Preprint Finds Narrow Routes to Fair Sharing as Items Arrive
A theoretical study identifies online TEF1 cases and sets a tight 1/n ceiling on the temporal maximin-share guarantee for additive goods.
Search
2,451 results
2141–2160
Preprint links native defects to α-MnO₂’s electronic and optical behavior
Calculations point to strong donor–acceptor compensation and direction-dependent light response, but not to a settled carrier type.
Preprint says strict conditions force a shrinking Ricci soliton into a flat-and-spherical product
A theorem for a narrowly defined class concludes that the space is isometric to R² × Sⁿ⁻²; the work is a mathematical proof, not an empirical study.
Preprint maps eight possible forms of agency for advanced AI
A conceptual framework separates legal questions from moral ones without establishing that any AI system qualifies as either.
Preprint reports less forgetting in a lightweight AI test of changing art styles
FedCurv-DR posted higher accuracy and better retention than FedAvg on a simulated WikiArt benchmark, but the clients were synthetic and the fairness findings were inconclusive.
Preprint reports conditional stability for delayed force-feedback robot control
A mathematical analysis and computer simulation explore a wave-based link for nonlinear master and slave manipulators, but provide no evidence from human operators or physical robots.
Preprint finds an exact boundary for critical graph families
For bipartite H and k at least 5, finiteness occurs exactly when H is 2P2-free; the study also gives finite subclasses, infinite constructions and an algorithmic consequence.
Preprint proposes a symmetry-agnostic test for stellarator confinement
The approach looks for a closed orbit-selected drift surface rather than constant local action, with model and trajectory tests supporting the idea in selected cases.
Preprint finds no single winner as time-series data shifts
DQeND led in incremental tests and abrupt location drift, while ELECTS performed best when class composition changed abruptly.
Preprint uses quantum-state geometry to bound the entangled part of a mixed state
An analytic study links the Best Separable Approximation to radial distances from the maximally mixed state and derives a separate result for two-qubit systems.
Preprint sets out a unique grid-based way to break down bipartite graphs
The formal framework divides each graph in scope into atomic blocks and links the pieces to maximum matchings, independent sets and a characteristic called m.
Preprint finds exact one-third threshold for directed cycles through every vertex
For finite oriented graphs, a rounded-up one-third minimum semidegree guarantees a directed cycle of length 3q through each vertex once q ≥ 2 and the graph meets an explicit size cutoff.
Preprint links stiff molecular constraints to different vibrations and plastic modes in model glasses
In matched simulations, the molecular model was stiffer and its plastic modes coupled more strongly to volume-changing deformation; a proposed correction narrowed several gaps.
Preprint reports stable liquid–vapor interfaces in a mathematical fluid model
Under small, constrained disturbances, a one-dimensional periodic model converges to a selected two-interface phase-transition state.
Preprint reports a way to tune large AI models from smaller test runs
A two-step method transferred learning-rate settings across mixture-of-experts models and extrapolated them to a much longer training horizon, but the final choice was not exhaustively tested.
Preprint: Restricted-Visibility AI Societies Outperformed Their Globally Visible Twins
In a synthetic test, matched systems that saw only assigned evidence generally solved unseen combinations more accurately, but the full preregistered battery narrowly missed its target.
Preprint reports near-verbatim recovery of hidden AI reasoning through tool calls
EchoCoT reached a 66.4% near-verbatim success rate on one accessible model, while proprietary-model tests relied mainly on reported lengths and summaries.
Correlated Catalysts Do Not Remove All Entanglement Limits, Preprint Reports
A mathematical analysis builds a hierarchy of constraints and finds a strict gap between distillation and cost for one explicit quantum state.
Preprint reports faster camera localization using gravity and image-feature geometry
Two proposed solvers estimate a camera’s position, orientation and unknown focal length from sparse feature information, with favorable results on synthetic tests and visual-localization benchmarks.
Preprint proposes a more stable way to rebuild signals from scattered samples
A nested sampling framework relaxes exact polynomial quadrature, then uses canonical inversion to improve reconstruction in sphere and torus tests.