Bee-Like Simulations Find Narrow Route to Gravity-Based Signals
Preprint: Direct pointing appeared in a middle range of simulated food conditions, while stable shifts to gravity-referenced signaling were found under selected settings.
Search
2,111 results
1081–1100
Study Finds Sparse Method Can Shrink Multimodal Search Costs
Preprint: PUMA often matched or exceeded dense retrieval across five benchmark datasets while using less storage and speeding exact search on large candidate pools.
ReRAM devices resist unwanted updates in in-memory AI study
Preprint: Array tests found no tendency to drift after 100,000 half-voltage pulses, while simulations reported successful convergence when k was below 0.01.
Holographic study finds different routes can carry the same flow
Preprint: Exact calculations show endpoint choices can change the route and sector split while leaving the total bottleneck flux unchanged.
Graph-based AI model reports 1.20% average cost gap in seven systems
Preprint: UNION reports low cost gaps and high feasibility in seven AC-OPF systems, but selected outages and a five-day temporal test narrow the evidence.
Geometric model traces entanglement through horizons and branes
Preprint: Analytic calculations use thread-like flows to test how entanglement is organized in models of curved space.
Coarse geometry offers tests for proper matrix-group actions
Preprint: A mathematical study gives sufficient checks for when subgroup actions behave properly on upper triangular matrix spaces.
Preprint finds local and renormalized alternatives to a failed global zeta
A mathematical study of the infinitely cusped tree lattice Yq derives an algebraic local zeta and a convergent finite part for its infinite tail.
New Number-Theory Estimate Allows Shorter Waring Intervals
Preprint: A uniform bound for short Weyl sums yields a sharper interval threshold for a restricted form of Waring’s problem.
Hybrid quantum model yields functions beyond determinants and permanents
Preprint: Models mixing fermionic and bosonic excitations yield generalized expressions, while a quasiparticle example goes beyond standard immanants.
Study maps when optimization solutions remain stable as data change
Preprint: An arXiv version-one manuscript sets conditions for stable values, solution sets, equilibria and finite-horizon policies under sequential model changes.
Preprint maps how DIS predictions handle scales and heavy quarks
Preprint: The methods notes explain how inclusive DIS calculations handle PDFs, scales and heavy quarks, while leaving important higher-order uncertainties unresolved.
New algebraic view of Collatz map leaves key question open
Preprint: A formal Boolean-polynomial construction derives carry formulas for an alternative Collatz map, but does not settle convergence for all positive integers.
New Count Maps the Growth of Minimally Transitive Groups
An arXiv preprint finds an explicit upper bound and matching n log(n)-scale growth along prime-power degrees, while exact constants remain open.
Enterprise Dockerfiles show widespread security and upkeep gaps
Preprint analysis finds widespread configuration warnings and a large gap between typical files and stronger internal examples.
Adaptive federated learning outscored a standard baseline in audio test
A Preprint study reports stronger pFedMARL scores across simulated non-IID clients, while a centralized model remained well ahead.
Extra √T Gate Linked to Lower Modeled Single-Qubit Synthesis Costs
Preprint: A study reports lower modeled costs for fine-grained single-qubit synthesis when the logical √T primitive is included.
Antarctic Micrometeorites Point to a Mixed Solar System Past
Preprint: Analysis of eight UCAMMs found three organic-matter types and mineral phases that may preserve histories from different parts of the early Solar System.
Study reports lower GMRES counts for fractional PDE systems
Preprint: A GLT-guided block method is tested on a one-dimensional variable-coefficient problem, with iteration counts that stabilize across displayed matrix sizes.
Robot policy scores higher with refreshed tactile feedback
Preprint: TacForcing reported 65% average success in simulation and 69% in real-world tests, without uncertainty estimates.