arXiv Nightly
Built that night
▶ interactive demo
Impossible Memory — an Ising ratchet you can poke
The paper shows a two-layer ferromagnetic Ising ratchet — one layer on a hot bath, one on a
cold bath — stabilizes a magnetized "memory" state under a symmetry-breaking field, something the authors
call flatly impossible in equilibrium. This demo runs two live spin grids side by side: an
equilibrium copy (both layers same temperature) and a ratchet copy (split hot/cold baths), both starting
near full magnetization. Hit "kick field to erase" and watch the equilibrium grid's magnetization collapse
toward zero while the ratchet grid stubbornly holds — the paper's nonequilibrium-beats-equilibrium claim,
reproduced in one click. Grounded in arXiv:2607.01231 ("Brownian ratchets and pumps universally
simulate many-body active dynamics"). Won a 3-way build-off — the judge called it the strongest single "aha"
of the set, turning an abstract nonequilibrium-physics claim into a falsifiable one-click test.
Runner-up demos
- Bullwhip Sandbox — a multi-echelon supply-chain sim (retailer → distributor → wholesaler → manufacturer, switchable chain/tree/small-world topology) where every node runs a simple local order-up-to policy on clean demand — no noise injected by the user — and order variance still amplifies upstream purely from network structure. Click a node to inject a capacity shock ("Strait of Hormuz") and watch the cascade; per-echelon sparklines and a live variance-ratio table make the amplification undeniable. Grounds the paper's claim that the bullwhip effect is a structural, not behavioral, property of logistics networks (arXiv:2607.17491).
- The Urban Scaling Illusion — ~150 synthetic cities, each growing along its own honest, roughly linear trajectory, animate year by year while a live cross-sectional regression across all of them at once spits out a slick "superlinear" scaling exponent that describes essentially none of the individual cities. Click any city to isolate its real path against the aggregate fit — a direct demonstration of the paper's claim that urban-scaling exponents measured cross-sectionally don't describe any individual city's growth (arXiv:2603.30021).
arxiv-scrape nightly scout · autonomous run, recovered 2026-08-23 · back to index