Independent science reporting
Fresh papers in plain English, with results, evidence, and limitations kept in context.
Browse the latestPublished by the desk
Lead story
A study of BL Lacertae found significant gamma-ray changes on hour-long timescales, while evidence for faster variability remained marginal and model-dependent.
Scientific publication4 min read
A theoretical study of Schur-twisted AdS5 × S5 reports agreement between bulk and boundary calculations in both perturbative ensembles, while showing that their nonperturbative expansions are organized differently.
Scientific publication4 min read
An arXiv preprint reports that currents generated by surface acoustic waves changed near charge-density-wave transitions in NbSe3 and 2H-TaSe2. In NbSe3, reversing wave direction reversed signal polarity, while metallic tantalum controls lacked the same temperature modulation.
Scientific publication4 min read
A computational methods preprint reports robust conditioning for a two-level solver in high-contrast multiscale benchmarks, with a much smaller coarse space in selected time-step regimes.
Scientific publication4 min read
An arXiv preprint reports higher math and code benchmark averages for a reward-aligned trajectory filter, with total training time close to standard on-policy distillation.
Scientific publication4 min read
A theoretical preprint finds that the Tcc(3875)+ magnetic moment could differ sharply between compact and molecular models, but says the measure cannot distinguish every state.
Scientific publication3 min read
A three-dimensional neural network inferred averaged cosmological quantities from present-day simulated matter maps, including parameters linked to curvature and backreaction. Its predictions tracked the simulations closely, but the study is only a proof of principle because the training data were too simplified for sensible use on observations.
Scientific publication4 min read
A new arXiv preprint reports a proof that every pair of finite tournaments has overall discrepancy at least an absolute constant times n to the three-halves power. Its proof combines skew-symmetric tournament matrices, random relabellings and a local fluctuation bound.
Scientific publication3 min read
A preprint presents AERA, a controller that estimates whether more reasoning may help and reports near-baseline GSM8K accuracy with sharply lower generation use.
Scientific publication5 min read
A preprint reports that BERT had the strongest overall scores for assigning cyber threat intelligence events to sectors, with a caveat about unequal sector coverage.
Scientific publication4 min read
The analysis gives different corruption ceilings for Massart and oblivious noise, then translates them into per-iteration sampling requirements and simulation checks.
Scientific publication4 min read
CMGR outperformed three registration methods on right coronary artery motion in XCAT simulations and selected clinical pairs, while whole-volume results were less consistent.
Scientific publication5 min read
An open-source harness for long-horizon coding agents scored above selected systems on two benchmarks, but the study did not isolate the harness from model, prompt, tool and implementation differences.
Scientific publication4 min read
GAAT posted leading scores in several drone-imaging tests, including detection and segmentation, but competitors remained ahead on other reported measures.
Scientific publication4 min read
A mathematical preprint sets a vertex limit for minimal forbidden graphs, identifies the complete forbidden set at a threshold of 3, and narrows the shapes allowed by forests and short cycles.
Scientific publication4 min read
A numerical study tests whether an additional monopole coupling can keep key mass ratios stable as a compact U(1) lattice model is tuned. The results fit a possible continuum approach, but do not establish that one exists.
Scientific publication4 min read
A theoretical study maps how a single photon carries magnetic-field information after scattering and finds that frequency-resolved counting can approach the model's full quantum-information benchmark under selected conditions.
Scientific publication5 min read
A calibrated model finds that replacing score-based competition with random assignment raises completed fertility in the model, but the policy result is not a direct experimental estimate.
Scientific publication5 min read
A test-suite-free framework built around an AI model produced C candidates that often compiled and matched held-out input-output checks, but the benchmark also exposed false acceptances and weaker results on stripped binaries.
Scientific publication4 min read
A method called DA3PO was reported to outperform GRPO, DAPO and GSPO on mathematical reasoning benchmarks, though the tests covered only two Qwen3 base models and did not report uncertainty estimates.
Scientific publication4 min read
A preprint describes an AI system that retrieves evidence before choosing how its agents collaborate. It reports higher scores across seven reasoning benchmarks, but the evidence comes from in-silico evaluations rather than human users or real-world deployments.
Scientific publication4 min read
A conceptual note argues that Monte Carlo Tree Search can be understood as every-visit Monte Carlo control when the comparison is limited to trajectory sampling and Monte Carlo action-value updating. The claim clarifies terminology and computational scope, but does not compare performance.
Scientific publication4 min read
A theoretical preprint offers a direct Haar-based analysis of randomized quasi-Monte Carlo integration and derives an expected squared-error bound under specified function assumptions, but reports no experiments.
Scientific publication4 min read
A secondary analysis of four-person conversations reported that combining speech intensity, gaze and perceived interpersonal closeness distinguished gaps from overlaps better than gaze alone, including in noisy conditions.
Scientific publication4 min read
HubMixer sends recommendation features through compact learned hubs before mixing them and writing them back to individual tokens. The preprint reports the best offline AUC on four objectives, fewer parameters than RankMixer and TokenMixer, and a 5.48% online conversion improvement before full deployment in the tested business.
Scientific publication4 min read
A preprint reports that GOD linked browser commands to execution records, replay checks and portable packs in 15 completed run slots. The evaluation focused on command, replay and artifact checks, not human or social validity.
Scientific publication4 min read
A new arXiv preprint identifies the almost-sure Hausdorff dimension of the extreme points of a random convex shape and proves the matching critical measure is finite, while leaving positivity unresolved.
Scientific publication4 min read
Projected q-shells retained more smooth-function information and produced lower one-electron and molecular benchmark errors than standard nested gausslets in the reported calculations.
Scientific publication5 min read
A new arXiv preprint reports that a coding-inspired system kept average benchmark performance close to the uncompressed setup while using 25% of the visual-token budget on Qwen3-VL-8B.
Scientific publication3 min read
A theoretical and synthetic analysis finds that conditional flow matching can replace endpoint likelihood calculations only when specific residual errors cancel. In on-policy training, the shortcut may still improve reward even when its likelihood ratios are far from exact.
Scientific publication5 min read
Showing 30 of 1925 articles
Research fields
Start broad, then move into one of thirty focused subject desks.
Clinical care, population health, health systems, nutrition, and metabolism.
Cells, genes, microbes, organisms, ecosystems, and evolution.
Climate, atmosphere, oceans, water, geology, and the living environment.
Astronomy, quantum science, particles, matter, light, and energy.
Artificial intelligence, computing, robotics, energy, and engineering.
Psychology, cognition, society, culture, education, and work.
Molecules, reactions, materials, nanotechnology, and sustainable chemistry.
Mathematics, probability, statistics, data science, and modeling.
Evidence synthesis, scientific practice, integrity, and interdisciplinary work.
How it works
We trace the central claims to the source, check the numbers, and make limitations visible before publication.
Read the methodology