Quantum code study ties 'magic' to word-set geometry
An arXiv preprint reports exact witness thresholds for selected code families, while identifying a major limit for larger ones.
Search
2,131 results
321–340
TailorCoPilot Linked to Higher Novice Success in Pattern Tests
Preprint: TailorCoPilot users finished garment-pattern revision tasks faster, with lower workload and higher expert-rated quality than baseline users.
AI Memory Stores Film Facts, But Struggles to Find Them
Preprint: LoRA adapters stored some film knowledge, but query-based retrieval could not reliably find the right memory.
Quantum model estimates tiny pair creation in binary stars
Preprint: A theoretical study derives formulas for scalar, fermionic and vector fields and applies them to an idealized weak-gravity binary-star model; its relativistic enhancement remains speculative.
Hybrid structural solver finds exact matches, but load limits remain
Preprint: The method joined reduced-order and full models in structural simulations, delivering exact results in some tests while losing accuracy under other loads.
Dark matter's clumpiness gap appears only at tiny scales
Preprint: Simulations found collisionless dark matter looked about twice as clumpy as self-interacting models around the most massive halos.
Tree-based method tackles homology of SL2 over localized integers
A preprint uses products of Bruhat–Tits trees and a staged calculation to study SL2(Z[1/n]) for general n.
Reacting-gas model matches reference simulations in key tests
An arXiv preprint reports an extended wave-particle method matching DSMC in cylinder and shock tests and modeling a reacting side jet over a blunt cone.
Multilingual Hallucination Detectors Show Modest Scores, Uncertain Ranks
This preprint reports on 623 submissions, finding middling scores, wide rank uncertainty and ranking differences linked to how test examples were built.
New ocean model targets global simulations without a global solve
An arXiv preprint estimates AC/DC at roughly 1.3 to 1.4 times hydrostatic runtime, while its deep-convection test exposed parameter sensitivity.
AI Model Reports Gains in Sign-Language Recognition and Spotting
An arXiv preprint reports benchmark results for a system that recognizes signs and estimates where they occur in continuous video.
Optimization preprint reports fast convergence, but tests remain limited
This preprint reports strong convergence rates for a specific class of constrained convex problems and compares the resulting algorithm with several numerical methods.
Preprint model points to better quantum phase sensing with repeated passes
A modeled interferometer study reports that rotated NOON outperformed other state families, with a proposed four-photon operating point.
Varying control signals can reveal hidden feedback networks
Preprint: Known changes in an input signal can separate feedback coupling from gain-invariant confounding, according to theory and simulations.
Math study finds uniform bounds for two-input operators on a complex sphere
This arXiv Preprint reports scale-independent bounds and a restricted-input smoothness threshold that falls from d+1 to d.
Preprint: A proof settles two colorful Helly conjectures
A theoretical arXiv study shows that regularity preservation fails for some multihomogeneous ideals under every monomial ordering, then proves face-count bounds for d-Leray complexes.
Rerunning AI agents often changes the outcome, not the failure
An arXiv preprint reports that controlled replay tests whether a multi-agent system was repaired by reusing the recorded prefix.
Physical AI benchmarks often rank models in similar ways
Preprint audit finds strong overlap among benchmark rankings and suggests that a smaller testing suite may retain much of the useful information.
Algebraic examples make a key reduction bound exact or off by one
Preprint: Constructed hypersurface families in every prime characteristic reach the bound exactly or fall one unit short.
Robot tests suggest a vegetation stiffness profile stays stable across heights
Preprint: A laboratory study of an artificial grass bush and a woody twig found that a distributed stiffness profile stayed consistent across push heights, while a single-number measure changed.