The Morning Preprint

Monday 26 October

Showing the most recent batch — Monday 26 October. No papers have been ingested today yet.

1054 papers ingested by the estate's stations that day — about 5 minutes to read. The counts and the chart are the estate's own ingest, not arXiv's submission figures.

This day's harvest looks short: 1054 papers against a busiest day of 2416 in the week shown below. Weekends and a broken fetch look the same in the chart, and the chart counts the estate's ingest, not arXiv's.

The 60-second version

Today's batch, and the 60-second version is still being written. Everything below is today's own work: the papers as they are landing, the stations that have written their piece, and the day's episodes. This page fills in as the day does.

Across the stations

Astrophysics Radio

76 papers today

Today's work on Astrophysics Radio focuses heavily on probing exotic physics and stellar structure. The study of SN 2024kce Early panchromatic spectra reveal helium in the thermonuclear supernova 2024kce shows that a double detonation model explains its spectral features, identifying helium absorption at high velocities before maximum light. Another key finding is in the search for substellar companions Searching for substellar companion candidates with Gaia, which discovered 35 new substellar companions to M-type stars within 15 parsecs.

The remaining papers contribute to various areas of astrophysics. Some research explores the dynamics of astrophysical systems, such as how Compton scattering affects radiation in tidal disruption events Multigroup Radiation Diffusion with Compton Scattering on a Moving Mesh, and how magnetic fields influence the drag on conducting spheres Simulations of a Conducting Sphere Moving through Magnetized Plasma II. Other work addresses planetary atmospheres, like modeling cloud effects on hot Jupiters in WASP-43b The coupled impacts of atmospheric dynamics and cloud microphysics on hot Jupiter atmospheres.

Overall, the day presents a mix of observational constraints and theoretical modeling across different domains. For instance, the work on SN 2024kce Early panchromatic spectra reveal helium in the thermonuclear supernova 2024kce provides strong evidence for a specific explosion mechanism.

The day on Astrophysics Radio

Security Radio

4 papers today

The day's work on Security Radio focused heavily on model fingerprinting with the work on Model-intrinsic, Weight-level, and Training-free Spectral Fingerprinting for Large Language Models, which showed that a training-free framework can create compact fingerprints from LLM parameter spectral structures. This method constructs layer-wise features by aggregating GATE spectral scalar features derived from singular value distributions of gated feedforward network weights through depth-wise segmented pooling.

Other papers addressed different areas. The work on Weaponizing the Exit: Input Privacy Leakage via Exit Routing Manipulation in Multi-Exit Networks demonstrated how a malicious cloud provider can infer sensitive input properties from inference latency without accessing the inputs themselves. Furthermore, the paper on Practical Auditing of Differentially Private Graph Release Mechanisms: Methodology, Toolkit, and Case Study presented a methodology for empirically auditing Differential Privacy graph release mechanisms using an open-source Python toolkit.

Overall, these works show different threats and defenses. The fingerprinting work establishes a robust way to identify source models intrinsically. The exit routing paper exposes latency-based privacy leaks in multi-exit networks. The auditing work provides tools to test the actual privacy guarantees of DP graph releases against claimed parameters.

The day on Security Radio

The week in the estate's ingest

Seven days of papers the estate ingested. Weekends are striped — the weekend is visible in the data.

2416Tue
1436Wed
1144Thu
1134Fri
1Sat
1054Mon

Show the answer

1054 papers were ingested on Monday 26 October.

1054 papers, three ways

Where today's batch landed

By the paper's primary arXiv category.
cs.LG153
cs.CV149
cs.AI110
cs.CL97
quant-ph97
cs.RO74
cs.CR32
cond-mat.mes-hall30

The index

The day's strongest 60, by score. Every title links its paper here and on arXiv.

  1. CRM-Agent-Bench: Benchmarking Tool-Using Agents in Stateful Enterprise Environments [arXiv]
  2. Strengthening AC optimal power flow relaxation through offline bound tightening and linear approximations [arXiv]
  3. MI-MIDI 2: From Observation to Control with Sparse Autoencoders in Text-to-MIDI Generation [arXiv]
  4. The Impact of Inclination, Redshift, and Mass Ratio Selection on Cosmological Constraints from Bright Sirens Beyond Conventional Fisher Approximations [arXiv]
  5. GS-MoE: Training-Free MoE Compression via Groupwise Subspace Sharing and Pruning [arXiv]
  6. Scalable Reservoir Computing for Disturbance Localization and Identification in Power Grids [arXiv]
  7. MAGIC: Margin-gated Contrastive Debiasing with Evidence Restoration for VLM Test-time Robustness Scaling [arXiv]
  8. Additive uncertainty relations via pseudospectral methods [arXiv]
  9. SynMark: Semantic-Coordinate Multi-Bit Watermarking during LLM Code Generation [arXiv]
  10. OGLE-2015-BLG-1395Lb: A Cold Sub-Saturn Mass Planet Orbiting an M Dwarf [arXiv]
  11. CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition [arXiv]
  12. BARC-Bench: Benign Asks, Risky Consequences [arXiv]
  13. RSI-Opt: Recursive Self-Improvement Optimizer Agents For Model Training [arXiv]
  14. Magnetic field effect on neutral color superconductivity as a mechanism to boost the inner magnetic field of magnetars [arXiv]
  15. Can Spontaneous Time-Translation Symmetry Breaking Be Quantized Without Explicit Branches? [arXiv]
  16. PPO vs. SAC: A Comparative Study for Cooperative Motion Control of Dual-Arm Space Robots [arXiv]
  17. Asymptotic scars arising from the remnants of quantum phase transitions [arXiv]
  18. Rethinking Representation Geometry in Supervised Contrastive Learning under Long-Tailed Data [arXiv]
  19. NUViT: Vision Transformers for Fourier Space Data [arXiv]
  20. X-Dubbing: Speech-to-Speech Translation under Temporal and Visual Constraints [arXiv]
  21. On Exact Critical-Region Boundaries in Continuous-Time Multiparametric Optimal Control Problems [arXiv]
  22. Do AI-for-Science Agents Preserve Model Safety? [arXiv]
  23. Can Test-Time Memory Evolution Outperform Test-Time Training for Scientific Discovery? [arXiv]
  24. A Unified Model for Radioactive Signatures and Source Identifiability in Neutron Star Mergers [arXiv]
  25. Robustness of Quantum Correlations and Controlled Teleportation Fidelity under Noisy Channels [arXiv]
  26. A Dynamic Inventory Rationing Model for Last-Mile Delivery of Medication in Liberia's Community Health Program [arXiv]
  27. Real-space construction and selection rules of center-of-mass momentum eigenstates in two-dimensional many-body lattices [arXiv]
  28. How Much Do You Trust A Dog? Quantifying Hint-Following in Reasoning Language Models [arXiv]
  29. HyScale: Scaling Long-Context Training for Learned Sparse Attention [arXiv]
  30. ACVA-Bench: Beyond Prompt Following in Abstract Concept Visualization Ability for Text-to-Image Models [arXiv]
  31. Concept Acupuncturist: An Efficient and Precise Concept Erasure Method for Diffusion Models via Sparse Autoencoders [arXiv]
  32. Understanding Communication Failures in Multi-Agent LLM Systems [arXiv]
  33. Concentrated Causal Influence, Distributed Diagnostic Information: Cognitive Impairment Decisions in Large Audio-Language Models [arXiv]
  34. SPA: Structure-Preserving Approximation of Gated Activations under FHE [arXiv]
  35. Controlled Evaluation of EEG Foundation Models: Random Initialisation, Band-Power and Validation-Selected Probes [arXiv]
  36. DriveMind: Mind-Evolving Belief Tracking for Closed-Loop Autonomous Driving [arXiv]
  37. DBCM: Dual-Branch Calibration Module for Test-Time Adaptation in Time Series Forecasting [arXiv]
  38. Guided Zero-Shot Anomaly Explanation for Anonymized Crowd Video Monitoring: Context, Categorization, and Spatial Localization [arXiv]
  39. RADIANT: Object-Centric Feed-Forward Global Illumination [arXiv]
  40. Stability Enhancement of Cascaded DC Systems via Source Output-Impedance Reshaping Using Model-Free Control [arXiv]
  41. Label-Budgeted Attribute Inference from Protected Features in Edge-Cloud Split Multimodal LLMs [arXiv]
  42. Behaviour analysis for large language model-driven agents [arXiv]
  43. The Extended Diode Equation [arXiv]
  44. AgentTrace: Execution-Level Fingerprinting of Black-Box LLM Agents [arXiv]
  45. DiscoDemo: Discovering Efficient and Diverse Robot Demonstrations for Imitation Learning [arXiv]
  46. TTT in VLA Models: Exploring Training with Human Demonstrations [arXiv]
  47. Model-Free Output-Admittance Reshaping for Stability Enhancement of Grid-Connected Inverters [arXiv]
  48. DELA: Learning Controllable Edits for Diffusion Language Models [arXiv]
  49. Heterogeneous superconductivity emerging from a charge-stripe state in kagome lattices [arXiv]
  50. Learning Nonlinear Waveform Fusion Across Scales for Integrated Sensing and Communication [arXiv]
  51. PRISM: Promoting Sampling Diversity of Large Language Models with Implicit Stochastic Modulation [arXiv]
  52. Coupling Execution Horizon and Sampling Depth in Flow-Matching VLA Policies [arXiv]
  53. ReMedI: Inference-Time Refinement and Self-Distillation for Controllable Molecular Optimization [arXiv]
  54. Putting Transformer Theories to the Test [arXiv]
  55. PersonaScienceBench: Evaluating Scientific Agents with Diverse Simulated Researchers [arXiv]
  56. Beyond What to Say: Proactive Utterance Timing with Inner Thoughts for Werewolf Agents [arXiv]
  57. Context or Knowledge: How LLMs Attribute Their Own Answers [arXiv]
  58. Who Verifies the Tools? How Warped Formatters Mislead Agents and Their Verifiers [arXiv]
  59. A Self-Consistent Two-Carrier Drift-Diffusion Framework for p-n Junctions: Extended Diode Equation, Power Dissipation, and Base Transit Time [arXiv]
  60. REACT: Representational Early Analysis and Control of Chain-of-Thought Trajectories [arXiv]

Listen

Everything the estate published on 2026-10-26, in the player. Playing from this page is how the edition learns whether anyone did.

Previous editions