Preprint reports a 260-fold speedup for simulated active-IRS beamforming
In synthetic blocked-link simulations, a rate-profile learning approach had a higher reported sum rate than WMMSE, but the result has not been tested on hardware.
Search
2,111 results
1761–1780
Preprint finds dense graphs have no nonsynchronized Kuramoto energy minima
A mathematical proof says that, for some unspecified positive η, every finite simple graph with minimum degree at least (3/4 − η)n has fully synchronized states as its only local energy minima.
Preprint: Time-series model posts stronger forecasts with fewer parameters
SATS reported leading scores on zero-shot and in-distribution tests, but the analysis did not include confidence intervals or repeated runs.
Preprint proves a degree threshold guarantees disjoint cycles of different lengths
The abstract theorem applies to finite simple directed graphs and extends to weighted cycles, but it does not provide the smallest or an optimized threshold.
Preprint reports wide-field infrared imaging without direct infrared detection
A benchtop experiment used visible interference and a CMOS camera to reconstruct images while an infrared beam passed through a test mask.
Preprint finds preference tuning can push flow-based image models off their data manifold
ThermoDPO-weighted improved reported toy and image-generation scores, but the evidence comes from offline benchmarks and a one-author human evaluation.
Preprint: Simulation reports abrupt cooperation collapse as defection becomes more tempting
A lattice-based model links reputation-weighted learning with sharp shifts, competing outcomes and the growth or extinction of cooperative clusters.
Preprint Reports Gains From Randomized Reservoir Ensembles in Chaotic-System Tests
Hénon and Lorenz 96 tests showed useful ensemble forecasts, while a folded-towel-map result focused on forecast geometry.
Preprint reports spin-torque signal in largely disordered YPtBi films
The response persisted through 400 °C, while chemical mixing at the upper interface emerged as a stronger clue than bulk crystallization.
Preprint reports stronger simultaneous approximations for selected number pairs
The proof-based study constructs common denominators and reaches classical and p-adic Littlewood conclusions only under stated mathematical conditions.
Preprint links faster changes in stock co-movement to a volatility premium
A monthly S&P 500 analysis found the association in implied variance before broader realized volatility, but found no timing advantage or crash protection.
Preprint: AI models were close forecasting heat demand in two German networks
Chronos-2 had lower overall error, while TabPFN-TS produced uncertainty intervals closer to their 80% target.
A mathematical threshold links stability to positive-curvature metrics
Preprint says a shifted stability measure can identify positive scalar curvature—and, above a key boundary, a unique constant-curvature Kähler metric.
Preprint: AI mapping finds just 10% of Nice’s suburban street network meets quality criteria
Street View images were used to compare vegetation, sidewalks and pedestrian-accessible entrances across 110.9 km of mapped streets.
Preprint reports fourfold higher single-atom detection with a nanoscale lens
A GaN waveguide was modeled and tested as a near-field optical lens, but the result comes from one alignment-sensitive laboratory setup.
Computer model handles changing-stiffness plates, but distorted meshes trip some versions
Across five computational cases, two virtual-element formulations retained accuracy, while standard stabilization failed in some distorted-mesh tests.
Preprint finds sharp thermodynamic peak in hot gluon matter
A pressure-constrained model follows lattice behavior across deconfinement but differs from independent specific-heat estimates around the peak.
Preprint finds one graph measure is never more than twice another
The result links broadcast domination and multipacking, and gives a factor-two approximation for finding a large multipacking.
Preprint Shows How a Curvature Flow Can Select One Shape at Infinity
A mathematical theorem says continuous negative boundary data determine a unique smooth convex solution, while a normalized inverse curvature flow converges to it under stated conditions.
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.