional identities for the optimum. Click the tick marks below the line to place M−1 dividers, then reveal the provably optimal packing. Grounded in arXiv:2606.04837 (Cunden, Cuppone, Gramegna, Vivo).
click a tick to place / remove a divider
Dividers placed: 0 / 0
Your bottleneck
—
Optimal bottleneck
—
You reached
—
Distribution of the optimal bottleneck over random trials0 trials
N+1 random points on the line → N i.i.d. gaps between consecutive points.
Pick M points (M−1 internal dividers) to maximize the smallest resulting gap.
Place M−1 dividers
Games played: 0 · avg score: —
The greedy optimum: binary-search a target value V, then walk the gaps left→right accumulating a running sum and cutting whenever it reaches V — the largest V for which this yields ≥ M blocks is provably the max-min block sum. That's the same accumulate-and-reset logic as the paper's random walk: P(optimal spacing ≥ V) = P(walk completes ≥ M reset cycles in N steps).

Try shrinking M relative to N — the optimal bottleneck grows fast, since you get to merge many small gaps into each block.